T20 Player Value: Part I
The Global T20 draft weekend was an opportunity to road-test my player evaluation method. An exercise in comparing my valuations for each pick against the available pool of players resulted Chris Morris and Colin Ingram rising to the top as the standout players of the draft, and two local bowlers, Ngidi and Leie, emerging as hidden gems deeper into the weekend
In this and subsequent posts, I aim to explain those evaluation methods. They are not set in stone. Any time I sit down to analyse a player, team, tournament, strategy, there is a decent chance that they will change. I would love to hear other people’s feedback and ideas. If nothing else, writing down my thoughts has forced me to be critical of my own work. Indeed, the methods changed several times even as I documented them
A pre-emptive note. Player evaluation for almost any purpose should welcome a menagerie of metrics. Contextual information can provide useful insight. Players may excel in certain situations and not others. There are many ways a player can add value. My aim here is not to establish best practice for player evaluation. I am simply outlining how I construct a single catch-all number that captures as much as possible about player performance as it can. It allows me to rank players, nerd recreation, and it allows a quick snapshot of roughly where the value lies in a team or draft before digging further
The world of cricket is migrating away from the traditional average as the ultimate performance measure in T20 cricket. Economy and strike rates should be the go-to stats for any self-respecting fan reading a scorecard. In a typical T20 innings with limited overs, the humble ball often surpasses the hallowed wicket as the most valuable resource in the game
My approach goes a step further and attempts to identify the exact number of runs that a player is worth to their team throughout a typical match. This distils strike rates, economies, and averages into one metric – naturally favouring the first two over the third
Measuring Performance in Runs
Win Probability and Expected Total models tell us how much a play impacts the overall outcome of a game or the eventual innings total. My approach uses these models to understand how much each player impacts the game, in terms of Win Probability Added or Runs Added. We simply compare the Win Probability / Expected Total before and after each ball that a player is involved in, and assign credit as appropriate
In the first innings, the objective is to set the highest total possible (or limit to the lowest total possible for the bowling side). Runs Added is the metric of choice. But in the second innings, Runs Added can make less sense, as a team deliberately paces the chase to reach the target using as many overs as they need. Win Probability Added is now more sensible and I use a relatively simple model to convert those Wins in the second innings into Runs
Let’s say a target of 158 represents a 50% chance of winning for the chasing team (it often does). Now imagine a team chasing 142, a much lower target. They have ~65% probability of winning the game. Therefore, we can equate 15% of a Win with roughly 16 Runs – or 1 Win with 107 runs. But this is not a linear relationship: the exact value of a Run will vary depending on the game. In an even game, each Run is worth more than in a lopsided game
Complete Innings Contribution and 4-Over Contribution
My final performance metric is designed to represent the average contribution (in runs) from a completed innings for batsmen – how much value they add over the course of a typical innings – and the average 4-over contribution (in runs) for bowlers – how much value they would add if required to bowl the full four overs
The approach describe above gives a baseline for performance but it is not the end of the story. Colin Ingram has one of the best strike rates in world over the last two years. His typical innings was worth over 7 runs to his team, significantly more than, for example, Chris Gayle. Does this make him the better batsman?
Possibly. But as Dan Weston notes, in his own, excellent piece on player valuation, Ingram’s high strike rate is partly a function of him playing in weaker, bat-friendly competitions such as the Blast and the Ram Slam. This underscores just one of several variables that must be considered when assessing player value across different domestic leagues
These variables also cause biases. Ingram’s inflated statistical profile may cause teams to over-estimate his value. Although, for the record, I still think he is rather valuable. In subsequent posts, I will cover the various adjustments that I make when assessing player value. Each adjustment represents a potential recruitment bias. Irrespective of whether they use a single metric or many, teams who make decisions cognizant of these biases are far more likely to make good than those who don’t