muCricket – Introduction (Real-Time Cricket Analytics)

• THE MU SIGMA TIMES
• August 6th, 2019
•   26781 Views

Yet the landmarks-way of judging made enough sense for stakeholders not to deviate much. But with the introduction of T20s in 2000s, cricket as a sport took a new avatar.

Match analysis is no longer on the couch-commentator’s plate!

Today, especially in ODIs where scores of 300+, once a rarity, are now chased down with precision. Therefore match-analysis also has to follow suit and in its wake, chase precision.

And that’s why muCricket!

So, is scoring 30 runs off 16 balls impactful? Ask muCricket.

No performance can be fairly judged without establishing the CONTEXT surrounding it. muCricket has been developed over the last 3 years with this sole objective of imbibing context into the game’s statistics.

We, at muCricket, don’t ask – “How many runs did you score?” The right question is – “Did you help your team win?”

How does muCricket capture the context?

muCricket captures context through several steps.

1. PAR SCORE: To begin with, recent ground history as well as the teams’ capabilities plays a key role in broadly establishing the par score for every match. A century on a high-scoring pitch is of course easier than on a low-scoring pitch, it’s only a wonder how averages and 100s/50s even existed for so long with such basic flaw!
2. MATCH CONTROL %: Next in line is understanding the context of every ball faced/bowled by the players and its direct impact on the game’s outcome. Historical, ball-by-ball, data helps muCricket understand detailed context for every ball by establishing relationship between wins and the on-going scores – the team’s control on the match at any given point of time. 30/3 in 10 overs in an ODI on a pitch where 300 runs is a par score signifies that the batting team is under pressure. muCricket captures the Match Control % for both the teams at every ball to be able to evaluate player’s performance with thorough context.
3. PERFORMANCE INDEX: While the last step helps us establish exact contributions, now in % improvement of their team’s control on the match instead of conventional numbers, this is not enough to identify the true impact of the performance on the match’s outcome. While scoring under pressure requires quite a lot of determination, ability to start pacing right-away when asked to bat in a comfortable situation is significantly challenging too.

So, what qualifies as enough effort to help teams win? It is therefore that muCricket understands the quantum of the effort put in     the players, through both qualitative and quantitative performance KPIs, through study of performances in winning causes in the past – vast data-points are available given the game’s rich historical database.

muCricket’s Performance Index is a measure of the player’s intend to win, without focusing on individual landmark-driven targets.

1. PERFORMANCE QUALITY: But is that enough? What about the role assigned to the player? And the situation under which he had to deliver it? Bringing performance with similar contexts together helps muCricket measure relative effort put in by the players as compared to their peers in similar situations. In doing so, muCricket identifies performances that were amongst the bests in each categories and the players associated with it! muCricket’s Player Quality is a measure of player’s capabilities as against his exact peers – performance by performance.

Note: muCricket has developed its methodology through rigorous use of muRx, Mu Sigma’s very own Decision Science Workbench. Right from defining the problem in extreme detail to model building and evaluations, it has enabled muCricket to structure the vast cricket data, extract insights and generate algorithms that now power its real-time abilities.

Why the effort?

The primary aim here is to be able to objectively capture the stories behind every performance which is otherwise recognized only in the forms of reports and write-ups. Such information helps muCricket provide actionable insights and thus build accurate predictive analytics around the game. Like other businesses, stakeholders in cricket too can gain a lot from forecasting their potential results.

How did muESP help deploy muCricket?

muCricket has been primarily developed through use of R-codes to run batch processing on entire historical data to test its hypothesis. While muCricket continued to evaluate its work in similar fashion, muESP provided it with platform to operationalize a combined real-time and batch systems architecture – thus enabling us to visualize the data real-time.

muStream’s agent-based paradigm, alongside robust computing capabilities, helped muCricket capture recognized KPIs in a sanitized and structured environment that it needed so as to be able to execute its methodology accurately.

muCricket built several rapid prototypes on muVCL, a quick UI builder tool with dynamic abilities, to both quality check its output in real time, as well as develop final dashboards with ease. Its ability to deploy apps on a single click has provided the pathway for muCricket to bring real-time analytics (through live match coverage) for all of us to experience. Do write to Samitha.Babu@mu-sigma.com for further details.

What next?

muCricket has already managed to capture all the qualitative data, in the form of quantifiable data – for both ODIs and T20s for more than 14 years of historical matches – we will be soon launching the muCricket’s highlights of historical matches, much like our live coverage, for public scrutiny. Till then, hope you enjoy our live coverage @ muCricket/muESP.

What more?

We are already on our way to add predictive layer to our live coverages too! To subscribe to alerts for on-going developments and our live coverage, do drop in a mail to Jatin.Thakkar@mu-sigma.com.