top of page

Cricket by the Numbers: Unveiling Winning Percentages & Predicted Scores using Statistics

By: Tayyab Khan


As fans, spectators, viewers and most importantly self

proclaimed pundits, almost all of us who’re into cricket and

sports in general have this desire to predict outcomes/

scores of games using our expertise for the source of

entertainments. On a professional level, teams,

broadcasters and betting companies in general also

anticipate in predicting scores and outcomes of games. For

teams, this serves the purpose to gain a competitive

advantage by making informed decisions about team selection, tactics, and resource allocation. Broadcasters on the other hand predict scores for engaging the audience into the game.

Data

With more advanced technology available, more advanced and hence detailed data and metrics are available. Event Data is the data that captures the key events that happen in a cricket match, such as balls bowled, runs scored, wickets taken, and field placements. Event data is collected in real time and is used to generate live scores and commentary. The statistics, like wickets taken, number of runs scored in a particular direction are derived from event data. Another more advanced, complex but less generally used is tracking data. This data tracks the movement of the ball and the players on the field. Tracking data is collected using cameras and sensors, and it can be used to generate replays and analyze player performance. This data is what’s used in DRS calls as well. This is the current method used by the ICC to revise the target for weather interrupted cricket matches. It is based on the idea that the batting team has two resources: (i) a certain number of overs to face and (ii) a limited number of wickets in hand. Based on a mathematical model, the Duckworth-Lewis method provides a table with the remaining percentage resources at any given stage of a game. This data in future might be used as a metric to analyze the performance of a fielder as well.


Predicting Cricket Scores

Broadcasters and teams essentially use game models from leading cricket data providers like cricviz, statsperform, ESPNcricinfo to name a few. These data

providers or agencies create the game models which are built around and trained using machine learning algorithms of historical cricket data. Game models can be used to predict the probability of a team winning a match, as well as the predicted score of the match. This information can be used by fans, bettors, and teams to make informed decisions. For example: A cricket data provider might use a game model to predict the probability of India winning their next match against Pakistan. The model would take into account a variety of factors, such as the pitch conditions, the weather conditions, the head-to-head record between the two teams, and the recent form of the teams. The model would then predict the probability of India winning the match. This information could then be used by fans, betters, and the Indian team to make informed decisions.

For instance, if India is playing against Australia, in Melbourne and India scores say 150 runs in 25 overs with the loss of 2 wickets, the algorithm will lookout for performance of India in last 25 overs when 8 wickets are in hand, then overall performance of India against Australia in general, India's performance in Melbourne, the usual performance in Melbourne by any average team, the impact of whether on game at that particular ground and lot lot other parameters. Once these are collected, all the parameters are bucketed and each bucket would have a specific weight, based on which the prediction is carried out and runs to be scored are calculated and accordingly the win/loss percentages are calculated using the appropriate machine learning / model algorithms.



"Win probability is a complex metric that is influenced by a variety of factors, including the pitch and weather conditions, the head-to-head record between the two teams, the recent form of the teams, and the team composition. CricViz uses a variety of statistical models and machine learning algorithms to take all of these factors into account when calculating win probability." ~CricViz

“The pitch is arguably the most important factor in determining the outcome of a cricket match. A good pitch will provide a balance between bat and ball, while a poor pitch can favor one side or the other. For example, a batting-friendly pitch will make it easier for batsmen to score runs, while a bowling-friendly pitch will make it easier for bowlers to take wickets." ~ESPNcricinfo


Guess you're now ready to predict who's going to win the world cup! ;) Let us know of your predictions in the comments below!!!






14 views0 comments

Comments


bottom of page