Anantha Narayanan

ODI overs analysis using ball-by-ball data: part 3

Detailed analyses of the scoring trends in ODIs, beginning with the 1999 World Cup. Plus, a Bradman surprise

South Africa are the possibly the best ODI team currently with no apparent chinks in their armour  AFP

Ball-by-ball data for ODI matches is available for 1784 matches, just over 50% of the matches played till date. The data is available for the 43 matches of the 1999 World Cup and then for 1741 matches from match no. 1719 onwards. For 70 matches in this lot, either there is no ball-by-ball data available or the available data is quite incorrect. This massive data-gathering exercise has been completed through Milind's sterling efforts, especially for the first 1700 matches or so.

Two articles, one part-analytical, part-anecdotal and the second one completely anecdotal, based on over as an entity were published recently. In this third part, I will look at this data from a purely analytical point of view.

The classifications are by Teams, Periods and OverGroups. There are eight teams (apologies to Bangladeshi and Irish followers et al), four time periods (with World Cups as group separators) and seven over groups. As anyone with an analytical bent will know this is a three-dimensional database and presents huge problems presenting data in the conventional two-dimensional tables. Graphs will be laborious to present and difficult to comprehend, which may also end up confusing the readers.

Hence I have adopted a via-media approach. For each of these three analytical areas, I will present a summary table and then present combinations of two in two-dimensional tables. Teams will be combined with remaining classifications. I will also present the Period and OverGroup combination.

The bowling analyses will contain both bowling strike rate (BpW) and bowling average (Avge) measures. This will let the readers look at different aspects of the analyses.

1. OverGroup summary
Over Group Inns1-RpO Inns2-RpO Inns1-BpW Inns1-Avge Inns2-BpW Inns2-Avge
1 - 104.354.7444.732.3941.332.61
11 - 154.494.7045.233.8744.034.41
16 - 254.314.5249.835.7447.135.45
26 - 354.634.8550.939.2840.933.11
36 - 405.375.2836.332.4429.726.17
41 - 456.225.8725.726.6723.222.73
46 - 508.387.0213.318.5914.817.34

The OverGroups have been determined based on information needs as well as current ODI laws. Thus overs 1-10 is the first PowerPlay (PP) and overs 11-15 is almost always the second PP. Overs 16-35 are split into two groups of middle overs. Overs 36-40 is separated since most of the times, the batting PP is taken during these overs. Then the last two sets of overs: one in which teams press the accelerator pedal and the concluding one in which the pedal stays at the floor.

During the first PP, the second innings scoring rate is about 10% higher. This is a reflection of the first innings situation: the uncertainty of the target. The bowler PP overs continue on similar lines but at a lower differential. In fact this trend of 4.xx scoring rates in the first innings and around 5-10% increased scoring rates in the second innings continue up to the fourth OverGroup: Overs 26-35. Then the Batting PP appears, hastening scoring rates for both sides. Teams batting first move couple of gears, averaging 5.36. Teams batting second move up by a lower value: by less than 10%. The first of the 40s overs has scoring rates either side of 6.0. In fact the last ten overs show a much higher first innings scoring rate: again a reflection of the first-innings scores being higher by winning margins.

The BpW starts at a middle level and increases to a maximum in the 26-35 OverGroup. There is a huge difference in this OverGroup between the first and second innings. The wickets continue to fall more rapidly in the second innings until the last group. There is a sudden drop in the second innings. The other OverGroups are comparable. But overs 26-35 are intriguing. A wicket falls every 51 balls in the first innings and every 41 balls in the second innings. Why? Is it because the second batting teams bat too cautiously in the first 25 overs and want to make amends? Is it because spinners bowl during these overs and chances are taken?

2. Period summary
Period Inns1-RpO Inns2-RpO Inns1-BpW Inns1-Avge Inns2-BpW Inns2-Avge
WC1999-20034.864.7037.130.0337.929.70
WC2003-20074.944.8336.530.0337.530.25
WC2007-20115.104.9535.229.9436.630.18
WC2011-20145.185.0235.930.9536.230.23

Now, for the period analysis. From now onwards, I will define the periods based on World Cup events. This will be better than arbitrary 5-10 year periods. Also, World Cups launch new strategies for teams: Examples are the way Martin Crowe used Mark Greatbatch in the 1992 World Cup, the all-out attack by the openers in the 1996 World Cup, the no-score-is-safe chases in the last two World Cups.

Although this analysis could have been done for the first six World Cups also (three periods), I have been consistent in restricting myself to the ball-by-ball data period only. I will summarise by saying that the scoring rates during the period from 1971 to 1999 was around 4.5. The overall scoring rates have moved up in the last-four periods steadily. What does this move signify? The average 50-over first innings score between 1999-2003 was 243 and the post-2011 period, 259. Over thousands of matches this is quite significant. The second-innings scoring rates are reflective of the lower target scores.

Surprisingly there is very little variation in the BpW values. The average BpW value oscillates around the 36-mark. The eight values range between 35.2 and 37.9. This is remarkable. Through thousands of matches the overall value to be either side 36.

3. Team summary
Team First Inns Inns1-RpO Second Inns Inns2-RpO Inns1-BpW Inns1-Avge Inns2-BpW Inns2-Avge
Australia217 5.38149 5.2833.226.1932.126.06
India186 5.47219 5.2337.632.3734.730.06
Sri Lanka205 5.04163 5.0435.529.2334.527.32
Pakistan173 5.11163 4.8936.229.8535.528.70
South Africa143 5.41152 5.1035.328.5032.825.80
West Indies135 4.97159 5.0036.530.3539.031.75
England150 5.14155 5.0234.928.9638.533.46
New Zealand133 5.09150 5.0135.629.1637.630.49

I have selected the top eight teams for this analysis. There are too many teams to present a complete analysis. The sequence of presentation is based on World Cup performances. Australia, with three World Cup wins is at the top, followed by India, with one World Cup win and one runner-up position. They are followed by Sri Lanka, with two runners-up positions and Pakistan, with one runner-up position.

The number of matches between batting first and second for Australia is dramatically lopsided. Australia batted first 60% of the times. They are quite confident of defending totals, with their bowling and fielding strengths. It could also be because the other big team, India, prefer to chase. Australia's average during the first-innings score is around 270. They have the best second-innings scoring rate. To boot, their wicket-taking rates are spectacular. The best amongst all teams. It is no wonder that they have been the most successful team, by the proverbial mile.

India averaged around 273 runs in the first innings. But their bowling in the first innings has been the worst amongst all teams featured here. They bowl better in the second innings. This is a paradox. Based on these numbers one should think that India would have been better off batting first and defending. Maybe MS Dhoni is too obsessed with his preconceived ideas. They bat first about 45% of the time.

The absence of heavy hitters is reflected in Sri Lanka's numbers. But their bowling has been excellent. Pakistan are also like Sri Lanka. Fair batting and good bowling. South Africa's numbers are excellent. Batting nearly as good as India and much better bowling. They should be sitting with a World Cup trophy or two. Let us see what happens next year. They should be joint favourites with Australia once they learn how to dismiss ten and Jack.

England's second-innings bowling lets them down. West Indies are average in batting and bowling. New Zealand's batting and bowling are right in the middle.

4. Period - OverGroup analysis
Period Over Group Inns1-RpO Inns2-RpO Inns1-BpW Inns1-Avge Inns2-BpW Inns2-Avge
WC1999-2003 1 - 104.214.6047.032.9542.032.18
WC1999-200311 - 154.724.8041.232.4844.135.26
WC1999-200316 - 254.204.2555.638.9450.835.98
WC1999-200326 - 354.584.6852.339.9447.336.88
WC1999-200336 - 404.994.8838.131.7129.023.60
WC1999-200341 - 455.735.2024.223.0721.919.03
WC1999-200346 - 508.076.0513.518.1816.917.03
WC2003-2007 1 - 104.264.6042.630.2040.330.91
WC2003-200711 - 154.514.7844.433.3642.934.15
WC2003-200716 - 254.254.5248.934.6349.837.53
WC2003-200726 - 354.414.6952.538.5740.731.78
WC2003-200736 - 405.155.0739.133.5932.727.66
WC2003-200741 - 456.165.8527.328.0125.925.24
WC2003-200746 - 508.826.3913.219.4816.317.41
WC2007-2011 1 - 104.534.9644.833.8041.234.03
WC2007-201111 - 154.464.7344.833.3142.733.66
WC2007-201116 - 254.304.4946.433.2744.933.66
WC2007-201126 - 354.734.8849.438.9240.032.52
WC2007-201136 - 405.365.2035.631.7829.725.74
WC2007-201141 - 456.355.8926.728.2224.023.51
WC2007-201146 - 509.086.5911.717.7015.817.30
WC2011-2014 1 - 104.314.7246.133.1542.233.21
WC2011-201411 - 154.384.4949.836.3747.135.26
WC2011-201416 - 254.464.6952.438.9945.035.13
WC2011-201426 - 354.835.0750.040.2239.733.53
WC2011-201436 - 405.895.7233.032.3727.926.63
WC2011-201441 - 456.405.9724.025.6021.621.47
WC2011-201446 - 509.166.8112.418.8915.417.47

This is first of the multi-classification pseudo three-dimensional analyses. Here I take the Period as the base and represent the Over groups at the next levels. This is linearly presented with each Period shown separately.

There are many numbers and I will come out with a few insights. I will present the data and let the readers come out with their own derivations.

Overs 36-40 deserve a careful look. The scoring rate has moved from 4.99 through 5.15, 5.36 to the current 5.89. The 10% jump in the last period is the effect of PP3. In the second innings the number moves up from 5.20 to 5.72. What about the bowling average? It hovers around the 32-mark. So there is no great change because of the PP, contrary to popular belief.

The last ten overs have seen a huge jump from around 6.8 during the first period, through 7.4, 7.6 to 7.7 currently. That is a significant change and reflects the current high average scores. The first ten overs has seen a drop in scoring rate during the current period.

The trend of BpW movements continues across the periods. However, there is a 10% drop in the first over values between the first and second periods. This is a significant increase in the number of wickets which were captured. Right through the years, the overs 26-35 have seen least number of wickets falling. However this trend seems to change during times with more wickets falling. The BpW figure has fallen from around 50 to 40.

>
5. Team - Period analysis
Team Period Inns1-RpO Inns2-RpO Inns1-BpW Inns1-Avge Inns2-BpW Inns2-Avge
AustraliaWC1999-20035.065.2334.425.6432.023.31
AustraliaWC2003-20075.485.2432.925.2331.425.64
AustraliaWC2007-20115.405.2931.725.9530.724.92
AustraliaWC2011-20145.345.4334.529.1135.429.73
IndiaWC1999-20035.204.9238.431.3337.330.69
IndiaWC2003-20075.255.1435.930.4335.529.50
IndiaWC2007-20115.705.3438.033.4434.731.59
IndiaWC2011-20145.705.4038.433.6832.028.72
Sri LankaWC1999-20034.864.7735.729.5236.727.47
Sri LankaWC2003-20074.964.8234.225.9437.629.55
Sri LankaWC2007-20115.135.1434.529.0431.824.96
Sri LankaWC2011-20145.135.2737.732.4032.827.41
PakistanWC1999-20035.154.5634.226.2033.827.17
PakistanWC2003-20075.015.0336.930.8136.429.51
PakistanWC2007-20115.424.9936.732.0335.929.85
PakistanWC2011-20144.894.8236.228.9735.527.77
South AfricaWC1999-20035.354.7436.627.3232.423.97
South AfricaWC2003-20075.115.1235.528.3736.428.02
South AfricaWC2007-20115.665.3535.129.5433.027.28
South AfricaWC2011-20145.535.0833.828.5029.723.55
West IndiesWC1999-20034.895.2233.225.6840.130.90
West IndiesWC2003-20074.914.9338.131.3539.031.95
West IndiesWC2007-20115.084.9737.031.0339.833.56
West IndiesWC2011-20144.995.0635.130.3737.830.71
EnglandWC1999-20035.084.5938.331.5139.033.24
EnglandWC2003-20074.925.0434.126.4741.935.21
EnglandWC2007-20115.274.9934.429.0036.632.32
EnglandWC2011-20145.305.2635.030.7436.632.99
New ZealandWC1999-20034.554.3935.827.6437.628.91
New ZealandWC2003-20075.005.0538.131.6236.528.81
New ZealandWC2007-20115.145.2934.027.7240.632.96
New ZealandWC2011-20145.645.1732.828.3735.631.34

This is the first of two team-based tables. The first is Team-Period and the second is Team-OverGroup.

What is obvious?

- Australia's recent decline in bowling standards.
- India's growing batting strengths, offset by poor first innings bowling.
- Sri Lanka's predicament similar to India: Good batting and poor first innings bowling.
- Pakistan's drop in batting but maintenance of good bowling.
- South Africa's outstanding numbers. They are the best ODI team in the world now. No weakness in sight.
- England and New Zealand in similar situations: good batting and poor defence of targets.

This is the Team-OverGroup analysis. I have split this into two tables: Four teams each because there are seven OverGroups.

6a. Team - OverGroup analysis (Top 4 teams)
Team Over Group Inns1-RpO Inns2-RpO Inns1-BpW Inns1-Avge Inns2-BpW Inns2-Avge
Australia 1 - 104.805.4238.625.6834.924.98
Australia11 - 154.905.1541.330.4236.627.91
Australia16 - 254.584.8446.732.2446.635.12
Australia26 - 354.855.0642.331.3435.829.02
Australia36 - 405.515.3035.530.1826.524.69
Australia41 - 456.706.0222.422.6519.719.58
Australia46 - 508.707.1412.516.4213.916.57
India 1 - 104.855.0747.936.2541.635.86
India11 - 155.044.8943.333.4238.931.14
India16 - 254.734.7554.840.4942.533.03
India26 - 355.055.3051.340.8240.234.05
India36 - 405.725.7134.932.8228.626.18
India41 - 456.556.0830.032.0523.123.23
India46 - 508.877.2413.218.4414.818.09
Sri Lanka 1 - 104.565.2139.528.4937.528.87
Sri Lanka11 - 154.564.7948.835.2941.332.54
Sri Lanka16 - 254.424.5543.331.5943.231.55
Sri Lanka26 - 354.834.7249.138.3836.228.22
Sri Lanka36 - 405.275.3634.629.8829.124.20
Sri Lanka41 - 455.945.8629.829.9622.320.89
Sri Lanka46 - 507.717.3414.018.5715.416.74
Pakistan 1 - 104.084.5041.130.4345.035.46
Pakistan11 - 154.274.3241.832.1947.237.27
Pakistan16 - 254.364.3155.739.8545.133.52
Pakistan26 - 354.935.0751.938.6239.630.58
Pakistan36 - 405.485.6841.735.2125.320.84
Pakistan41 - 456.626.2525.524.5721.220.42
Pakistan46 - 508.786.8613.218.2213.916.72

- Australia's outstanding bowling efforts right through the innings.
- India move into higher gears early in the innings.
- India's opening bowling attack lacks sting. A wicket every 36 balls only.
- Sri Lanka's finish is quite ordinary.

6b. Team - OverGroup analysis (Next 4 teams)
Team Over Group Inns1-RpO Inns2-RpO Inns1-BpW Inns1-Avge Inns2-BpW Inns2-Avge
South Africa 1 - 104.504.8647.632.6137.026.12
South Africa11 - 154.935.2953.940.1742.530.91
South Africa16 - 254.534.6249.935.4441.930.47
South Africa26 - 355.094.9449.037.6635.428.89
South Africa36 - 405.755.4428.725.2327.525.08
South Africa41 - 456.515.7624.823.8018.416.96
South Africa46 - 509.367.6712.516.2315.917.11
West Indies 1 - 104.164.5844.431.8839.530.98
West Indies11 - 154.194.8249.237.1844.034.26
West Indies16 - 254.094.6846.233.4550.337.77
West Indies26 - 354.664.9060.647.1645.436.74
West Indies36 - 405.855.7338.033.5034.028.49
West Indies41 - 456.526.2126.627.4431.531.17
West Indies46 - 508.527.1112.516.8914.216.27
England 1 - 104.614.7743.930.7645.438.96
England11 - 154.855.2038.227.6656.647.44
England16 - 254.504.6748.235.0344.134.53
England26 - 354.554.8252.240.6543.236.92
England36 - 405.595.2036.632.3132.730.35
England41 - 456.225.9124.526.6323.222.81
England46 - 508.317.2712.716.9116.719.83
New Zealand 1 - 104.264.8742.130.2242.733.44
New Zealand11 - 154.704.5047.134.8142.932.48
New Zealand16 - 254.254.6044.231.1553.138.57
New Zealand26 - 354.685.0446.635.0141.032.67
New Zealand36 - 405.595.3335.530.5631.528.05
New Zealand41 - 456.256.0224.424.6422.823.00
New Zealand46 - 509.047.4314.321.2614.016.75

- South Africa's finish is spectacular. Well above 9 RpO.
- New Zealand are also quite good.
- South Africa's defence of targets is extraordinary.
- England's defence is awful. Look at the first 15 overs.

For the time being this concludes my ODI overs-based analyses. There is a lot more that can be done. I will keep these for a later day.

A bonus for the readers. I have a rare two-volume collection of Bradman memorabilia. It is a fantastic set of books and contains scorecards, telegrams, photographs, letters et al. Along with each article I will scan one such rare document and attach for user viewing. I had presented the chart for Bradman's 334 in the previous article. To view the next scan, please click HERE. This is a 1931 telegram offering Bradman £500 for a five-month playing assignment, by Accrington, a Lancashire club. This was after Bradman's successful tour during 1930. Note the £2 per week charge for accommodation. Did Bradman accept the offer? There is a story behind this and I will provide the answer if some reader scans and elicits a query.

Alternately here are a few links, courtesy Milind, on this topic.

Wiki

Background

Offer

Decision

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