It was interesting to read a preview ahead of the KKR v Delhi match the other night highlighting some key “contests” for the game ahead. The contests were as follows: Gautam Gambhir v Quinton de Kock. Robin Uthappa v Shreyas Iyer. Yusuf Pathan v Karun Nair. Andre Russell v Carlos Brathwaite. Shakib Al Hasan v Pawan Negi.
Were these the real “contests” in the game? Gambhir v de Kock? Uthappa v Iyer? How much was each player’s performance affected by the other person in the “contest”? Probably none at all. Of course we understand the reviewers intention in suggesting that these were contests, however it is a shame that the real contests in that match and the other matches played to date in IPL2016 have been played out almost invisibly to the spectator, but with real consequence to the outcome of the game.
So what are these contests I am referring to? I am referring to the match-ups between the various batsmen and bowlers that played in the game. How well do team management select their bowling line up taking into account the likely batting line up of the opposition and their strengths and weaknesses? During the game, how well does the captain use his bowlers to match their strengths against the weaknesses of the batsmen at the crease? We would like to think that in this day and age there would be quite a bit of science behind these decisions – and this may well be the case – but there is some evidence to suggest otherwise.
Before sharing some results from the first four games of IPL2016, the background behind match up analysis needs to be introduced. The first component is a players Contribution Score.
Contribution Score Definition
Traditional metrics in T20 cricket such as average and strike rate do not alone capture one of the more critical aspects of player performance, which is the state of the match at which a player makes his or her contribution. Take the following example.
Suppose a batsman scores 50 runs off 50 balls. While this is a reasonable contribution in any form of cricket, it would generally be considered slightly below average in T20 cricket given that the average strike rate is closer to 130 runs per 100 balls faced. Now take two different game situations in which this score might be made:
- An opening batsman in a team batting first scores 50 off 50 balls and is the first wicket to fall, in the 18th over with the score at 115
- An opening batsman finds his team at 20-6 chasing 80 for victory. He scores 50 off 50 balls and although he is dismissed in the last over, his team hangs on for a 2 wicket win
Which inning is more important for the team? Clearly the second innings was more important and should be rated higher than the first. Most available metrics in cricket will not recognize the difference in value of these two innings. The Contribution Score (CS) however identifies every match state in every game, and assesses the players’ performance accordingly. As a result, over the course of an inning, a match, a series, a year or a career, we can use Contribution Scores to assess how a well a player has performed given the situations he or she has been in.
CSs are measured in terms of runs. If a batsman’s CS for an innings is 0, it means that the player performed in line with average given the state of the match during which the inning was played. A CS of say +10 for an innings means the player contributed 10 more runs than expected given the state of the match. For example a batsman who averages a CS of +10 effectively contributes +10 runs above average to the team performance per innings. If the average score for a team batting first in T20 cricket is 151, then this players expected contribution is to increase the team’s average total by +10 runs to 161.
The same approach is applied for bowlers. A score of +1 for a batsman for a given delivery automatically means a score of -1 is given to the bowler of that delivery. So a bowler who averages a CS of -5 runs a game effectively gives away 5 runs on average to the opposition per game. If the average score for a team batting first in T20 cricket is 151, then this players expected contribution is the increase the opposition team’s average total to 156.
Another benefit of CS being measured on the same scale is that a players batting CS can be added to his/her bowling CS to arrive at his/her overall contribution to a match. This is useful for assessing who was the most influential player in a match. Any contribution score greater than zero should be interpreted as above average performance. The higher the CS, the more above average.
Contribution Scores are highly correlated with the team’s success, more so that any combination of strike rate, average etc because of the additional match state information encapsulated by the Contribution Score. The ease of having one number for a player that is both relevant to a team’s success and captures all aspects of a player’s performance allows for some interesting analysis.
Bowling Match Up Analysis
So how can Contribution Scores be used for match up analysis?
Bowling Match Up analysis looks at each individual bowler’s historical performance against each batsman type (i.e. right handed or left handed batsman) and each individual batsman’s historical performance against each bowler type (i.e fast, fast/medium, medium, finger spin or wrist spin delivered by either a right arm or a left arm bowler). Match Up analysis uses Contribution Scores (CS) as the underlying metric. Take the following example.
Martin Guptill’s CS per ball as a batsman against left arm orthodox bowling is -0.2. I.e. Guptill historically performs below average against left arm orthodox bowling in T20s. RA Jadeja is a left arm orthodox bowler whose CS per ball as a bowler against RHB (Guptill is a RHB) is 0.1. I.e. Jadeja historically performs above average against RHB. Intuitively it would be a positive match up for the bowling team to have Jadeja bowl to Guptill. How positive would the match up be? It can be quantified by taking Guptill batting CS per ball against left arm orthodox bowling (-0.2) and subtracting from it Jadeja’s CS per ball bowling to right hand batsmen (0.1) which equals -0.3. How do we interpret this? On average, every ball that Jadeja bowls to Guptill will reduce the teams final total, compared to the total that should be expected given the state of the game, by 0.3 runs. From the batsman’s point of view the CS is worth -0.3 runs per ball, but from the bowlers point of view, the match up is worth +0.3 runs per ball. The more balls bowled with this kind of match up, the better it is for the bowling team!
A positive CS based match up from the bowling team’s perspective is defined as a match up where the difference between the batsman’s historical CS per ball against that type of bowler and the bowler’s historical CS per ball against that type of batsman is more than +0.1. A negative CS based match up from the bowling team’s perspective is defined as a match up where the difference between the batsman and bowler’s historical CS per ball is less than -0.1, and a neutral match up if the difference between the batsman’s and bowler’s CS is between -0.1 and +0.1.
Of course these results are what is expected, on average. There will be times that a match up seems positive from a bowler’s point of view, but the actual the result is negative. Maybe the bowler bowls poorly, maybe the batsman has worked on his prior weakness. This uncertainty is part of the beauty of sport. But you might be surprised how clearly over the long run the expected positive match up actually does deliver a positive result for the team.
IPL2016
So, four matches into IPL2016 and what do we see? Refer to the table below:
The observations are as follows:
1. 30% of balls bowled in the 2016IPL to date have been in match ups that were expected to be positive from the bowlers point of view. The actual average contribution score per ball in these match ups was +0.5 runs per ball worth of benefit to the bowler!
2. 47% of balls bowled in the 2016IPL to date have been in match ups that were expected to be negative from the bowlers point of view. The actual average contribution score per ball in these match ups was -0.3 runs per ball worth of benefit to the bowler, i.e. +0.3 runs per ball of benefit to the batsman!
i.e. For each ball the bowling team is able to either better manage team selection and/or game day bowling strategy by substituting one ball of negative match up to positive match up is worth 0.8 runs (almost 1 run) of benefit to the team! Imagine if a team could reduce its 47% negative match up to 30%, with corresponding increase in positive match ups. This equates to a 16 run reduction in the opposition batting team’s score, which is huge!
3. The team with the higher percentage of positive bowling match ups has won three of the four games so far this IPL. Match 3 was the only match that bucked this trend – and in that match both teams had a very poor positive match up rate.
Again these results are on average and it requires bowlers to bowl up to their own standard and opposition batsmen to play according to theirs. Each will have good and bad days, but on average – and it is a great start to picking a team or developing a strategy – the match ups work!
Look out for a follow up piece on this where we share some of the key match ups at player level.