A Discussion About Match-Ups

Email: Sportsanalyticsadvantage@gmail.com

If this article has given you insight into the data that Sports Analytics Advantage can offer cricket teams around the world in formulating team strategies, selection, draft or auction plans, or any other work, please feel free to get in touch at sportsanalyticsadvantage@gmail.com.

11th November, 2020.

For yesterday's IPL Final, the Mumbai Indians sprung a surprise by leaving out leg-spinner Rahul Chahar and, in his place, including off-spinner Jayant Yadav.  This decision wasn't necessarily performance related - Chahar had taken 15 wickets during the tournament - while Yadav had bowled just 18 balls previously in his one appearance of the tournament, which also came against Delhi Capitals on the 31st October during the group stages.  

To call Yadav a journeyman would be a little unfair - he's played for India in both Test and ODI cricket, and made a century batting in a Test match - but he's had limited impact at IPL level, making a grand total of 14 career appearances across six seasons since his debut in 2015, at the age of 30.  Yadav's overall T20 record is certainly pretty unspectacular from a wicket-taking perspective, although he has offered solid economy for his teams at various levels.  However, it would probably be reasonable to assert that the majority of regular IPL watchers would consider Chahar to be the better spinner, without any further reasoning.

It was not a coincidence that Yadav's previous appearance for Mumbai this season came against Delhi.  Delhi have more left-hand batsmen than most teams (four in their top seven yesterday) and Yadav, as an off-spinner, should benefit from this with some positive match-ups - in fact, he's taken all of his career IPL wickets against left-handers.  Conversely, Chahar's IPL economy rate against left-handers is over a run worse than against right-handers, and he's taken over 80% of his career wickets against right-handers despite bowling only 65% of his overs against right-handers.  So, it is clear that Yadav is relatively better against left-handers, while Chahar is relatively better against right-handers.

Mumbai had the additional benefit of having the spin option of all-rounder Krunal Pandya, who is a left-arm orthodox spinner.  Krunal's IPL record against right-handers is very strong, while he has struggled against left-handers, running at an economy rate of around 2.5 runs per over worse against left-handers.  Given that Krunal was unlikely to be dropped from the Mumbai side due to the extra benefit of his batting, it was logical that Yadav should be included at the expense of Chahar, to give Mumbai strong options against both left-hand and right-hand batsmen.  It should also be noted that all three of these bowlers data here broadly goes along with general player population tendencies, with right-arm off-spinners generally performing better against left-hand batsmen, and vice versa with left-arm orthodox spinners.  Considering all of the above, picking Chahar and Krunal - both very strong against right-handed batsmen - seemed unnecessary given that Delhi only possessed three right-handers in their top seven.

While this approach seems absolutely logical to someone like myself who is extremely data-driven and who has spent a huge amount of time studying the value of match-up data, it also goes against what a lot of 'experts' - often prominent ex-players now in the media - would suggest.  Their counter-argument would be that teams should simply possess their best starting XI comprised of their best players, perhaps because match-ups were not such a prominent area of exploitation of opposition in their playing days.  I would argue against this, suggesting that the 11 best players available don't always make the best starting 11.  Interestingly, on the subject of ex-players, in 'Moneyball', Billy Beane suggested that it was a positive if someone tasked with recruiting players hadn't played the game, and I think a lot of ex-players are far too reliant on their own experiences and aren't always willing to embrace new methodology which could enhance their understanding of the game.  However, some are, and they are likely to be the ex-players who rise to the highest level of their subsequent careers.

Despite the ability to understand the benefits that match-up analysis can bring, both in terms of exploiting opposition weaknesses and negating their strengths, quantifying this is a tricky process.  Some people seem to think that a reasonable approach is to look at individual batsmen versus bowler career history, but this method is fraught with huge dangers due to the tiny sample sizes which are almost always present.  A recent example which got a lot of coverage on social media was Virat Kohli versus Sandeep Sharma in advance of the recent IPL Eliminator match between Royal Challengers Bangalore and Sunrisers Hyderabad.  Kohli was detailed as facing 50 balls, scoring 69 runs and being dismissed seven times in advance of the match.  Looking at this data, there are no issues from a strike rate perspective, with Kohli striking at a shade below 140, but there have been previous issues for the RCB captain from a dismissals point of view, averaging just over seven balls per dismissal against Sharma in this sample.  

There are two problems with such an approach.  Firstly, there is no context given here.  So, how many of these seven dismissals were 'worldie' catches, dodgy lbws, or unlucky inside edges onto the stumps?   It may be none, but it's worth considering.  If, say, four of the dismissals were from very unlikely scenarios, the match-up looks far better for Kohli.  Furthermore, how many of the balls in the match-up were when both players were at similar ability levels to their current ability level?  We need this extra context at the very least when such a small sample is being used.  The second problem, and one which is even bigger, is that it is very difficult indeed to predict whether the next 50 balls from Sandeep Sharma to Virat Kohli will produce similar results.  A 50 ball sample size is tiny, and is unlikely to yield much in the way of future predictive value, particularly given that players careers also often follow very different trajectories.  However, it may well be, of course, that some people are simply using these individual player head to head figures simply to promote content on media and social media, as opposed to actually believing in the utility of it.

In my view, there are better approaches.  Unfortunately, those hoping to get detailed analysis of my preferred approach are going to be disappointed - it's far too commercially sensitive.  However, broadly speaking, I focus on player records versus grouped batsmen and bowlers, and general population tendencies.  I provide the teams I currently work for detailed opposition match-up analysis, and specific bowling plans to opposition batsmen, and even go as far as offering printable grids for captains and coaches to quickly refer to when necessary.

As I don't currently work for any IPL teams, I did one of these as an example of my work for the recent match between Rajasthan Royals and Chennai Super Kings on the 19th October, and posted this on both Twitter and Linkedin:-

Rajasthan Batting, Chennai Bowling:-

Chennai Batting, Rajasthan Bowling:-

In the match, the balls bowled by bowlers with green (the most positive) match-ups had an economy rate of just over four runs per over which is a superb economy rate in T20 cricket by anyone's standards.  Conversely, in a low-scoring match, bowlers used for the red (the least positive) match-ups generated an economy rate of almost eight runs per over - virtually double that of the green match-ups.  Yes, it's just one match, but it illustrates the benefits that finding these positive match-ups (and avoiding the negative match-ups) can bring.

A negative bowling match-up in yesterday's final would have been Krunal Pandya versus Rishabh Pant.  We've already discussed how Krunal has much worse economy against left-handed batsmen, and left-hander Pant has a strike rate around the 160 mark against left-arm orthodox spinners in T20 cricket.  Pant, indeed, did perform well against Krunal, scoring 22 runs from the nine balls he faced against Mumbai's left-arm spinner yesterday.  The key for batting teams trying to combat the match-ups derived via astute bowling teams is for their positive match-up players to attack, and try to hit the bowler out of the attack - basically trying to help their team-mates who might struggle more against that particular bowler.  For example, given that Jayant Yadav was brought in by Mumbai Indians in order to try and exploit Delhi's four left-handers in the top seven, Delhi's right-handed batsmen, such as captain Shreyas Iyer, Marcus Stoinis and Ajinkya Rahane, needed to have a plan to try and hit Yadav out of the attack so that he's not brought back on to bowl to the left-handers.  

However, despite how obvious this looks, batting teams often don't do this.  In my view, a great deal of batsmen aren't self-aware enough to take such an approach - the quote suggesting that cricket is a team sport played by individuals springs to mind - and this is where discussing this type of approach with an analyst is extremely useful.  I have had a number of players approach me independently to discuss their own data, and they are very interested in the insights that it can bring to make them more self aware about their strengths, weaknesses and different approaches that they can adopt.  This is obviously great in terms of these players wanting to try and upskill, but they represent a tiny percentage of the player pool.  Interestingly, in the ball prior to Yadav dismissing left-hander Shikhar Dhawan yesterday, right-hander Shreyas Iyer knocked the ball into a gap for a single to put Dhawan back on strike.  This result for an isolated ball for a right-arm off-spinner is a great outcome - they are conceding below a par scoring rate (as they conceded a single) and got a generally positive match-up for the subsequent ball from it.

Players need to adopt such a level of thought to maximise successful T20 strategies - it's like a game of chess, with astute players thinking numerous scenarios ahead.  The T20 format, initially derided as a 'hit and giggle' and not taken particularly seriously when it was first introduced, is actually the format with the greatest need for strategic input, with each ball of a team's innings representing 0.83% of the batting and bowling resources available.  Being able to save an expected value of around 0.3 runs on a ball via a match-up might not sound much, but multiply this by 120 balls and you get an expected saving of 36 runs - a figure which would have a huge impact on a T20 team's expected win percentage.  While it might not be possible to save 0.3 runs per ball on a match-up across the 120 balls, even if a team managed to save 5-10 runs per innings via positive match-ups - very achievable indeed - this is considerably more than a marginal gain.  

Why wouldn't a team try and achieve this?

If this article has given you insight into the data that Sports Analytics Advantage can offer cricket teams around the world in formulating team strategies, selection, draft or auction plans, or any other work, please feel free to get in touch at sportsanalyticsadvantage@gmail.com.