ODIs December 7, 2009

Innings Power Factor: a new measure for ODI innings

Alex Tierno had suggested that I create a new factor for ODI innings, incorporating three features of an ODI innings: runs, scoring rate and contribution to team score
83

This piece was written in collaboration with Alex Tierno

I have attempted something new for "It Figures" in this article. Almost on a continuous basis, many of the readers have offered suggestions for analysis. Some of these have been answered as a response to the comment. Some require creation and publishing of tables in existing articles. Once in a while I get a suggestion which warrants a separate article. This is the first one created based on this premise. In future when such an idea comes up, I will do a similar publishing.

This is based on a suggestion made by Alex Tierno a few months back. I was tied up with various things and only now could I do justice to the suggestion. Alex, in consultation with me, has also has polished the idea with some tweaking recently.

Alex has suggested that I create a new factor for ODI innings which he called "Destructive index". I have called that the "Innings Power Factor". This is a single factor which incorporates the three major features of an ODI innings: runs, scoring rate and contribution to team score.

I will respond to reader comments in a general manner prior to publishing. However Alex can respond to these in a summary fashion after publishing.

The formula used is

Innings Power Factor (IPF) = Runs scored * Scoring rate * % of Team score.

The more I studied this the more I was impressed with the simplicity and effectiveness of this as a measure of ODI innings. The higher each of these factor is, the more the value of the innings. At the same time, the introduction of the % of Team score moderates the factor as exampled below.

Let us take two examples. a 50 in 20 balls would get 125 points using the first two factors. A 125 in 125 balls would also result in a value of 125 points. However the % of Team score for the first innings is likely to be 15-25% and 40-50% for the second. This takes care of higher valuation of higher scores.

It should be noted that this factor, being a pure batting one, does not take into account team strengths, bowling quality, pitch type, innings status, result et al. If all these factors are introduced it will become another Ratings exercise. So please do not send any comments on the exclusion of these factors. In a way this is similar to the 100s-50s. A 100 is a hundred irrespective of when, where and who it was scored against. I also like this measure since it does not have the 99 to 100 problem I have earlier talked about.

This is an unforgiving measure and requires all three factors to work together to finish with a reasonable value. Cameos tend to lose out. At the end of the article I have done a table which takes into account only the first two values.

I briefly toyed with the idea of having a fourth factor, the Result (1.1/1.0 or 1.0/0.9). I gave up for two reasons. It penalizes Coventry/Tendulkar/Hayden/RASmith/Ponting et al unfairly. They could not have done anything more. Also in the top-100, 85 are wins, so this factor will not have any great impact.

The analysis is done in two parts. In the first part, all the innings are analysed and the IPF calculated, sequenced and the table drawn up. By a perusal of this table I have determined that an IPF of 70 (100 off 60 out of 240) translates into an outstanding performance and one above 40 (80 off 50 out of 250) is a very good performance. Also IPF values above 10 (50 off 50 out of 250) translate into good performances. At the other end, only IPF values of below 2.0 might be termed unsuccessful innings. These summaries are posted into the player data and Player tables are drawn up.

1. Top ODI performances ordered by IPF (Runs * S/R * % TS) : > 50.0

No MtId Year Player Name          IPF  For  Vs I Runs(Balls) S/R  %TS TmScre Res

1.2660 2007 McCullum B.B 192.5 Nzl Bng 2 80*( 28) 285.7 84.2% [ 95] Won 2.1209 1997 Saeed Anwar 152.9 Pak Ind 1 194 (146) 132.9 59.3% [327] Won 3.2873 2009 Coventry C.K 150.0 Zim Bng 1 194*(156) 124.4 62.2% [312] 4.0264 1984 Richards I.V.A 146.0 Win Eng 1 189*(170) 111.2 69.5% [272] Won 5.0216 1983 Kapil Dev N 146.0 Ind Zim 1 175*(138) 126.8 65.8% [266] Won 6.1236 1997 Ijaz Ahmed 146.0 Pak Ind 2 139*( 84) 165.5 63.5% [219] Won 7.1652 2000 Jayasuriya S.T 140.2 Slk Ind 1 189 (161) 117.4 63.2% [299] Won 8.2290 2005 Dhoni M.S 139.5 Ind Slk 2 183*(145) 126.2 60.4% [303] Won 9.0457 1987 Richards I.V.A 131.8 Win Slk 1 181 (125) 144.8 50.3% [360] Won 10.2824 2009 Sehwag V 131.3 Ind Nzl 2 125*( 74) 168.9 62.2% [201] Won 11.1049 1996 Kirsten G 130.2 Saf Uae 1 188*(159) 118.2 58.6% [321] Won 12.1207 1997 Jayasuriya S.T 125.3 Slk Ind 2 151*(120) 125.8 65.9% [229] Won 13.0169 1983 Gower D.I 125.2 Eng Nzl 1 158 (118) 133.9 59.2% [267] Won 14.2082 2004 Gilchrist A.C 117.4 Aus Zim 1 172 (126) 136.5 50.0% [344] Won 15.1523 1999 Tendulkar S.R 113.3 Ind Nzl 1 186*(151) 123.2 49.5% [376] Won 16.2581 2007 Gilchrist A.C 113.2 Aus Slk 1 149 (104) 143.3 53.0% [281] Won 17.1010 1995 Lara B.C 112.4 Win Slk 1 169 (129) 131.0 50.8% [333] Won 18.2420 2006 Boucher M.V 111.8 Saf Zim 1 147*( 68) 216.2 35.2% [418] Won 19.2349 2006 Gibbs H.H 110.2 Saf Aus 2 175 (111) 157.7 40.0% [438] Won 20.2923 2009 Tendulkar S.R 109.5 Ind Aus 2 175 (141) 124.1 50.4% [347]

Readers, myself included, would be surprised at the top entry. However a careful perusal of McCullum's blitzkrieg, about which I had come out with an article recently, justifies the position. A middling score, boosted by an unbelievable scoring rate and an amazingly high % of Team score has propelled this innings to the top. Those who question why McCullum's innings outshines Richards'/Jayasuriya's/Anwar's/Kapil's masterpieces should note that McCullum's innings meets Alex's "destructiveness" characteristic to a T.

Then come the 6 famous innings by Anwar/Coventry/Richards/Jayasuriya/Dhoni/Kapil. The odd innings which splits these six is Izaz Ahmed's Lahore demolition of India.

In the top 37 100+ innings, Richards and Jayasuriya have four innings each, Tendulkar has three and Gilchrist and Lara have two each.

Out of the 446 70+ innings, 238 (53.4%) are in the first innings.

To view the complete IPF list, please click here.

Now let us see the player tables.

2. IPF Summary by Batsmen: Ordered by average IPF value

SNo.Batsman             Cty Inns  Runs <-Innings Power Factor->
> 10 75+ 50+ 10+   Avge

1.Zaheer Abbas Pak 60 2572 20 0 5 15 13.29 2.Richards I.V.A Win 167 6721 58 5 5 48 13.10 3.Tendulkar S.R Ind 425 17178 138 7 20 111 12.53 4.Gayle C.H Win 200 7430 55 5 11 39 11.89 5.Trescothick M.E Eng 122 4335 34 1 5 28 10.52 6.Gilchrist A.C Aus 279 9619 72 5 9 58 10.42 7.Smith G.C Saf 147 5613 49 2 4 43 10.17 8.Lara B.C Win 289 10406 80 6 6 68 10.00 9.Pietersen K.P Eng 85 3179 28 0 1 27 9.93 10.Hayden M.L Aus 155 6132 41 2 1 38 9.90

Two 80s greats, Zaheer Abbas and Richards lead this table with averages exceeding 13.00. Richards has achieved this in over 150 innings. However note the high average of Tendulkar, 12.53 achieved in 425 innings. Then come a string of modern ODI stalwarts.

To view the complete file, please click here.

3. IPF Summary: Ordered by number of above average IPF values ( > 10)

SNo.Batsman             Cty Inns  Runs  
> 10(!!)  75+ 50+ 10+
No   %

1.Tendulkar S.R Ind 425 17178 138-32.5 7 20 111 2.Jayasuriya S.T Slk 429 13377 99-23.1 8 14 77 3.Ponting R.T Aus 321 12310 90-28.0 1 9 80 4.Inzamam-ul-Haq Pak 350 11739 86-24.6 0 7 79 5.Ganguly S.C Ind 300 11363 83-27.7 3 7 73 6.Lara B.C Win 289 10406 80-27.7 6 6 68 7.Kallis J.H Saf 281 10410 78-27.8 0 5 73 8.Dravid R Ind 313 10765 76-24.3 0 3 73 9.Gilchrist A.C Aus 279 9619 72-25.8 5 9 58 10.de Silva P.A Slk 296 9284 71-24.0 0 5 66

Tendulkar has 138 innings which meet the 10 points cut-off, 7 of these are the higher level performances exceeding 70 points. Jayasuriya follows with 99 performances, 8 at the top level and then follows Ponting with 90. Inzamam-ul-haq, Ganguly and Lara follow next. The presence of that Sri Lankan great, de Silva, in no.10 position is heart-warming. Two days during early-1996 are justification for this place.

To view the complete file, please click here.

4. IPF Summary: Ordered by % of differential (success-failure) performances

(IPF values < 2.0) - (IPF values > 10.0)

SNo.Batsman Cty Inns Runs <-Innings Power Factor--> > 10(!!) < 2 (??) Diff No % No % %

1.Zaheer Abbas Pak 60 2572 20-33.3 23-38.3 5.00 2.Richards I.V.A Win 167 6721 58-34.7 69-41.3 6.59 3.Hussey M.E.K Aus 102 3623 29-28.4 38-37.3 8.82 4.Pietersen K.P Eng 85 3179 28-32.9 37-43.5 10.59 5.Smith G.C Saf 147 5613 49-33.3 66-44.9 11.56 6.Greenidge C.G Win 127 5134 39-30.7 55-43.3 12.60 7.Tendulkar S.R Ind 425 17178 138-32.5 192-45.2 12.71 8.Hayden M.L Aus 155 6132 41-26.5 63-40.6 14.19 9.Ponting R.T Aus 321 12310 90-28.0 137-42.7 14.64 10.Jones D.M Aus 161 6068 45-28.0 70-43.5 15.53

Since the number of matches varies considerably, I have also introduced a % value, which is the IPFs / ODI Inns. Richards leads in this measure with 34.7%, followed by Pietersen with 34.1%, Smith with 34.0%, Zaheer Abbas with 32.5% and Tendulkar with a high 32.5% despite playing 425 innings. This means that these great players produced a very good batting performance once in three innings. That is really something. Now the chronicle of failures. Mike Hussey has failed to deliver in only 38.2% of the innings, Zaheer Abbas 38.3% and Michael Bevan, 39.8%. It can be seen that most of these batsmen play in the middle order.

Now comes a composite value which is the % failure - % success. The lower this value is the more effective the batsman is. The above table has been ordered in the increasing order of this difference %.

Zaheer Abbas is the top batsman with a differential % value of just 5%. The great Richards follows next with 6.59% and then two modern greats, Pietersen and Hussey, with differential % below 10. Two olden day greats, Greenidge and Jones, split the four modern giants, Smith, Tendulkar, Hayden and Ponting.

The more I see the table the more I feel that this is the single table which encompasses the ODI greats in full.

To view the complete file, please click here.

5. Top ODI performances ordered by IPF-2 (Runs * S/R) : > 125.0

No MtId Year Player Name         IPF-1 For Vs  I Runs(Balls) S/R  Res

1.2420 2006 Boucher M.V 317.8 Saf Zim 1 147* ( 68) 216.2 Won 2.1090 1996 Jayasuriya S.T 276.2 Slk Pak 1 134 ( 65) 206.2 Won 3.2349 2006 Gibbs H.H 275.9 Saf Aus 2 175 (111) 157.7 Won 4.0457 1987 Richards I.V.A 262.1 Win Slk 1 181 (125) 144.8 Won 5.1125 1996 Shahid Afridi 260.1 Pak Slk 1 102 ( 40) 255.0 Won 6.1209 1997 Saeed Anwar 257.8 Pak Ind 1 194 (146) 132.9 Won 7.2349 2006 Ponting R.T 256.2 Aus Saf 1 164 (105) 156.2 8.2272 2005 Vincent L 246.5 Nzl Zim 1 172 (120) 143.3 Won 9.2774 2008 Yuvraj Singh 244.2 Ind Eng 1 138* ( 78) 176.9 Won 10.2873 2009 Coventry C.K 241.3 Zim Bng 1 194* (156) 124.4

Alex also wanted me to do a calculation excluding the third moderating factor, the % of Team score. In other words make this a pure batting individual factor. The above table is an intriguing one. Again, it is a surprise to see Boucher on top. However his is a big century at a scoring rate above 2.00. Similarly Jayasuriya's innings finds its place. It is amazing that a score as low as Afridi's 102 has found its place in the top-10.

Out of the 379 125+ innings, 264 (70%) are in the first innings. This is a marked change to the reasonably equal split for IPF. Possibly the uncertainty of the target for the first innings might have contributed to this disparity.

In the revised table there is only one change. Kapil Dev has secured 221.92 points for his 175* and moves to 21st place. The complete table has not been replaced.

To view the complete file, please click here.

Let me thank Alex for an excellent idea. I request the readers to come out with a similar factor for bowling performances. Wickets/Economy rate seems quite simple but possibly the readers could improve on this. No outside-bowling parameters please.

I will now give serious considerations to some of Seshasayee's excellent suggestions. I have already done the one on Test players' continuous streaks. The tables have been incorporated in my previous article.

The next one is an intriguing one. Sesha wanted me to analyse the next 2/3 years' programmes and do a projection of Test runs and wickets. My initial reaction was to avoid opening this Pandora's box because of the expected reactions of certain types of readers. Then I realized that this would only be an analysis and I should do this without worrying about the reactions of readers. However this is a tough one and the first thing I have to do is to prepare a complete matrix of tours for the next 2/3 years.

Another of Sesha's suggestions is for me to an analysis on player combinations (2-11) who have played in most number of Tests. Again, another tough one but worth doing because of the novelty and insights it would bring.

Anantha Narayanan has written for ESPNcricinfo and CastrolCricket and worked with a number of companies on their cricket performance ratings-related systems

Comments have now been closed for this article

  • alex on February 25, 2010, 10:47 GMT

    SRT's 147-ball 200* ranks #9 in Table 1 @136, and #4 in Table 5 @272. It lasted full 50 overs, and 200 is the highest score so far. Was it his greatest ever ODI knock? Of the recent vintage, I think 117* and 138 supercede it, considering the importance of occasion & circumstances.

    I think 150-ball 225 is a possibility in ODI's. That will score 337 on Table 5, and somewhere between 150 and 200 on Table 1. [[ Alex I am not sure how many people would be reading the blog published a few months back. I think his 175 is a greater innings considering all points. In fact I am thinking of doing an analysis of 175+ innings. Ananth: ]]

  • Michael on January 6, 2010, 16:16 GMT

    Hi Ananth, I know that things like 'bowling strength' takes this into consideration in some of your analyses, but can we all agree that it is time to ignore all matches that include Zim and Bang in genuine criket stats?? In this analysis, yes the McCullum inns was amazing, but it is really more like an A-list than an true ODI inns. When considering personal stats, players like Murali and Kallis have major Z/B boosts, and really, it does not reflect on true career averages when Murali takes 176 wickets@15 in these games. I wouldn't mind keeping Z/B if these teams simply went through a transition period before becoming true test powers (Similar to Sri Lanka in the 80's), but apart from a couple of years for Zim, these teams have never been remotely competitive. Ho about you just do analyses that exclude Z/B in the future?? [[ Michael The Bangladeshi bowling, the McCullum mauling notwithstanding, is quite good and ranks at par with many an estabilished side. Mortaza, Shahdat, Abdul Razzaq, Md Rafeeque and now Shakib-al hassan. It is their batting which is below-par. Ananth: ]]

  • Anshu Jain on December 28, 2009, 19:17 GMT

    cont...

    The top 2 bowling performances in ODI history have the Bowling Performance Factor calculated and listed below:

    McGrath 7/15 vs Namibia - (3.21/2.14)*(8.4/6)*(7)*(25)/(10)= 36.75 Vaas 8/19 vs Zimbabwe - (2.42/2.37)*(9.4/6)*(8)*(30)/(10)= 38.39

    The huge difference between the ranges compared to the earlier listed performances and their BPFs is interesting, and worth exploring.

    Also, while the avg. per wicket figure at first appeared important, it somehow felt right to leave out from the BPF calculations.

    Havent tested the formula extensively yet, but I feel it would be worth getting others' and your views on.

  • Anshu Jain on December 28, 2009, 19:16 GMT

    The top 2 bowling performances in ODI history have the Bowling Performance Factor calculated and listed below:

    McGrath 7/15 vs Namibia - (3.21/2.14)*(8.4/6)*(7)*(25)/(10)= 36.75 Vaas 8/19 vs Zimbabwe - (2.42/2.37)*(9.4/6)*(8)*(30)/(10)= 38.39

    The huge difference between the ranges compared to the earlier listed performances and their BPFs is interesting, and worth exploring.

    Also, while the avg. per wicket figure at first appeared important, it somehow felt right to leave out from the BPF calculations.

    Havent tested the formula extensively yet, but I feel it would be worth getting others' and your views on.

  • Anshu Jain on December 28, 2009, 19:13 GMT

    Ananth and Alex - great work with the IPF/DI. For the Bowling Power Factor, the following formula may be looked at:

    Bowling Performance Factor = (Inns Economy/Bowler's Economy)*(Inns SR/Bowler's SR)*(No. of wickets taken)*(Sum of Points of Wicket-Positions taken)/(Total Innings Wickets) where:

    Wicket-Positions 1-4 are worth 5 points each Wicket-Positions 5-7 are worth 3 points each Wicket-Positions 8-11 are worth 1 point each

    For example, taking Anil Kumble's 6/12 against the West Indies in the Hero Cup 1994, Kumble's Bowling Performance Factor would have been

    (3.06/1.94)*(24.1/6.16)*(6)*(12)/(10)= 44.43

    A few other top bowling performances' (in my opinion) factors are calculated and listed:

    Walsh 5/1 vs Sri Lanka - (1.92/0.22)*(17.1/5.4)*(5)*(7)/(10)= 96.72 Bond 6/23 vs Australia - (4.16/2.3)*(33.3/10)*(6)*(24)/(9)= 96.36 Mendis 6/13 vs India - (4.37/1.62)*(23.7/8)*(6)*(20)/(10)= 95.89 Murali 7/30 vs India - (4.62/3)*(29.3/8.57)*(7)*(21)/10= 77.39

    cont...

  • Pankaj Joshi on December 19, 2009, 13:59 GMT

    Ananth, your point on quality of wickets and destructive impact is well taken. Kapil and co in WC 85 were terrific in opening inroads so the spinners were never under pressure. One thought - could you just look at the top 25 wicket takers from this perspective? If batsman #1 is dismissed the bowler gets 11 points and if batsman #11 is dismissed the bowler gets 1 point. The aggregate then is divided by the number of wickets taken in career and you get the quality index of the bowler. Basic kernel, surely can be developed further. Wait your thoughts.

  • Naren on December 19, 2009, 4:48 GMT

    xolile: Quite correct about the stage of an innings as far as determining "destructiveness" of a spell is concerned.A 5 wicket haul early up would normally be worth more than a similar haul at the end of an innings. Abhi: Quite correct about runs given away during a destructive spell, especially early up. The flip side,however,is say a 5 wicket haul at the death when batsmen are simply throwing their bats around. So, again "xolile"'s argument needs to be factored in.

  • keyur on December 18, 2009, 10:01 GMT

    good analysis but i have one problem with it. the analysis counts the destructiveness of the inning and for the same the formula runs*sr*% of team score seems right

    But the average destructiveness of a batsman over his entire career can't be calculated by his average ipf score

    this is because your ipf is a exponential index (because ipf = runs scored are cubed and divided by balls faced & team score)

    so i strongly disagree with table 2 ordered by average ipf value and suggest this:

    career ipf = career runs*career strike rate*(career runs/total team runs in matches in which player was part of).

    these info must be easily available by statsguru and will be a better index for career destructiveness!

    the reason is that suppose a player with 1 100+ ipf inn & plenty of 7 ipf innings gets rewarded over a player with plenty of 9 ipf innings on basis of a single innings

    this gets corrected in the new formula as suggested & is based on the very parameters you have used 4 ipf

  • alex on December 15, 2009, 10:20 GMT

    Ananth - on bowler's destructiveness, please see if the following appeals to you, assuming the bowler takes N wickets while defending a total:

    destructiveness = (a_1+...a_N) + overs bowled*(#overs bowled/allowed quota)*(#overs bowled* RRR- runs given)

    where a_i: expected "net" runs saved by taking i-th wicket, RRR: run-rate required by the opposition to win the match. Now, a_i subject to interpretation. For example,

    a_i = 0.3*(Term 1+ Term2 + Term 3), where Term 1 = average*average SR*average %TS Term 2 = runs*SR*%TS Term 3 = Term2 applied to the current partnership.

    Term 1 captures how well he bats on average, Term 2 captures how well he was in this innings, and Term 3 captures how dangerous the partnership was. Now, many variations can be tried. Suppose you are defending 250 in 50 overs. Hence, on average, you can allow at most 25 runs/partnership, 25 runs/batsman, 30 deliveries/wkt. You can use this to calibrate a_i. [[ Alex The one thing I admired about your Batting destructiveness factor was its inherent simpilicity and the abiliity for people like Satvik to say he would create the index for ODI innings while watching the match. That is what people would have done while watching yesterday's run-feast which passed off as a match. The formula you have suggested goes the other way. It is too complicated. We have to come out with an easily understandable and calculable factor. Ananth: ]]

  • Pankaj Joshi on December 14, 2009, 13:44 GMT

    Ref Navin Agarwal's comment. Its a good case study of the hypothesis I had suggested. I had searched on my account and generated one example. India vs Aus WC 1983 first match. Aus 320-9 in 60 overs, Kapil 5-43 in 12. THAT should get somewhere on the bowling index. Ananth, your take please. [[ Pankaj I am afraid this is only an average performance. 1. The Australian scoring rate was 5.33 and Kapil's was 3.5. Certainly better but not the huge difference we see in many other innspells. 2. Kapil took the first wicket, then came back towards the end and picked up wkts 6-9. He came back at 250 or so for 5. So this is a below-avge collection of wickets. Compare this with Kapil's own 4 for 30 against Australia in the 1985 B&H. 3 of the 4 wickets were top wickets and India won. That Australia was also a stronger team. Let me give you the Rating pts for some other project I am doing. The 5 for 43 gets 290 and 4 for 30 gets 350. Just to give you a comparison. Ananth: ]]

  • alex on February 25, 2010, 10:47 GMT

    SRT's 147-ball 200* ranks #9 in Table 1 @136, and #4 in Table 5 @272. It lasted full 50 overs, and 200 is the highest score so far. Was it his greatest ever ODI knock? Of the recent vintage, I think 117* and 138 supercede it, considering the importance of occasion & circumstances.

    I think 150-ball 225 is a possibility in ODI's. That will score 337 on Table 5, and somewhere between 150 and 200 on Table 1. [[ Alex I am not sure how many people would be reading the blog published a few months back. I think his 175 is a greater innings considering all points. In fact I am thinking of doing an analysis of 175+ innings. Ananth: ]]

  • Michael on January 6, 2010, 16:16 GMT

    Hi Ananth, I know that things like 'bowling strength' takes this into consideration in some of your analyses, but can we all agree that it is time to ignore all matches that include Zim and Bang in genuine criket stats?? In this analysis, yes the McCullum inns was amazing, but it is really more like an A-list than an true ODI inns. When considering personal stats, players like Murali and Kallis have major Z/B boosts, and really, it does not reflect on true career averages when Murali takes 176 wickets@15 in these games. I wouldn't mind keeping Z/B if these teams simply went through a transition period before becoming true test powers (Similar to Sri Lanka in the 80's), but apart from a couple of years for Zim, these teams have never been remotely competitive. Ho about you just do analyses that exclude Z/B in the future?? [[ Michael The Bangladeshi bowling, the McCullum mauling notwithstanding, is quite good and ranks at par with many an estabilished side. Mortaza, Shahdat, Abdul Razzaq, Md Rafeeque and now Shakib-al hassan. It is their batting which is below-par. Ananth: ]]

  • Anshu Jain on December 28, 2009, 19:17 GMT

    cont...

    The top 2 bowling performances in ODI history have the Bowling Performance Factor calculated and listed below:

    McGrath 7/15 vs Namibia - (3.21/2.14)*(8.4/6)*(7)*(25)/(10)= 36.75 Vaas 8/19 vs Zimbabwe - (2.42/2.37)*(9.4/6)*(8)*(30)/(10)= 38.39

    The huge difference between the ranges compared to the earlier listed performances and their BPFs is interesting, and worth exploring.

    Also, while the avg. per wicket figure at first appeared important, it somehow felt right to leave out from the BPF calculations.

    Havent tested the formula extensively yet, but I feel it would be worth getting others' and your views on.

  • Anshu Jain on December 28, 2009, 19:16 GMT

    The top 2 bowling performances in ODI history have the Bowling Performance Factor calculated and listed below:

    McGrath 7/15 vs Namibia - (3.21/2.14)*(8.4/6)*(7)*(25)/(10)= 36.75 Vaas 8/19 vs Zimbabwe - (2.42/2.37)*(9.4/6)*(8)*(30)/(10)= 38.39

    The huge difference between the ranges compared to the earlier listed performances and their BPFs is interesting, and worth exploring.

    Also, while the avg. per wicket figure at first appeared important, it somehow felt right to leave out from the BPF calculations.

    Havent tested the formula extensively yet, but I feel it would be worth getting others' and your views on.

  • Anshu Jain on December 28, 2009, 19:13 GMT

    Ananth and Alex - great work with the IPF/DI. For the Bowling Power Factor, the following formula may be looked at:

    Bowling Performance Factor = (Inns Economy/Bowler's Economy)*(Inns SR/Bowler's SR)*(No. of wickets taken)*(Sum of Points of Wicket-Positions taken)/(Total Innings Wickets) where:

    Wicket-Positions 1-4 are worth 5 points each Wicket-Positions 5-7 are worth 3 points each Wicket-Positions 8-11 are worth 1 point each

    For example, taking Anil Kumble's 6/12 against the West Indies in the Hero Cup 1994, Kumble's Bowling Performance Factor would have been

    (3.06/1.94)*(24.1/6.16)*(6)*(12)/(10)= 44.43

    A few other top bowling performances' (in my opinion) factors are calculated and listed:

    Walsh 5/1 vs Sri Lanka - (1.92/0.22)*(17.1/5.4)*(5)*(7)/(10)= 96.72 Bond 6/23 vs Australia - (4.16/2.3)*(33.3/10)*(6)*(24)/(9)= 96.36 Mendis 6/13 vs India - (4.37/1.62)*(23.7/8)*(6)*(20)/(10)= 95.89 Murali 7/30 vs India - (4.62/3)*(29.3/8.57)*(7)*(21)/10= 77.39

    cont...

  • Pankaj Joshi on December 19, 2009, 13:59 GMT

    Ananth, your point on quality of wickets and destructive impact is well taken. Kapil and co in WC 85 were terrific in opening inroads so the spinners were never under pressure. One thought - could you just look at the top 25 wicket takers from this perspective? If batsman #1 is dismissed the bowler gets 11 points and if batsman #11 is dismissed the bowler gets 1 point. The aggregate then is divided by the number of wickets taken in career and you get the quality index of the bowler. Basic kernel, surely can be developed further. Wait your thoughts.

  • Naren on December 19, 2009, 4:48 GMT

    xolile: Quite correct about the stage of an innings as far as determining "destructiveness" of a spell is concerned.A 5 wicket haul early up would normally be worth more than a similar haul at the end of an innings. Abhi: Quite correct about runs given away during a destructive spell, especially early up. The flip side,however,is say a 5 wicket haul at the death when batsmen are simply throwing their bats around. So, again "xolile"'s argument needs to be factored in.

  • keyur on December 18, 2009, 10:01 GMT

    good analysis but i have one problem with it. the analysis counts the destructiveness of the inning and for the same the formula runs*sr*% of team score seems right

    But the average destructiveness of a batsman over his entire career can't be calculated by his average ipf score

    this is because your ipf is a exponential index (because ipf = runs scored are cubed and divided by balls faced & team score)

    so i strongly disagree with table 2 ordered by average ipf value and suggest this:

    career ipf = career runs*career strike rate*(career runs/total team runs in matches in which player was part of).

    these info must be easily available by statsguru and will be a better index for career destructiveness!

    the reason is that suppose a player with 1 100+ ipf inn & plenty of 7 ipf innings gets rewarded over a player with plenty of 9 ipf innings on basis of a single innings

    this gets corrected in the new formula as suggested & is based on the very parameters you have used 4 ipf

  • alex on December 15, 2009, 10:20 GMT

    Ananth - on bowler's destructiveness, please see if the following appeals to you, assuming the bowler takes N wickets while defending a total:

    destructiveness = (a_1+...a_N) + overs bowled*(#overs bowled/allowed quota)*(#overs bowled* RRR- runs given)

    where a_i: expected "net" runs saved by taking i-th wicket, RRR: run-rate required by the opposition to win the match. Now, a_i subject to interpretation. For example,

    a_i = 0.3*(Term 1+ Term2 + Term 3), where Term 1 = average*average SR*average %TS Term 2 = runs*SR*%TS Term 3 = Term2 applied to the current partnership.

    Term 1 captures how well he bats on average, Term 2 captures how well he was in this innings, and Term 3 captures how dangerous the partnership was. Now, many variations can be tried. Suppose you are defending 250 in 50 overs. Hence, on average, you can allow at most 25 runs/partnership, 25 runs/batsman, 30 deliveries/wkt. You can use this to calibrate a_i. [[ Alex The one thing I admired about your Batting destructiveness factor was its inherent simpilicity and the abiliity for people like Satvik to say he would create the index for ODI innings while watching the match. That is what people would have done while watching yesterday's run-feast which passed off as a match. The formula you have suggested goes the other way. It is too complicated. We have to come out with an easily understandable and calculable factor. Ananth: ]]

  • Pankaj Joshi on December 14, 2009, 13:44 GMT

    Ref Navin Agarwal's comment. Its a good case study of the hypothesis I had suggested. I had searched on my account and generated one example. India vs Aus WC 1983 first match. Aus 320-9 in 60 overs, Kapil 5-43 in 12. THAT should get somewhere on the bowling index. Ananth, your take please. [[ Pankaj I am afraid this is only an average performance. 1. The Australian scoring rate was 5.33 and Kapil's was 3.5. Certainly better but not the huge difference we see in many other innspells. 2. Kapil took the first wicket, then came back towards the end and picked up wkts 6-9. He came back at 250 or so for 5. So this is a below-avge collection of wickets. Compare this with Kapil's own 4 for 30 against Australia in the 1985 B&H. 3 of the 4 wickets were top wickets and India won. That Australia was also a stronger team. Let me give you the Rating pts for some other project I am doing. The 5 for 43 gets 290 and 4 for 30 gets 350. Just to give you a comparison. Ananth: ]]

  • Abhi on December 13, 2009, 7:08 GMT

    @xolile quite right(as usual!)... my point was that runs leaked are often not necessarily an indicator of a "less" destructive spell. Especially since most of the runs leaked of a bowler during a particularly good spell may be edges/squirts and just generally "fluke" runs...Not necessarily indicators of a poor ball/bowling.

  • Navin Agarwal on December 12, 2009, 3:56 GMT

    I haven't gone through all the comments on bowling. But one spell I remember is Harbhajan Singh's. England scored 307-5 in 50 overs and Harbhajan Singh's spell was 10-2-14-2. There is a diffrence of 4.74 in innings rate and bowlere econ rate. It was not destructive as it did not take wickets but very effective. Without which England might have scored over 350. The match link is here http://www.cricinfo.com/ci/engine/match/65033.html

    Calculate the effectiveness of this spell and tell where it is in the bowling index.

  • love goel on December 11, 2009, 14:14 GMT

    May be, we can divide the innings into different classes on the number of runs scored. So we can can have a list of most destructive 50, another set of most destructive 100, and a third set of set of most desturctive 150's I think that a classification on 40 runs, say a list of <40,40-80 and so on will be good.

    If not runs then certainly balls played can be used as classification.Innings with lesser balls played should be judged seperately while comapring 'destructiveness'. Impact on the match,however,is clearly best considered by the present IPF formula(Runs*SR*%TS).

    For me, the best part of current IPF is its simplicity, and if you apply it , you will almost certainly get the Man of the Match(not considering the bowlers)

  • Xolile on December 11, 2009, 9:33 GMT

    Ananth, Abh, Goel,

    The destructiveness of taking wickets generally decreases as the innings progresses, whereas the destructiveness of bowling dot balls and preventing boundaries generally increases.

    The only truly effective way to measure a bowler’s destructiveness would therefore involve looking at the innings on a ball by ball basis. Ananth has already pointed out that this information is not available for the first 1,500 ODIs. We could however ignore these early matches and focus on the subsequent ones. Since ODI tactics changed radically in around 1995 this may in fact be advisable.

    Only problem is I have generally used StatsGuru – and StatsGuru does not provide ball-by-ball information. Do you perhaps know where I could find ball-by-ball information – preferably in an Excel-friendly format?

  • alex on December 11, 2009, 8:53 GMT

    Ananth - in response to Anirudh's Dec 10 comment.

    While suggesting your IPF, I too had Anirudh's suggestion in mind on a number (actually, a vector) that can be displayed on TV screens alongside average and other stats. A vector containing 4-5 such simple-to-compute IPF scores can give a more complete picture nicely.

    Also, an innings that features in Top 100 in all 4-5 of these IPF tables must be really special. Likewise, for batsmen. That can help calibrate the media hype and fan fanatism in general.

  • Anirudh on December 11, 2009, 8:18 GMT

    Ananth, Thanks for your feedback - I agree with you on the point of Career vs. Innings Analysis. Looking forward to your career note soon. But I hope you can figure out something for the large scores in small chases finishing in sub-35 overs - I personally do not feel they match up to your description of Destructive or Effective innings.

    While we're at it.. you may consider getting a little TM sign on your IPF number.

  • Ananth on December 11, 2009, 7:20 GMT

    Anirudh/Raghav I feel too many factors are hitting here. - Simplicity of calculation - Contribution to team cause - Top order vs Finishers - Career analysis - Purpose of IPF My take is that if the purpose is ONLY determination of destructive innings we have to adopt certain methodologies. On the other hand if the purpose is to use this in a career analysis, the methodology should be different. Sub-10 ball innings would never make the first list while they should have an impact on the later. When we take averages we do not exclude low innings. Possibly, for the career analysis, I should exclude sub-10 ball innings at the end of the innings where the batsman has remained not out. It is essential that the basic IPF factor should be simple to understand, easy to calculate, comparable and genuinely contribute to the team cause. I will also make required adjustment when I use this for career analysis.

  • Anirudh on December 11, 2009, 7:06 GMT

    Going through the table of Top-IPF innings I observed 13 out of the top 50 innings are in chases below 220. Mostly all are recent matches where it's safe to say these scores are "sub-par". Makes me wonder if we should wither qualify this - multiply by overs batted/50 overs - or maybe even ignore them completely. Because that batting effort, albeit superlative, doesn't compare to the other 37 and doesn't win the match itself. Further to solve issue of opener vs finisher - if we remove the small innings altogether (10balls or less not out, or those entering after 45/47th over) - it will put the finishers on a more even keel even though it will reduce their sample size - for their innings that qualify, they may have less runs and %TS but higher strike rates. Moreover, if a team bats someone at 6/7 in a limited over contest he can't be that important to the team over the long term as the number 1/2 batter with the bat at least hope to see your response to these.

  • Anirudh on December 11, 2009, 6:22 GMT

    Contd. So some cutoff or additional measure to temper the IPF for such innings would really give an even more accurate picture. Just to demonstrate - Sachin's innings against Zimbabwe in Sharjah is ranked pretty high too (I use this because Sachin is my absolute favourote so I wouldn't want any of his innings to lose out on a high spot) - but with a small target chased down in 35 odd overs, is it really one of his most destructive or impactdul? Furthermore even if we tweak the definition slightly away from destructive to "effective" I wouldn't want these scores high up. Since I am really pressing you hard to use this as a measure of a batsman's ability/impact in the game over a long period - I.e career, this suggestion becoes more relevant. I also agree with Raghav about excluding certain innings below 10 balls or so - this would neither reward nor penalize batsmen - but for the batsman in the top 6 who really matter, the IPF could eventually be a defining statistic. Thanks.

  • Anirudh on December 11, 2009, 6:13 GMT

    Anantha, Thanks for reading my comments. Maybe I am taking the subject off to a slight tangent but I really feel that you have something going with this number which can go beyond comparing some innings to even defining the impact of a player in the game. Maybe use it instead of averages on TV -see my comment on baseball where they have a Slugging number to define how good a hitter is. Career figures would be a very good start for that. Regarding the use of %team score - it really goes on to address many issues - particularly Not outs, featherbed pitches where a big innings by the opener to set up the total can be overshadowed by a quick fifty in the slogs, etc. There are plenty of reasons so I request to let this remain. But if you see the table there are many innings which are chasing small totals - while I feel they are excellent efforts, I feel the high %TS really carries them there though the bowlers really worked hard on this to restrict the opposition. Contd

  • Abhi on December 11, 2009, 4:38 GMT

    Ananth, Depends on what we are trying to measure , I suppose. Depends on how many balls Gilmours 6 took vs. Sreeshants 6…which would be perhaps more indicative of “destructiveness”. The Gilmour spell would perhaps be the better “allround “ spell.If Sreesanth took out the better batsmen in quick time,say 5 overs (inspite of leaking more runs), it would more destructive. Whereas if Gilmour took 10 overs for his 6, that means that there were several more overs bowled at the other end whereby the batsmen may have been able to score of another bowler- so , the Sreesanth spell may actually have more value.

    As we know sheer luck plays a huge role in a sport such as cricket, especially when considering individual innings. Very often, during a particulary good spell a bowler may leak a lot of runs inspite of beating the batsmen ,via edges/squirts through slips and gully. We are assuming that all runs scored are middled by the batsmen and sent straight back past the bowler. So, on a given day with poor luck a bowler may leak a lot of runs via mishits/dropped catches etc , but on another day the very same quality of bowling may result in a pot of wickets- so pure runs leaked may not necessarily be indicative of “destructiveness”…so , if we do decide to use this factor a very small weightage should be given to the same.

  • alex on December 11, 2009, 4:03 GMT

    Ananth - this is in response to Raghav's Dec 10 comment.

    I have suggested an IPF_Finishers table earlier in the comments. Hopefully, Ananth will work on it. However, even that metric (like all metrics) has its loop-holes. E.g., suppose a team scores 240 in 50 overs with an opener X scoring 80 ball 80 and a batsman Y scoring 40 ball 40 after arriving at crease when the score was 170. In this example (commonly encountered), both X and Y have an IPF_Finishers score of 27!! I guess most people will agree that X was more destructive that Y, and that IPF does justice to it. So, two points:

    1. Earlier, I had suggested a multiplicative factor of (batsman SR/ team SR) to smooth out the %TS in IPF (rather than take the sq root of %TS).

    2. As you suggest, I think IPF_Finishers merits its own tables (without the square-root added ... because, now, you expect the batsman to stay there and finish the job). Again, the factor of (batsman SR/team SR) may make sense here.

  • Raghav on December 10, 2009, 18:59 GMT

    Still thinking of the Batting IPF...

    For pure DESTRUCTION, I find Runs * SR which gives the R^2 factor the best measure.

    Now, many of have been looking for an IMPACTFUL player/innings in between the lines. For this we got %TS factor. As discussed this needs to be tempered using Log/sqrt etc..

    It should be tempered to be fair to notouts and lower order batsmen. For chases take %TS = Runs/Runs needed when batsmen comes in. For 1st Innings take %TS=Runs/Runs scored after batsmen arrives at crease.

    The big problem. When only 12 runs are need or a wicket falls in 50th, one can hit 2 sixes. This gives a SR of 600 and TS of 1. I feel we shall have to ignore all scores of below 20 as being insignifcant. I cannot recall many impactful 20s by top batsman. We could also take 10 balls as a cutoff to count the innings. This will work well for Innings analysis where anyways we are keen on the top 100 rather than below 100 which are 0 runs.

    For career we take % impactful inns and their avg IPF.

  • Abh on December 10, 2009, 16:24 GMT

    xolile,goel etc I thought the intention was to get the bowlers "destructive" index( not allround bowling performance)...so,we can simply use similar factors as with the batsmen. And ignore runs leaked, dot balls etc...all these things relate more to a sort of "allround" good bowling performance. If a bowler gets say 5 wickets in 5 overs and leaks pots of runs even 50 or more , the spell can definitely be considered "destructive". So, runs leaked etc are relatively side issues. [[ Abh Gilmour's 6 for 14 has to be distinguished from Sreesanth's 6 for 55. Hence a simple rpb factor has to be part of the index. Ananth: ]]

  • Pelham Barton on December 10, 2009, 15:28 GMT

    Others have commented on the fact that this measure can be calculated in a way that includes the cube of the runs scored. This bothered me when I first saw it, but it is mitigated to some extent by dividing by balls faced and the team innings and I think that on the whole this measure gets innings into the right order. That means that all your statistics about percentage of innings above and below given thresholds are completely valid.

    The only one with which I have a problem is the last column of your Table 2, where you calculate the average (mean) innings power factor. This treats (for example) two innings with IPF 100 each as equivalent to one innings of IPF 150 followed by an innings of IPF 50. I think that is a strong assumption. It might be better to use the median as a measure of each player's typical IPF. You could perhaps add this to Table 2, so anyone who prefers the mean can still use that. [[ Pelham If you take one example (it could even be made more striking at 190 and 10) it might look odd. However taken over 100/200/300/400+ innings, the average value is an excellent indicator of the batsman's effectiveness. See the players in that table. That represents ODI batting at its best. Ananth: ]]

  • love goel on December 10, 2009, 14:46 GMT

    "However, it is interesting that (a) is about a million times less probable than (b). And rarity usually implies value."

    I believe the value in the second case(1.4-1-0-10) comes from the fact that when you bowl out a team, the remaining overs become maidens in themselves, it is just that they weren't bowled and hence not recorded.

    Also the performance A would be very useful anywhere, at any over mark in the inning. However the performance B will be the best performance ever if it came in 1st over, but will totally lose its lustre if it were to happen in 48th over

  • love goel on December 10, 2009, 14:35 GMT

    This is in response to Raghav; "By adding %TS the formula looks like (Runs * Runs * Runs)/(Balls * Team Score)"

    The team score is not independent of Batsmen runs scored. It is more like Batsmen run+other batsmen runs. Assuming Other batsmen score equal to the batsmen being considered(though if you look at the top innings,except for McCullum, all are around 60% of the team score), the formula reduces to Runs*Runs/(Balls*2) which is same as Runs*Strike rate. Very,Very simple to calculate.But again we can't get away with the fact that runs are of paramount importance here.

    The difficulty is to judge the importance of runs scored, whether it Runs^2 or Runs^3. The answer may lie in between , but then Satvir will need a scientific calculator instead of plain one! [[ Goel I am going to work on the basis, at the risk of over-simplication, that the formula has to be simple and easy for anyone to calculate. I will borrow ideas from all but form my own formula on this basis. Satvir should be able to do the calculation that de Villiers' 121 in 85 out of 354 works out to a IPF value of 58.9 in 30 seconds after the end of the innings. One could even do a ball-park calculation of a third of 170 (121*1.4). Ananth: ]]

  • Anirudh on December 10, 2009, 9:43 GMT

    Contd.

    3. The table favours innings while chasing small targets - these are good efforts but are they "destructive"?? The bowlers had done that already. - this needs to be addressed with some criterion such as No. of overs taken while chasing or a qualifier to limit chases below 200. 4. The scale is obviously exponential so it needs to be simplified - i.e. if anything from 10 to 75 is good, it gets confusing. Maybe you can use logs, roots, to make it easier to interpret while maintaining the simplicity.

    Overall, your simple analysis is the best part of the article which makes this measure so user-friendly and representative of true skill. However, rating of each Innings may be a good exercise in itself and for history but cannot be used for rating players. Also, the distribution table is fantastic - but again very difficult to calculate and to represent as a number.

    I hope you can accordingly come up with a IPF that we could possibly see on our TV screens in the future. Thx [[ Anirudh McCullum's innings is an aberration. If you take that innings away there is not a single innings which does not deserve its place. The second table allows Afridi's century to move up. That is not bad. Possibly the solution is to do some root/log work. Ananth: ]]

  • Anirudh on December 10, 2009, 9:36 GMT

    Ananth/Alex: Great article. Watching the T20 game last night, i was wondering if Cricket needs a better "number" than the Average, Strike Rate, Economy Rate etc. This is even more relevant with the advent of T20s where viewers are a little confused as to who is really good? Your analysis is really getting onto something - this number or something similar could be used to "define" or "quantify" batsman similar to a Slugging Number for Baseball hitters... esp in T20s. However, some things need to be addressed: 1. Career NOT Inning-by-Inning numbers must be made - i think this should be possible with the same formula 2. Smaller innings (e.g. 10 balls and below NOT OUT or Batsman coming in after the 47th over but OUT) to be excluded. If a batter isn't good enough to be sent up the order, then we can ignore the innings for this purpose

    Contd.

  • alex on December 10, 2009, 9:10 GMT

    Ananth - this is in response to Raghav. The destructiveness metric was chosen to reflect the F=ma law from physics with m = # runs; a = double time derivative of x, where x = # runs * time spent at crease/(#deliveries faced per time).

    The denominator is the intensity of the threat to destruction. So, a factor of "time spent at the crease" was added to show how solid this destruction was --- threat is zero if you cannot bat. So, it is important to carry on longer, which leads to a higher %TS. So, %TS factor was added ... Ananth's sentence on double-PF and triple-PF says it all.

    However, it is a bit of shock that the average triple-PF does not distinguish between Bevan and Gavaskar. As a compromise between double-PF and triple-PF, you could try using the square-root of %TS. Then, I think, Richards (189*), Jayasuriya (134), and Anwar (194) might be the Top 3 innings. That might also put the openers and the middle order batsmen on a more equal footing. [[ Alex As Anirudh has pointed out and you have earlier suggested we should reduce the impact of % TS by scaling down the range through taking the root of or taking log. The great thing about this blog is the way people come together to contribute in an ego-less manner. A few months back I almost stopped any writing here because of the very strong and adverse comments made by set of readers. Now that has gone off (for how long, I wonder). Ananth: ]]

  • Xolile on December 10, 2009, 8:58 GMT

    Ananth,

    From a team perspective I agree that (b) is certainly preferable and should therefore be considered more destructive.

    However, it is interesting that (a) is about a million times less probable than (b). And rarity usually implies value.

    You could also say that both wickets and dot balls are achievements from a bowler’s perspective. The most a bowler can achieve in an ODI match is to bowl 60 dot balls and take 10 wickets. It’s a bit like winning 10 gold and 60 bronze medals compared to winning just the 10 gold medals.

  • Raghav on December 10, 2009, 6:34 GMT

    By introducing %team score you have ventured away from the goal. It is neither an analysis of "destructive" nor "powerful, impactful, best etc.". It is somewhere in between.

    For destructive I would go with just plain Runs*SR. A more 50 off 25 is equally destructive as a run a ball 100.

    By adding %TS the formula looks like (Runs * Runs * Runs)/(Balls * Team Score). Unknowingly it just adds to the importance of Runs 3 times and you see that most destructive innings are the really high ones. [[ Raghav You are correct. In fact I have set up Runs(cubed) / Balls * TS as my cross-check factor. It was done only to moderate the explosive effects of the Runs * SR value. You will see that the alternative is also provided as a table. However where this triple product helps is in the Player wise consolidation. You can see that the Player IPF average is a clear pointer to ODI greatness. Also we have been able to do the cut-offs (lt 2 and gt 10) in a more comfortable manner. In other words, by itself, the double-product factor is more meaningful while as a means of measuring ODI batsmen across careers the moderated triple-product factor is more meaningful. Thanks for your insight. Ananth: ]]

  • Xolile on December 9, 2009, 17:04 GMT

    @ love goel

    Thank you for considering the formula I have proposed. In response I would like to ask you a question. Which of the following do you consider the better performance?

    a) 10-10-0-10 b) 1.4-1-0-10

    [[ X: I am answering on my behalf and not Goel's. Both are bloody good performances and I would never venture to select. Although it is obvious that innings (a) would have lasted a longer time than (b). From the point of view that the overall innings must be shorter in (b), hence lower runs would have been conceded, (b) would be considered better. Having said that, these are extreme cases and one could easily do some outrageous guessing that bowler (b) came in at 300 for no loss in the 48th over and polished off the innings at 310. Bowler (a) might have opened the bowling and bowled straight through for 50 all out. That is the problem with extreme examples. Ananth: ]]

  • love goel on December 9, 2009, 15:12 GMT

    IPF = 48 – 48*R/B + 0.2B – 0.2R + 7W^1.18 1. This factor doesn't consider strike rate at all. 2 bowlers with with equal wickets , say 5-2-30-5 and 10-4-60-5 are ranked equally. I would say less overs is preferable any day,any time as it also means less runs(and in most cases early dismissal of the batting team). Strike rate needs to be included.

    One should also consider whether wickets were taken in a bunch or after intervals. If you were to see Hattricks compared to 3 wickets hauls, the percentage of wins will be significantly higher in hatricks(although in case of hattricks bowlers may have taken more wickets, in which case they need to comapred to 4/5/6 wicket hauls). I would say award extra points for wickets in consecutive over(same over may be considered consecutive over). But again , this info may not be available on scorecards , so it may not be possible [[ Goel As I have already said, I will take into account all contributions, make sure that all key factors are included and then come out with a Bowling IPF which I can see would be a greater challenge than the Batting one. Ananth: ]]

  • Surya Narayanan on December 9, 2009, 11:55 GMT

    This analysis is excellent. But it seems to overlook the impact of truncated matches on the strike rate. Both matches with reduced overs where such decision is made before an innings starts and those where overs are reduced midway have different effect on strike rates of the batsmen & the teams. [[ Surya D/L does not lend itself to any type of structured analysis. And if I try to fit in a formula which is common for all matches., the D/L changes will distort the other matches which are over 90%. Ananth: ]]

  • Surya Narayanan on December 9, 2009, 11:19 GMT

    Regarding 'etc.' & 'et al', the former means 'and other things' & hence should be used to represent anything (including living beings other than humans) while the latter is specifically to mean 'and other people'. [[ Surya Almost always I use "et al" when I refer to people. I use that in the sense "amongst others". Thanks. Ananth: ]]

  • Ananth on December 9, 2009, 11:14 GMT

    I have updated the article with the following changes. 1. Lowered the bar to 1977 so that Clive Lloyd is in. 2. Show the Team Score in the main table. 3. Treat a 0 in 0 as a still-born innings, not a failure. 4. Change Kapil Dev's innings from 150 to 138 balls. The download files have also been uploaded to thirdslip.com so that the users can download the text files without problem.

  • MUHAMMAD Imran Dada on December 9, 2009, 11:11 GMT

    Where is Javed Miandad ???????????? [[ Imran He is in all the tables if you look for him !!!!!!!!!!!!!!!! Ananth: ]]

  • Xolile on December 9, 2009, 8:36 GMT

    (Continued)

    10 BAW Mendis (SL) [8-1-13-6 v India] IPF:99.99 11 M Muralitharan (SL) [10-1-30-7 v India] IPF:99.55 12 Shoaib Akhtar (Pak) [9-1-16-6 v New Zealand] IPF:99.36 13 CA Walsh (WI) [4.3-3-1-5 v Sri Lanka] IPF:98.18 14 BC Strang (Zim) [10-2-20-6 v Bangladesh] IPF:97.99 15 M Muralitharan (SL) [10-3-9-5 v New Zealand] IPF:97.76 16 SE Bond (NZ) [9-3-19-6 v India] IPF:96.1 17 A Kumble (India) [6.1-2-12-6 v West Indies] IPF:95.42 18 A Nehra (India) [10-2-23-6 v England] IPF:94.99 19 SE Bond (NZ) [10-2-23-6 v Australia] IPF:94.99 20 M Ntini (SA) [9.3-4-22-6 v Australia] IPF:94.46 21 HK Olonga (Zim) [8.2-3-19-6 v England] IPF:93.95 22 Waqar Younis (Pak) [10-0-36-7 v England] IPF:93.55 23 PV Simmons (WI) [10-8-3-4 v Pakistan] IPF:92.94 24 DT Johnston (Ire) [10-4-14-5 v Canada] IPF:92.76 25 Aaqib Javed (Pak) [10-1-37-7 v India] IPF:92.55

  • Xolile on December 9, 2009, 8:35 GMT

    Okay, my last attempt. The new formula is as follows:

    IPF = 48 – 48*R/B + 0.2B – 0.2R + 7W^1.18

    It rewards bowlers progressively for wickets taken. It also awards dot balls and boundary prevention. It can be meaningfully applied to any bowling figures. Although it does not refer to DL tables it is based on the DL principle of “resources”.

    Below is the top 25 of all time. All the InnSpells mentioned by you and the readers feature in the list. Let me know what you think.

    1 WPUJC Vaas (SL) [8-3-19-8 v Zimbabwe] IPF:116.22 2 AJ Bichel (Aus) [10-0-20-7 v England] IPF:109.55 3 GJ Gilmour (Aus) [12-6-14-6 v England] IPF:108.25 4 GD McGrath (Aus) [7-4-15-7 v Namibia] IPF:105.81 5 Imran Khan (Pak) [10-2-14-6 v India] IPF:103.99 6 MF Maharoof (SL) [9-2-14-6 v West Indies] IPF:101.54 7 SB Joshi (India) [10-6-6-5 v South Africa] IPF:100.76 8 CEH Croft (WI) [9-4-15-6 v England] IPF:100.45 9 Azhar Mahmood (Pak) [10-2-18-6 v West Indies] IPF:99.99 [[ X: An excellent attempt. Let me look at it along with all other suggestions. Thanks. Ananth: ]]

  • Abhi on December 9, 2009, 7:00 GMT

    Right thanks, just to put it in a formula: bowler IPF= “wickets taken” * "No.of wickets / total wickets that fell in the inn." / “stike rate”,

  • Abh on December 9, 2009, 6:47 GMT

    V.good article on the batsmen- with the usual suspects starring. As rgds. the bowling - for "Destructiveness"...then we can sort of "invert" the batsmen parameters and still keep it simple.

    So, if the batsmens Innings Power Factor (IPF) = Runs scored * Scoring rate * % of Team score, Then for the bowlers we may use “wickets taken”, “stike rate”, "No.of wickets / total wickets that fell in the inn." [[ Abhi/All others who have contributed on Bowler PIF. Thank you all. I am not going to do this in a hurry. I will look at all the contributions peacefully, in detail, well after my next post (an important and long-awaited one, I can assure you). I am immersed in so many tasks over the next 10 days and want to do justice to a complex algorithm. Ananth: ]]

  • Praveen on December 9, 2009, 6:41 GMT

    I enjoyed this article, credit to the two authors, fantastic stuff. Inspired by Chris Gayle's knock the other day, I wonder if we could go in the other direction? E.g. in test matches, the best rock solid innings where a batsman bats through in difficult conditions. But perhaps instead of strike rate use a function of the number of wickets that fell whilst that player was at the crease?

    Just a thought.

  • Pankaj Joshi on December 9, 2009, 5:27 GMT

    Thanks Ananth. On the bowling discussion, KiwiMarcus has the right idea. Consider a batting side putting up 300-5 in 50 overs and one bowler getting 2 for 40 in his ten overs, it definitely deserves weightage. Lastly one could give a factor to the game's result, bowlers are expected to win matches within the day's run budget. Would appreciate more thoughts. In batting, Ol Man River Lloyd is so visible for all of us in 20-20 hindsight. Would it make sense to give the stage a weightage? I am being mean but the World Cup is by far a huge stage and the impact of Lloyd's knock in 1975 on the result did go much beyond the numbers.

  • alex on December 9, 2009, 3:54 GMT

    Ananth - this is in reply to BJ's comment. It is also important not to give the wicket away early. I believe a truly destructive batsman should also face enough number of deliveries (> 50/6=8.33 overs ... assuming 6 specialist batsmen/team). As regards Viru, consider a hypothetical opener who hits 5 sixes in 5 balls and gets out on the 6th with a handsome IPF of 150. Many opponents would be glad to bowl to him rather than a 60 ball 60 guy. This is an extreme argument but should get the point across on why a 60 ball 60 could be more destructive that a 30 ball 40.

  • Sesha on December 9, 2009, 3:45 GMT

    I strongly think that bowling spell destructiveness should not be reasonably measured without considering Batsman quality.

    Run scored by batsman before the wicket can be ignored as it would be factored into the Bowlers Economy rate.

    Ananth did you have a chance to look into the Bowler power factor I mailed you? You can help improve on it if you think my formula could lead to something meaningful..

    Taking into consideration the quality of wickets would eliminate the skew in freak spells like that of Walsh's where he took batsman # 6,8,9,10,11 whereas Phil Simmons took Batsman # 1,3,4,5. In my opinion the destructiveness of both the spells should not be much diff if not weighing more towards Simmons'

  • Jeremiah Kuruppanattu on December 8, 2009, 23:22 GMT

    This proves that Dravid was always a great one-day player, even more effective than Sehwag. Dravid's strike rate is much higher than most people think, about . After settling in he has always hit more than his own fair share of boundaries. Average of 40 with strike rate of 70 is better than most of India's cuurent players. When you see the number of runs he has scored over the years, he should definitely still be on the team. How do you drop Dravid for a flash in pan player like Kohli. Dravid has certainly declined but it is heavily exaggerated. And who has complained about Dravid's fielding? Dravid is certainly not the best fielder, but he is far from the worst. The sheer number of catches he has taken proves that his reflexes are on par with anyone's. It is a shame that his ability is being wasted just to say that India is a young team.

  • KiwiMarcus on December 8, 2009, 22:15 GMT

    For the bowling analysis, I'd argue that the 3 factors that go into making a Destructive bowling spell are a lot of wickets in a short space of time, in a spell that stands out from the other bowlers. To try and capture that, perhaps a measure like:

    Index = (wickets/balls) * (innings runs/runs against bowler)

    I haven't put these into any significant preliminary analysis , but an average 10 over spell of 1/40 out of a total of 240 gives a score of 0.1, a spell of 1/100 out of 300 gives 0.05, while for some of the examples above:

    Kumble 6/12 - Index = 1.66 Gilmour 6/14 - Index = 0.55 Walsh 5/1 - Index = 10.19 Simmons 4/3 - Index = 1.80

    Note that this puts perhaps a heavier premium on runs than might be desirable, but to me the rankings of the mentioned spells look sensible - Walsh's spell was freakish in such a short space of time, Kumble and Simmons were amazing in different ways, and Gilmour's excellent but less 'destructive' as it was spread over a longer spell.

  • Lallu lal on December 8, 2009, 21:53 GMT

    One of my fav bowling performances (must have been one of the more destructive ones) Imran's 6/14 against India in Sharjah. I think his figures read 10-6-14-6. I think India scored 125 and still managed to win when Pak folded around 87. Oh those days!

  • BJ on December 8, 2009, 19:20 GMT

    I'm a constant reader, though this is my first post. Excellent analysis and I agree with everything except that I think it's more an 'influence' or 'MVP' analysis than destructiveness analysis. I've read your earlier response where you stated that if you exclude % of TS then '30 off 10' and '90 off 90' will get the same scores. But, I'd argue that there's nothing wrong if they get same scores when analysing 'destructiveness'. If anything, I would argue that 30 off 10 is more destructive than 90 off 90. 90 off 90 may have been more valuable, but not destructive...

    I also have a personal agenda in this. My favorite Sehwag doesn't figure a lot if you apply % of TS since he gets out in his 40s and 60s in ODIs. Wouldn't you agree that he is one of the most destructive players currently, even if not consistent enough in ODIs?

  • Xolile on December 8, 2009, 19:11 GMT

    Ananth,

    I have tweaked my suggested IPF bowling formula to take into consideration InnSpells < 60 balls (such as Walsh’s 1 for 5 in 4.2 overs):

    IPF = 48 - 48R/B + 0.2B – 0.2R + 6W

    B = Balls bowled R = Runs Conceded W = Wickets taken

    Applying this formula, the 10 most destructive ODI InnSpells of all time are:

    1 GJ Gilmour (Aus) Inn: 12-6-14-6 IPF: 86.27 2 SB Joshi (India) Inn: 10-6-6-5 IPF: 84 3 WPUJC Vaas (SL) Inn: 8-3-19-8 IPF: 82.8 4 Imran Khan (Pak) Inn: 10-2-14-6 IPF: 82 5 AJ Bichel (Aus) Inn: 10-0-20-7 IPF: 82 6 CA Walsh (WI) Inn: 4.3-3-1-5 IPF: 81.42 7 PV Simmons (WI) Inn: 10-8-3-4 IPF: 81 8 M Muralitharan (SL) Inn: 10-3-9-5 IPF: 81 9 MF Maharoof (SL) Inn: 9-2-14-6 IPF: 79.56 10 CEH Croft (WI) Inn: 9-4-15-6 IPF: 78.47

    I am happy with the end results. All three the InnSpells you have mentioned feature in the top 10. Hopefully the formula is still simple enough. [[ X, My only concern is taht all these are outstandingly economical wicket-taking spells. I am concerned that the two 7-wkt hauls by Aaqib and Murali against India are not there. Maybe we should do the wickets differently. Give progressively higher weight for wickets 6/7/8. Ananth: ]]

  • love goel on December 8, 2009, 17:50 GMT

    Thanks. I also thought it wouldn't be practical to go Duckworth-Lewis way. But still suggested as it takes care of a lot of factors in itself. May be one-day. For simplcity,nothing can beat runs/overs/wickets based formula

    And I expect to get a few surprising spells which nobody really cares for, like when Kenya bowled WI for 73 in a WC 1996 match. Or when India crumbled from 98/1 t0 120/8 in WC 1996 semi-final.Looking forward to some unknown great spells

  • love goel on December 8, 2009, 17:19 GMT

    I am sorry for the comment. May be ,I did read too much into the statement. No intention to question anybody's integrity.

    I have a totally different idea about bowlers IPF, which I feel is very simple and most difficult at the same time. One can use DL method to figure out how much resources the batting team had before the Bowlers started over and after he finished the over. Sum up the resources used for the bowlers spell. Divide the number of runs given away/resources used. There you get the value of the spell.

    Though I believe this will be computationally prohibitive, may be it can be used to compare only a small set of bowling spells on a more detailed level [[ Goel If xyz or abc or pqr (you know them) had made the comment I would have simply trashed the comments and gone on to the next one. But I respect your comments, hence the (somewhat strong) reaction. No offence meant from my side. Unfortunately the analysis you have suggested requires a lot more information than found in the scorecard and is either proprietary (and hence unavailable) or not available at all. Hence the need to use the available factors, which are wickets, bowling accuracy, strike rate, team wickets, resources (indirectly) etc. At the same time avoid the Ratings factors such as quality of wickets, team strengths, batsmen quality, result, pitch conditions, match importance etc. I don't want to get out of Alex's somewhat simple and effective starting premise. Ananth: ]]

  • love goel on December 8, 2009, 14:27 GMT

    For bowlers, I think it is also important to take into factor whether wickets were taken earlier in the innings or later. Sometimes there is a batting collapse in the slog overs. These wickets can't be considered equal to a wickets in the first few overs,for they have much more impact on the innings.

    Also you should consider whether wickets were taken in one single spell or in many spells. for me, Malinga's 4 in 4 balls remains one of the most destrcutive spells. Almost all of the bowling world record like Aaquib Javed, Muralitharan, Vaas, Kumble 6/14 vs Wi, Andy Bichel performance vs England in 2003 wc,not to mention Sunil Joshi bowling vs SA where he bowled maidens after maidens etc should be high in the ranking.

    "Probably three of the most destructive bowling performances of all times was Gilmour's 6 for 14, Walsh's 5 for 1 and Simmons' 4 for 3. Apply your formula to these and see."

    Sorry to say, but didn't you just biased the hypothesis to draw a particular result [[ Goel You read too much into statements designed to help people in their thought process. You are completely and utterly wrong if you think that I would bias my analysis to bring out a particular result. It is a conclusion which insults my integrity. These three performances, if you care to go through them, are examples, just as Joshi's might be of bowling performances which could be called devastating. I wanted to give a few unknown spells for consideration. I could very well have exampled Aaqib Javed's 7-wkts and Murali's 7 wkts against India. These are just examples. I would appreciate if you, who I think is a balanced and insightful reader, avoid making such comments. I never EVER start with a result and twist the analysis to arrive at that result. Ananth: ]]

  • Gizza on December 8, 2009, 13:28 GMT

    Very insightful analysis. However I would change "Destructiveness" to "Influence". The last % of innings factor is indirectly saying how influential that innings was on the match, not that it was destructive.

    eg. If a Sehwag or Tendulkar race to 100 in 60 balls then the rest of the Indian batsmen falls like dominos, it is not more destructive than a Gilchrist scoring 100 off 60 balls but then the Australian middle order backing him up and racking a score of 350. The bowling side would regard both Sehwag's and Gilchrist's innings as equally destructive and probably both as equally psychologically deflating. Just because eg. Ponting and Symonds continued the destructiveness does not make the initial innings less destructive. [[ Gizza What you say makes sense. However the % TS was meant to moderate the explosive effect of Runs x Scoring Rate. Otherwise we will have a 30 in 10 balls ending up with the same PIF as a 90 in 90. Ananth: ]]

  • alex on December 8, 2009, 12:33 GMT

    Ananth - this is in reply to Som's comment.

    1. I already have suggested a formula on the lines Som mentioned ... I have suggested a table based on it titled "IPF_Finishers".

    2. Openers may get an opportunity to record a big IPF score but a top 5 batsman almost always gets an opportunity to score > 10 on this metric. E.g. a 50 ball 50 in 250 from 50 overs rates 10 on this metric. To get this opportunity, a batsman has to arrive at the crease by the 35th over. In most cases, a team loses 3 wickets by the 35th over. So, I suggest that you look at the frequency with which a batsman scores over 10 (provided by Ananth). If a posn 3-11 batsman scores > 10 with >20% freq, that is quite exeptional ... notice the results for Bevan, Dhoni, Symond, and Hussey! [[ Alex/Som The table is ordered by the difference between the two % values. Hence one would have to see the complete table to get an idea of the > 10 % figure. I would say that a value above 25% represents a top-class middle order batsman. See how good Ganguly and Dean Jones are. Ananth: ]]

  • romel roy on December 8, 2009, 10:58 GMT

    No. 1 Table says it all; Excellent analysis Ananth; congratualtions to you and alex.

  • alex on December 8, 2009, 10:27 GMT

    Ananth - I see no mention of Clive Lloyd in these tables. He was absolutely destructive (and, may be, the most important player of the WI champion teams). I understand that this metric will not do justice to him (and players like Bevan) but how come he doesn't figure at all? Even Gus Logie features in Table 2 but not Lloyd!! [[ Alex You have made me feel like a fool. How did I miss Lloyd. Problem is he scored 1977 runs and just missed the cut. Within 24 hours the bar will be lowered by 23 runs and new tables put up. To think that there is Vaas there but not Lloyd. In the bargain, Sanjay Manjrekar and Vaughan would also get in. Ananth: ]]

  • Vijay on December 8, 2009, 9:24 GMT

    Ananth, this message is not relevant to the article. I did not know of any other way to get in touch with you hence this: I wanted to suggest an article on "the world's best batting line up". During the Ind-Aus series last year, India's top 7 were Viru, Gauti, Jammy, Sachin, Lax, Dada, and Mahi. Was this the best batting line up in history? Against the game against SL India had Yuvi in place of Dada. Also where does the Aussie line up of 2004 fit: Hayden, Langer, Ponting, Martyn, Waugh, Katich, Gilchrist? Probably Katich brings this down but what about the 2001 team of Slater, Hayden, Langer, Waugh, Waugh, Ponting, Gilchrist? I know Hayden discovered himself on that tour so he wouldn't have been all that good till then, but still. And what about the WI - Greenidge, Haynes, Richardson, Viv, Lloyd, Logie, Dujon? [[ Vijay I have done this earlier but will attempt to do it again with newer factors and better methods. I will do this after clearing my plate. Ananth: ]]

  • Pankaj Joshi on December 8, 2009, 8:18 GMT

    Hi just one possible correction. Kapil Dev's 175 had come, if memory is right, in 138 balls. If that is the case, his knock would move up the pecking order because its IPF would be 146, as Richards and Ejaz. [[ Pankaj The 150 was picked up from an earlier Frindall book. I will correct and re-post the tables within a day or two. Ananth: ]]

  • Kishore on December 8, 2009, 7:58 GMT

    Awesome man....can't get better than this...excellent work done and well appreciated. Lokking for more from you.

  • alex on December 8, 2009, 6:52 GMT

    Ananth - I forgot to point this out earlier. The IPF score can also be viewed on a log scale with, say, 2 as the base (since 2 is a reasonable cut-off to weed out negligible contributions).

    Thus, IPF of 2 is 1 on the log scale, IPF of 4 (which is an OK performance) is 2 on the log scale, IPF of 8 (which is a strong performance) is 3 on the log scale, etc. Only 11 performances rate more than 7 on this log scale while 187 rate more than 6 on the log scale.

    To even out the advantage enjoyed by the openers or simply a longer innings (in terms of opportunity to have a higher %TS), the IPF could be rescaled by multiplying it with the following factor: SR/(rest of the team's SR). Thus an 100 ball 80 in a score of 300 (in 50 overs) rates about 18 on the old IPF while it rates only 13 on the revised scale ... however, this construct unnecessarily punishes good batsmen in good teams while rewarding good batsmen in poorer teams. [[ Alex The great thing about your factor was its simplicity. As the first comment mentioned nicely, anyone watching an innings on TV could work out the PIF in seconds. I might use the Log scale for the other analysis I have mentioned to you in which I will be using PIF. Ananth: ]]

  • alex on December 8, 2009, 5:49 GMT

    Ananth - this is in reply to Sesha. I think, Bevan was probably the best ever ODI finisher. My dream team will have him in it anyday. To rate the finishers better, I suggest the following separate measure:

    IPF_Finishers = runs * SR * (runs/RF), where

    1. RF = runs scored after the batsman arrived at the crease (while batting 1st) 2. RF = runs needed runs when batsman arrived at the crease (while batting 2nd)

    For added delight, multiply this IPF by:

    - SR/(team SR before the batsmen got to crease) ... while batting 1st

    - SR/(required SR when the batsman got to crease) ... while batting 2nd

    I think Gibbs' 175 was the best chasing innings ever, given the improbability of the run chase; SRT might have bettered it if he had recd the strike when he was hot (@169) in his 175. [[ Alex You must have seen some of the replies. This information (when did the player start the innings) is not available at all for the first 1000 or so matches. Afterwards in a sporadic manner until match 1500 or thereabouts. Afterwards it is fine. However the score is available and the first part is possible. Let me look at this. Possible that Gibbs might just about edge out Dhoni. Ananth: ]]

  • Marcus on December 8, 2009, 4:01 GMT

    One thing I've noticed is that a lot of the time the top scorer in the innings actually scores slower than the team (eg. Shaun Marsh recently scored 112 at about 100, while the team scored well over 300). This makes sense, as the centurion often will have to be the "lynchpin" of the innings and therefore the size of the innings will be the result of runs scored quickly by suppoting batsmen at the other end. So I think a column could be added for the team total in the first table, just to give us a more complete idea as to the style of the innings. [[ Marcus Good idea although the % of TS column gives a clear, although indirect, indication of the same. Ananth: ]]

  • Unni on December 8, 2009, 1:41 GMT

    Great article, Ananth... and idea was simple...

    However, I had the same feeling as xolile when I read this. I know I'm trying to make things complicated. If you see the tables, you can see most of them are openers. They can be destroyers, agree. But that means, the subsequent ones cannot be. So, if we take the team share factor, consider 280/2 scoreline and somebody scored 120 in 100 balls. And consider 280/8 and the same person scored the same. Now, in my opinion, the second performance has more destructiveness than the first one. If the number of players batted comes into picture, the late order cameos will have some more weightage. (I can imagine your response on this :-)) [[ Unni I agree that a 120 in 100 out of 280 for 8 is FAR MORE VALUABLE than out of 280 for 2. That will be the case when I do a Ratings exercise. However both are equally destructive because they are analyzed within the same three factors. For that matter the 280 for 8 team might have had far better bowlers than the 280 for 2 team. Ananth: ]]

  • Sesha on December 8, 2009, 1:20 GMT

    Thought on Formula for ODI bowling performance

    Innings Strike rate for a bowler * Innings Economy Rate for a bowler * % of ER on Batting team Run rate for the innings * Sum of Batting positions of wickets

    Unlike Batsman who wants all Key Performance indicators (KPI) to be high, Bowlers are slightly diff, they want everything to be low except number of wickets. Hence SR is taken in the formlula. In ODI performance SR alone does not count ER is equally important ans so it is also used. To measure how good a bowler was compared to other bowlers in an inning and to see his impact % of ER on Innings Run rate is used. A bowling spell should also be rated based on quality of batsman whose wicket was taken. Luckily the lower the batting position number (1-6) the better the batsman. Hence that factor is also considered.

    Let me know your thoughts.... [[ Sesha/Xolile You have given two totally differeent Bowling factors. Probably the solution lies in a combination. Can both of you do something. Probably three of the most destructive bowling performances of all times was Gilmour's 6 for 14, Walsh's 5 for 1 and Simmons' 4 for 3. Apply your formula to these and see. You could mail me the results. Ananth: ]]

  • Abs on December 8, 2009, 0:58 GMT

    No doubt Richards and Abbas were great players. But one point that is often forgotten is that a number of matches that they played were 60 over innings compared to the 50 that modern players get (almost 20% more!). I wonder what their respective averages are if you only count 50 over matches. [[ Som Not all matches were 60 overs. The 50 over limit came in sporadically as early as 1982. Ananth: ]]

  • Matt G on December 7, 2009, 22:04 GMT

    This is completely off-topic, but I'm asking it here because I can't see a way to send a general email question to the "It Figures" folks.

    Why is it that the scoreboard in the background of the "It Figures" column head is clearly from an (American) football stadium? The visible parts of the scoreboard show "qtr", "down" and "visitor", only the last of which is relevant in cricket. With so many cricket scoreboards available, why, oh why was one from a grossly inferior sport (hey, I've lived in the US for 30 years, i've earned the right to say that!) chosen?

  • Som on December 7, 2009, 21:09 GMT

    Instead of the '% of runs scored' shouldn't it be '% of runs scored of the runs the team scored after the batsman got in', that way batsman coming lower down the order will have better representation. At this point, the result is so top heavy and big innings played by openers or top order batsmen allow them to tick all the parameters.

  • Xolile on December 7, 2009, 18:29 GMT

    Ananth,

    May I suggest the following simple formula for bowling: Balls Bowled – Runs Conceded + (Wickets Taken * 5)

    If effectively rewards wickets, dot balls and the prevention of boundaries.

    The possible problem with the formula you have suggested is that a bowler that bowls 10 overs and concedes 10 runs but takes no wickets will score zero points.

  • love goel on December 7, 2009, 18:13 GMT

    Ananth, I was never asking you to leave out Not Out innings. But like while calculating batting average, it is run scored/number of innings out, for calculating career IPF average(Table 2) it should be Total IPF(all innings)/number of innings out. No need to change number of innings played. For that even 0/0 is considered an innings. So 189* will add to IPF, innings played but not to innings out.

    Also Xolile exmaple is very valid. 0/0 is just a extreme case of the same example. Imagine a batsmen coming only for the last ball , hitting a six to take the team total to 300. His IPF will be 6*6*6/300 = .72. Go tell that batsmen he underperformed in that innings! [[ Goel An extreme example can be used to invalidate anything. Anyhow the guy who hit a six off the only ball he faced has done well, that is all. He will have a strike rate of 6.00, the highest in ODI history. Let him be satisfied with that. Ananth: ]]

  • Xolile on December 7, 2009, 17:01 GMT

    As always it is an interesting idea and a well written article. Nevertheless, this measure does through out a few apparent anomalies.

    For instance, a batsman who scores 20* runs of 15 balls in a successful chase of 270 will score less than 2 IPF points. He is therefore deemed to have failed.

    And if McCullum replicated his table-topping innings in the famous 438-run match between Aus and SA, he would have scored 41.75 IPF points instead of 192. [[ Deon You probably have hit the nail on the head. McCullum's innings is no.1 ONLY because it was out of 95. This is not a list of the top ODI innings. In which case all other factors come in. If you remember Alex's original name "Destructiveness Factor", could there have been an innings more devastating than McCullum's. Nzl: 95 scored in 6.0 overs against an above-average bowling attack (Mashrafe/Shahdat/Shakib/Razzak, there have been many worse attacks amongst top teams over the years) and McCullum 80 in 28 balls. More than 50% of the deliveries he received were hit for boundaries. Consider from that point of view. Ananth: ]]

  • Lahiru Mampitiya on December 7, 2009, 16:02 GMT

    Excellent work Ananth, Numbers never lie :) It would be great to see the same analysis for the bowlers too. [[ Lahiru Pl come out with your suggestions on a similar formula for bowliers. Ananth: ]]

  • love goel on December 7, 2009, 15:54 GMT

    "An innings is counted only if he gets to start his innings. In other words, he could be not out with 0 in 0. That is an innings if it appears in the scorecard as such"

    Why count a not out inning? It may be valid when you are doing an inning based analysis. But on a career averages based analysis(Table 2),shouldn't the not out inning be ignored? After all,cricket batting averages also exclude not innings.

    Also in case of an inning of 0/0, if you take IPF to be zero, it becomes an IPF<2 innings for the batsmen, without any input from the batsmen. Why not exclude this innings from the list of bad performances(Table 4)? Although any not out innings with good IPF has to be still measure in; we cant penalize batsmen for staying not out. [[ Goel That is the convention and I do not want to change it. If I do that there will be inconsistency between my database and other recognized databases. If a batsman walks in, he has played an innings, that is convention. How can I exclude Not outs. It could be 189* or 0*. But your last point is valid. A still-born innings need not be considered a failure. I will make that correction and post the changed table. Ananth: ]]

  • love goel on December 7, 2009, 15:37 GMT

    The formula is really simple and elegant. It leads itself to great result, especially Table2.

    However I had a question on the formula. If %TS is included, why not include Batsmen Strike Rate/Team strike rate also. This will balance out the effect of high scoring matches, just like %TS score

    And Ananth,For table 4 ,you have written "The more I see the table the more I feel that this is the single table which encompasses the ODI greats in full"

    To me Table 2 seems to be more important if you wish to look at great ODI batsmen. Table 4 to me, favours consistent players over inconsistent ones. Any particalur reason you chose table 4? [[ Goel No particular reason other than the fact that at one shot we have rewarded high level of success and low level of failure. But I can see the table 2 being the more relevant one. Ananth: ]]

  • Sesha on December 7, 2009, 15:35 GMT

    Contd...Observations

    6. Totally agree with Mr.A on table 4. Except for Z.Abbas and KP, I guess Bevan, Gilly could be replaced just to see the top 10 batsmen..:-)

    7. If my eye-balling is correct Dhoni 183 & Gibbs, Sach's 175 have been scored while chasing...Mr.A, could you consider adding any weightage to chasing...which anyday makes a good score even sexier...

    Some thoughts reading the blog:

    1. Mr.A, Could you come out with what is the least strike rate a batsman shall have based on avg winning scores. I remember you doing a similar work for T20s.

    2. How has SR & Avg of top batsman developed over the years...Sachin & Sanath with SR of >85 even after 400 matches is one of the unbelievable numbers...more so Sach's because of his Avg too.. [[ Sesha First Zaheer. Poor guy played for 12 years and appeared in 62 matches while Pakistan played 88 matches. Contrast this with Pollock who played for the same 12 years and played in 303 matches while SA played 335 matches. Zaheer did wonderfully well in the matches which were available to be played. Bevan has not played many such explosive innings. As I have already explained to Arjun, download the data file, export to Excel and you can do what you want. Separate the innings as first or chasing and so on. Dhoni's is the best chasing innings ever, by a mile, no, 5 miles. If you have a downloading problem I can upload to Thirdslip.com as a text file for easier downloading. In one of my earlier articles titled "ODI matches - an analysis through 4 decades", I have done an evolution comparison of S/R and Avge over the years. Although now I can do a far superior article. Ananth: ]]

  • Deepak Shah on December 7, 2009, 15:24 GMT

    This is a great analysis. One interesting observation that I had was the absence of Michael Bevan from all of the tables. How do you explain that?

    Maybe, if you brought in the "not-out" factor, it would change the lists quite a bit, and make it more fair for those batsmen who batted later in the innings. [[ Deepak Bevan was a great finisher. He was the 80 in 100 batsman, not the 120 in 80 batsman although he has done that also. Also being not out does not get into the equation. Pl remember that we are analyzing individual innings. Ananth: ]]

  • Sesha on December 7, 2009, 15:09 GMT

    Hi Ananth & Alex,

    One of the most satisfying articles to read because I was able to connect to the article and ideas so well. Special mention to Ananth for using appropriate words to get the message across in a slightly tricky article.

    In my first comment, I would like to mention a few of my observations... 1. Except for McCullum & Sehwag's innings All other entries in top 20 are their respective highest followed by 2/3 highest.

    2. All batsmen in the top 20 have a SR of >80 except Kirsten and Gower. Gower's SR of 75 is exceptional for the era he belonged to

    3. Notable Exceptions in the first table from the modern era who are famous for their destructive abilities are Afridi, Gayle, Treschk, Graeme Smith, KP & Hayden who aptly feature in 2nd table

    4. With due respect to Z.Abbas, he has played in too little innigs to be heading table 2. Mr.Ananth you can consider some entry barrier.

    5. Another proof y Sach & Viv are the best ODI batsmen ever..

    Contd...

  • Arjun on December 7, 2009, 13:29 GMT

    I think IPF should be done separately for Ist and 2nd innings since while batting in 2nd innings not always full resources(overs) are utilised. [[ Arjun I have given the innings info. So it would not be difficult for anyone to view these separately. The problem is that I already have 5 tables and the article would have been quite lengthy. The other point is that the % TS is based on Runs which is a complete measure even for the second innings. Ananth: ]]

  • Pranesh on December 7, 2009, 13:21 GMT

    "It should be noted that this factor, being a pure batting one, does not take into account team strengths, bowling quality, pitch type, innings status, result et al." Sorry, but it should be result etc. et al is used when you cite one example. [[ Praneshgreat I have always used et al as "and others". Here is the Dictionary definition. "" et al means roughly "and others". It is written at the end of a list of names to indicate that others are related to the same subject. "" Ananth: ]]

  • Devendra on December 7, 2009, 12:48 GMT

    Very Interesting way of measuring "Destruction". Maybe ..Instead of considering the team score .. the score made by the team when the Batsmen was on field may hold more relevance. [[ Devendra The correct information on who was out at what score is available only from around match # 1000 onwards and with complete accuracy from match # 1500 onwards. Otherwise it is a good suggestion. Ananth: ]]

  • alex on December 7, 2009, 12:28 GMT

    Ananth - in computing %, is an innings counted in your analysis if a batsman doesn't get to bat in it? [[ Alex An innings is counted only if he gets to start his innings. In other words, he could be not out with 0 in 0. That is an innings if it appears in the scorecard as such. Ananth: ]]

  • satvir on December 7, 2009, 12:19 GMT

    amazing, man. that is great. now, i will watch every match having calculator in my hand. I am really stunned at effectiveness and simplicity of this formula [[ Satvir Thanks. I think Alex deserves a pat on his back for the very words you have used, effectiveness and simplicity. Ananth: ]]

  • No featured comments at the moment.

  • satvir on December 7, 2009, 12:19 GMT

    amazing, man. that is great. now, i will watch every match having calculator in my hand. I am really stunned at effectiveness and simplicity of this formula [[ Satvir Thanks. I think Alex deserves a pat on his back for the very words you have used, effectiveness and simplicity. Ananth: ]]

  • alex on December 7, 2009, 12:28 GMT

    Ananth - in computing %, is an innings counted in your analysis if a batsman doesn't get to bat in it? [[ Alex An innings is counted only if he gets to start his innings. In other words, he could be not out with 0 in 0. That is an innings if it appears in the scorecard as such. Ananth: ]]

  • Devendra on December 7, 2009, 12:48 GMT

    Very Interesting way of measuring "Destruction". Maybe ..Instead of considering the team score .. the score made by the team when the Batsmen was on field may hold more relevance. [[ Devendra The correct information on who was out at what score is available only from around match # 1000 onwards and with complete accuracy from match # 1500 onwards. Otherwise it is a good suggestion. Ananth: ]]

  • Pranesh on December 7, 2009, 13:21 GMT

    "It should be noted that this factor, being a pure batting one, does not take into account team strengths, bowling quality, pitch type, innings status, result et al." Sorry, but it should be result etc. et al is used when you cite one example. [[ Praneshgreat I have always used et al as "and others". Here is the Dictionary definition. "" et al means roughly "and others". It is written at the end of a list of names to indicate that others are related to the same subject. "" Ananth: ]]

  • Arjun on December 7, 2009, 13:29 GMT

    I think IPF should be done separately for Ist and 2nd innings since while batting in 2nd innings not always full resources(overs) are utilised. [[ Arjun I have given the innings info. So it would not be difficult for anyone to view these separately. The problem is that I already have 5 tables and the article would have been quite lengthy. The other point is that the % TS is based on Runs which is a complete measure even for the second innings. Ananth: ]]

  • Sesha on December 7, 2009, 15:09 GMT

    Hi Ananth & Alex,

    One of the most satisfying articles to read because I was able to connect to the article and ideas so well. Special mention to Ananth for using appropriate words to get the message across in a slightly tricky article.

    In my first comment, I would like to mention a few of my observations... 1. Except for McCullum & Sehwag's innings All other entries in top 20 are their respective highest followed by 2/3 highest.

    2. All batsmen in the top 20 have a SR of >80 except Kirsten and Gower. Gower's SR of 75 is exceptional for the era he belonged to

    3. Notable Exceptions in the first table from the modern era who are famous for their destructive abilities are Afridi, Gayle, Treschk, Graeme Smith, KP & Hayden who aptly feature in 2nd table

    4. With due respect to Z.Abbas, he has played in too little innigs to be heading table 2. Mr.Ananth you can consider some entry barrier.

    5. Another proof y Sach & Viv are the best ODI batsmen ever..

    Contd...

  • Deepak Shah on December 7, 2009, 15:24 GMT

    This is a great analysis. One interesting observation that I had was the absence of Michael Bevan from all of the tables. How do you explain that?

    Maybe, if you brought in the "not-out" factor, it would change the lists quite a bit, and make it more fair for those batsmen who batted later in the innings. [[ Deepak Bevan was a great finisher. He was the 80 in 100 batsman, not the 120 in 80 batsman although he has done that also. Also being not out does not get into the equation. Pl remember that we are analyzing individual innings. Ananth: ]]

  • Sesha on December 7, 2009, 15:35 GMT

    Contd...Observations

    6. Totally agree with Mr.A on table 4. Except for Z.Abbas and KP, I guess Bevan, Gilly could be replaced just to see the top 10 batsmen..:-)

    7. If my eye-balling is correct Dhoni 183 & Gibbs, Sach's 175 have been scored while chasing...Mr.A, could you consider adding any weightage to chasing...which anyday makes a good score even sexier...

    Some thoughts reading the blog:

    1. Mr.A, Could you come out with what is the least strike rate a batsman shall have based on avg winning scores. I remember you doing a similar work for T20s.

    2. How has SR & Avg of top batsman developed over the years...Sachin & Sanath with SR of >85 even after 400 matches is one of the unbelievable numbers...more so Sach's because of his Avg too.. [[ Sesha First Zaheer. Poor guy played for 12 years and appeared in 62 matches while Pakistan played 88 matches. Contrast this with Pollock who played for the same 12 years and played in 303 matches while SA played 335 matches. Zaheer did wonderfully well in the matches which were available to be played. Bevan has not played many such explosive innings. As I have already explained to Arjun, download the data file, export to Excel and you can do what you want. Separate the innings as first or chasing and so on. Dhoni's is the best chasing innings ever, by a mile, no, 5 miles. If you have a downloading problem I can upload to Thirdslip.com as a text file for easier downloading. In one of my earlier articles titled "ODI matches - an analysis through 4 decades", I have done an evolution comparison of S/R and Avge over the years. Although now I can do a far superior article. Ananth: ]]

  • love goel on December 7, 2009, 15:37 GMT

    The formula is really simple and elegant. It leads itself to great result, especially Table2.

    However I had a question on the formula. If %TS is included, why not include Batsmen Strike Rate/Team strike rate also. This will balance out the effect of high scoring matches, just like %TS score

    And Ananth,For table 4 ,you have written "The more I see the table the more I feel that this is the single table which encompasses the ODI greats in full"

    To me Table 2 seems to be more important if you wish to look at great ODI batsmen. Table 4 to me, favours consistent players over inconsistent ones. Any particalur reason you chose table 4? [[ Goel No particular reason other than the fact that at one shot we have rewarded high level of success and low level of failure. But I can see the table 2 being the more relevant one. Ananth: ]]

  • love goel on December 7, 2009, 15:54 GMT

    "An innings is counted only if he gets to start his innings. In other words, he could be not out with 0 in 0. That is an innings if it appears in the scorecard as such"

    Why count a not out inning? It may be valid when you are doing an inning based analysis. But on a career averages based analysis(Table 2),shouldn't the not out inning be ignored? After all,cricket batting averages also exclude not innings.

    Also in case of an inning of 0/0, if you take IPF to be zero, it becomes an IPF<2 innings for the batsmen, without any input from the batsmen. Why not exclude this innings from the list of bad performances(Table 4)? Although any not out innings with good IPF has to be still measure in; we cant penalize batsmen for staying not out. [[ Goel That is the convention and I do not want to change it. If I do that there will be inconsistency between my database and other recognized databases. If a batsman walks in, he has played an innings, that is convention. How can I exclude Not outs. It could be 189* or 0*. But your last point is valid. A still-born innings need not be considered a failure. I will make that correction and post the changed table. Ananth: ]]