Smart Stats: the HD version of conventional cricket stats

Data analytics hasn't perhaps taken a strong foothold in cricket as they have in some other sports, yet. However, cricket lends itself to data analytics like few other sports do. The sheer volume of variables - some more quantifiable than others - that influence the outcome off each ball pose as many challenges to data analysts as they offer opportunities.

In its briefest form, cricket has 240 units of action events that generate quantifiable variables. Hundreds of games are played every year that generate thousands of data points that can be analysed and dissected to help improve the way the game is followed and experienced by fans.

Smart Stats, launched as a standalone tool in 2018, is a suite of metrics that helps fans judge performances in limited-overs cricket, especially T20s, in a far more nuanced manner than conventional metrics do. Smart Stats brings fans and experts closer to the action using numbers - they are the 'high definition' version of conventional stats.

Last year, ESPNcricinfo had introduced metrics such as Smart Runs, Smart Strike Rate, Smart Economy and Batting and Bowling Indices. Now, we bring a new and improved version of the same. Sophisticated algorithms make sense of various input parameters to churn out numbers that quantify contexts and add meaning to performances like never before. Additional metrics such as Smart Wickets, Player Quality Index and Pressure Indices have been added to cover all possible facets of the game.

We also introduce Forecaster - a tool that predicts the scores and the outcome of games. At any stage in a match, fans can now know the chances of their team winning or the total their team is likely to put up.

So what do Smart Stats tell about the IPL 2018 season? Who were really the best performers? How do the leaderboards look as seen through the lens of Smart Stats? We present a few of the finding in this article.

Ever felt that a wicket is actually worth more than just another wicket? And not all wickets are worth the same? Smart Wickets puts a number to that notion. It values each wicket by looking at the quality of the batsman dismissed, the stage of the game, and impact of the wicket on the outcome of the match.

To illustrate how Smart Wickets work, consider this: both Andrew Tye - the Purple Cap winner - and Rashid Khan got the wicket of AB de Villiers in last year's IPL. But Tye gets only 0.95 Smart Wicket value for his wicket, while Rashid gets 1.64. Why?

Because by the time Tye dismissed de Villiers in the 19th over of the chase in Bengaluru, the batsman had already done the damage with a 39-ball 57. Royal Challengers Bangalore needed just ten runs off 11 balls at that stage. On the other hand, Rashid took de Villiers' wicket early in his innings when RCB required just 67 runs from 56 balls with seven wickets in hand. As it turned out, de Villiers' wicket was crucial to the outcome of the match, with the chasing team falling short by just five runs.

Another example is from the RCB v Kings XI Punjab match in the 2017 season in Bengaluru that the visitors won defending a middling total. Both Sandeep Sharma and Axar Patel took three wickets each for Punjab. However, while Axar took the wickets of Shane Watson, Pawan Negi and Samuel Badree, Sandeep was the bowler to derail RCB's chase with wickets of Chris Gayle, Virat Kohli and de Villiers inside the Powerplay. Sandeep's three wickets were worth 4.86 on Smart Wickets. Axar's three were worth 2.85.

Rashid's 21 wickets get him a Smart Wicket value of 26, making him the Purple Cap winner according to Smart Stats. Umesh Yadav pushes the official Purple Cap winner of IPL 2018 to third place place with 23.85. Tye, Siddarth Kaul and Hardik Pandya complete the top five. Rashid's Smart Wicket value is 5 points more than his actual wickets, which is the highest additional points any bowler earned for his wickets last season. Smart Wickets reinforces the popular opinion that Rashid is by far the best bowler in the T20 format.

Smart Runs value the 'quality' of runs that were scored by batsmen. For example, runs scored by batsmen under high pressure that help their team win matches are considered more valuable. Kane Williamson was the official Orange Cap winner in IPL 2018, having scored the most runs - 735 in 17 innings at an average of 52.50 and a strike rate of 142.44. Rishabh Pant of Delhi Daredevils (now Delhi Capitals) ended at the second place with 684 runs from 14 innings at a strike rate of 173.60. Pant, however, tops our charts with 850 Smart Runs. He gets this because of, among other innings, an unbeaten 63-ball 128 against Sunrisers Hyderabad, who boasted the best bowling attack last season. KL Rahul, Ambati Rayudu and Shane Watson complete the top five.

An important input that goes into deciding the 'quality' of runs scored by batsmen is the pressure created by the match situation. At each ball that a batsman faces, the Pressure Index algorithm quantifies the notion of pressure on him (likewise on the batting team) between values of 0 and 10. Runs scored off each ball are weighted up based on the Pressure Index value on that ball. As a result, runs scored in high-pressure situations will be worth more than otherwise.

In the qualifier between Chennai Super Kings and SRH last year, the Pressure Index graph below shows the pressure CSK faced during the chase. Faf du Plessis - who won the match for them single-handedly - batted through the innings. The pressure on CSK, and hence on du Plessis, is seen rising slowly as wickets fall, before peaking in the 17th over when they were seven down and managed just four runs in the over with the ask at nearly 11.

Any runs scored by du Plessis towards the end of the chase will be valued more than the runs he scored earlier in the innings. To illustrate this, let's take the two sixes du Plessis hit off Carlos Brathwaite in the match. The first one came in the ninth over when the pressure value was under six. The other one came in the 18th over when the pressure value was well over nine. Du Plessis gets a Smart Runs value of 8.5 for the first six and a much higher value of 11.7 for the second six.

The Smart Stats algorithms also put a number to the impact of each run scored (or conceded) and each wicket taken, thereby making it possible to quantify and rank each player's contribution to a match. This not only results in objective identification of the Player of the Match but an aggregation of these contributions over a season also tells the fans who the Player of the Tournament is.

Sunil Narine, with his performances with both the bat and the ball, tops the leaderboard in terms of the overall impact he had in the IPL 2018 season. A 22% lead over the second-placed Pant shows that Narine was well ahead of the field in terms of the impact he had on the season. Rahul, Williamson and Watson complete the top five.

Forecaster employs a powerful algorithm that uses historical probabilities as bases to predict teams reaching a score or overhauling a target. Historical probabilities, however, play only an indicative role. The algorithm rationalises the numbers by considering match situations, batting and bowling strength of sides left in the game, and conditions to make the predictions more representative of the situation in the match when the prediction is made.

For example, in the Rajasthan Royals v SRH match in Jaipur in 2018, it seemed that Royals were comfortably placed when, chasing 152, they required 50 from 30 balls with six wickets in hand. Historically in the IPL, teams had won 60% of the matches from this situation. However, Forecaster gave Royals only a 41% chance of winning considering Rashid had two overs left and the Royals didn't have a strong line-up of batsmen to follow. As predicted, in the last five overs, the Royals managed only 38 runs and lost the match by 11 runs.