Alb's Thoughts

The Indian Premier League is a salary cap league. In 2023 each team was able to spend roughly USD$11.5m (950 million rupees) for the season. A player auction is held prior to each new season with all teams bidding for player contracts. The amount the player is won for at auction is the amount they’re paid for the season. The lowest paid contracts are USD$26,000 (2 million rupees), and the highest paid contract for the 2023 season was Sam Curran from England who was paid USD$2.256m (185 million rupees). Contract length is not publicly available information however it is highlighted when a player is retained from one year to another. It can then be inferred how long the player was under contract for with contract lengths appearing to vary from 1 to 4 years, however this has not been able to be confirmed.

The aim of this article was to investigate whether there is any correlation between money spent on player contracts in the Indian Premier League, and winning. I acknowledge that winning may not be the only consideration given to how much teams bid on a player – marketing may also play a part. Contract information was available for 7 seasons from 2017-2023, therefore this was the time period used for this analysis.

At a team level, the amount spent on Player Contracts does not appear to be correlated with Team Winning Percentage. This is what a salary cap in any league should be trying to achieve, parity across payrolls.

Indian Premier League contracts are heavily positively skewed – the median contract costing $328,000USD, and the mean contract costing $537,000USD.

The above two plots both represent the same information, but in two different visualisations: Indian Premier League contract amounts by position. My main takeaway here is how much more densely populated the Bowler contracts are. The median contract for Bowlers is $275,000USD compared to $388,000USD for non-Bowlers. What’s more shocking is when comparing 75th percentile contracts: $583,000 for Bowlers, compared to $975,000 for Non-Bowlers. This presents two questions for me:

1. What contributes more to winning games: Bowling or Batting?
2. Is bowling contract amount correlated with performance?

To quote Wikipedia: Win probability added (WPA) is a sport statistic which attempts to measure a player’s contribution to a win by figuring the factor by which each specific play made by that player has altered the outcome of a game. In the sense of Twenty20 cricket – at each ball during the game the win probability can be calculated by applying a model. Win probability added is the difference between the win probability prior to a ball being bowled, and prior to the next ball being bowled. This can then be evenly split to Batting WPA, and Bowling WPA. Other factors such as fielder position, and skill, are also likely to play a part in Win Probability Added, however current open source Ball by Ball data does not include fielding information. Win Probabilty accounts that a wicket, or runs, at one point in the game, does not have the same impact on a team’s likelihood to win a game, as it does at another point in the game. A wicket, or a boundary (4 runs or 6 runs) early in an innings, usually is less impactful than latter in an innings, as the opponent still has lots of time to adjust for the new information provided by that event, whereas later in the game, there is less opportunity to make up for that event.

Using data from both the Indian Premier League, and the Major Twenty20 situations across the world – it appears that Bowling Win Probability Added is more highly positively correlated with Team Win Percentage than Batting Win Probability Added. Total Win Probability Added is highly positively correlated with Team Win Percentage in both the Indian Premier League (R^2: 0.82) and across the major Twenty20 competitions (R^2: 0.76). Therefore, we can use Win Probability Added as a fairly reasonable proxy for team success. 

The second part of our first question is: are high contract values correlated with higher Bowling performance?

Using Win Probability Added as our primary metric it does not appear that more expensive contracts provider to Bowlers or All Rounders results in higher levels of Win Probability Added through their bowling efforts.This is reflected on both a Total Bowling WPA level, and on a Bowling WPA/Ball level.

So far we have: Bowling WPA is more correlated with Team Win percentage than Batting WPA, however by the same token, between 2017-2023, contract value was not correlated with Total Bowling WPA or Bowling WPA/Ball. Therefore, at this stage of the analysis, perhaps I’m content to go pretty cheap when paying bowlers? Given Bowler contracts are already substantially lower than other contracts, it appears the IPL decision makers agree with this strategy, however perhaps I’d propose they could be even more aggressive in their how little they target the bowlers? Ultimately though, you still need to pick the bowlers who are going to provide Win Probability – can we know that ahead of time? Is bowling WPA stable year to year?

Based off these charts: No. It does not appear Bowling WPA is stable year to year, either in total, or on a per ball basis. Therefore, if I’m an IPL team, my current plan would be to go very cheap when it comes to Bowlers.

Over the past 7 years worth of data, which bowlers have been the highest paid in the IPL? How much Bowling WPA have they provided?

Out of the top 10 most expensive bowler contracts since 2017, only 3 of them have provided a positive Bowling WPA, with the highest coming from Sunil Narine in 2019 with a Bowling WPA of 43 points. 

What about for All Rounders? Keeping in mind that all rounders are also employed for their batting.

A wider range of bowling WPA in the All Rounders group. Ben Stokes in 2017 and Pat Cummings in 2021 provided significant benefit to their respective teams with their arms.

However, since 2017 which bowlers have provided the most total Bowling WPA in each season? And what have they been paid?

As you can see, the most productive bowlers since 2017 have not been the highest paid. Only one of them has been paid over USD$1 million, Trent Boult in 2022. The average contract of the Top 10 WPA Bowlers has been USD$531,200.

So teams shouldn’t be using high amounts of their cap on Bowlers, who should they be using their cap money on?

The above charts demonstrate that there does not appear to be any correlation between Contract amounts of those tasked with Batting, and their total Batting WPA, or their Batting WPA/Ball.

For the Batters I also reviewed whether there was any correlation between contract amounts of Expected Runs Added (XRA) whilst batting. 

Expected Runs Added is a new term in this article. Similar to Win Probability, at each point in a Twenty20 match, based on historical data, each ball can be associated with a number of runs that should be expected to be contributed at that point in the game. The number of expected runs depends on the innings number, how many balls left in the innings, the amount of runs already scored, the amount of wickets already lost, and if in the second innings, how many runs needed to win. Expected runs added is the amount of runs scored on a given ball, minus the expected runs of that ball given that situation. The mean expected runs on a ball are 1.09 runs. I prefer to only use XRA for Batters and not for Bowlers, as on an average play a bowler can only contribute a bowling expected runs added anywhere between 0 and 1.09, this is a small range. Batters however could contribute up to 4.81 Expected Runs Added on an average play. Therefore, it’s not fair to compare Batting and Bowling XRA.

The first chart indicates that there has been some correlation between IPL Contract Amounts and Total Batting XRA. The second chart however indicates that on a per ball basis, there doesn’t appear to be any correlation.

Interestingly, there is correlation between IPL Contract amount and the amount of balls they’ve faced whilst batting. This helps explain why there is some correlation between Contract amount of Batting XRA, however that there isn’t between Contract amount and Batting XRA/Ball. Is there a correlation between Contract amount and and amount of balls faced because those that are paid more are more likely to bat earlier in the order? Or do they keep their wicket for longer? Even if they did keep their wicket longer than those on cheaper contracts, given there doesn’t appear to be any correlation between Contract amount and Batting WPA, Batting WPA/Ball or Batting XRA/Ball, are perhaps these expensive Batters merely taking up space at the top of the batting order without actually contributing meaningfully to their team’s success?

This phenomena seems similar to how it seems like NFL (American Football) teams seem to remain committed to playing their highly paid running backs, even if their cheaper, backup running back appears to be more efficient (see Ezekiel Elliot & Tony Pollard the last few years), possibly because of sunk cost fallacy with the contract they provided them, I wonder if IPL teams should be challenged to play their best Batters earlier in the order even if they’re cheaper, and less “sexy”. This comes with some caveats though – Win Probability is generally more likely to have larger swings later on in an innings which could explain why some of the cheaper batters, if they are truly batting further down the order, may have more opportunity to accrue WPA than those batting earlier in the order. In saying that, the opposite is also true – those batting later in the order could be prone to losing more WPA in high leverage situations.

As I did for bowlers, is Batting WPA or Batting XRA stable year on year? As this information is crucial to know from a team building perspective.

Batting WPA, total, and on a per ball basis, appears to be unstable year to year.

Interestingly, in the IPL, batting XRA appears to have much more stability (although, remains low) from year to year than WPA, and this is maintained on a per ball basis too.

It’s understandable that total batting XRA has some stability year on year as number of balls faced is actually very stable year on year across the four major Twenty20 competitions.

Virat Kohli has 4 of the 10 most expensive Batting Contracts since 2017. Although he has primarily been a net positive for his teams from a Batting perspective, on a per ball basis he’s been middling.

High batting WPA seasons have come from a range of contract amounts. Two of the top 10 were paid less than USD$70,000 for the season, and three were paid over USD$1.7m. The average contract of the Top 10 Batting WPA Players has been USD$931,700, this is $400,000 higher than the average of the Top 10 Bowling WPA Players.

Summary

• Bowlers are the cheapest position in the IPL
• Bowling WPA is slightly more correlated with Team Win Percentage than Batting WPA
• Contract amounts are not correlated with Bowling WPA
• Bowling WPA is not stable year to year
• Contract amounts are not correlated with Batting WPA
• Batting WPA is not stable year to year
• Contract amounts have some positive correlation to Total Batting XRA, but not to Batting XRA/Ball
• Contract amounts are positively correlated with number of balls faced as a batter, which helps explain the positive correlation of contract amounts and Total Batting XRA
• Batting XRA has a small amount of stability year to year
• Number of balls faced as a batter is fairly stable year to year

Further questions that need answering:

• Why is there a correlation between IPL Contract amount and balls faced whilst batting? Are those that are paid more likely to bat earlier in the batting order?
• Is there a correlation with Batting WPA and Batting WPA/Ball and position in the batting order?
• Considering there is some stability of Batting WPA and XRA from year to year, is there a correlation between Batting WPA and XRA and their contract in Year N+1? This is a difficult question to answer as we would need to know when a player was eligible for a new contract – information that is not publicly available for the IPL.

Conclusion – Which positions should IPL teams pay the most? Which should they pay the least?

I’m not sure. However, I do think there presents an opportunity in the IPL for a team to attempt to “moneyball” their way to a championship without breaking the bank.

Acknowledgements

Game data was sourced using the cricketdata R package (http://pkg.robjhyndman.com/cricketdata/)

Contract amounts were sourced through ESPNCRICINFO.

Win Probability and Expected Run Models were created myself.