How Analytics is Changing The Way Sports is Played, Watched and Broadcast

Introduction

Do you know why coaches ask for timeouts in basketball when the opposition keeps scoring consecutive points? “We are stopping the momentum”, they say. Cent per cent Analytics. Check if this explanation fits, Momentum = Force x Time. When an opponent is on a roll their force is high, you can’t stop that, but you can certainly put a brake on how long they are exerting the force and take the momentum to 0! That is why a batsman on a hitting streak never wants strategic timeouts.

Analytics has been applied in many businesses, functions and circumstances. Any event that generates data can be used for analysis to aid better decision making. Sports & Games is one such domain, it generates precise data, it generates high volumes of data, it has multiple parties involved and it has a lot of money riding on. However, with so many opportunities presenting in different sports, the journey of analytics is not pacy or smooth.

Machine Learning

A forecast on the sports analytics market, 3 years from now estimates to around $4 bn which augurs well for all the companies specialising in the data capture, storage and streaming. And post these steps, for the companies performing data analytics services to assist the teams, players and management.

Why Sports analytics has improved the playing quality?

The immediate audience for Sports analysis is either the sports teams involved in the games or for the sports betting firms. Although these are the major parties involved, there are others who consume analytics such as viewers, broadcasters, columnists, consultants.

Sports analytics uses in-matches data such as player statistics, in-game statistics; causes & effects of events; environmental conditions such as weather, home/away, playing surface; team’s win/ loss; etc. With this data, there are enough opportunities to derive insights on, but not limited to

  1. BI (Game Intelligence? Team Intelligence?) to track how teams have performed
  2. Tracking reports on player performance, strengths and weaknesses
  3. Predictive models to forecast how teams/ players will perform.

The ultimate objective of winning the game has improved the playing levels. The value of a win trickles down to the fans filling the stadium seats, players bagging the television contracts, broadcasters increasing the viewership, clubs boosting the merchandise sales, associations planning vehicles parking and pricing tickets, teams improving fan retention, and more importantly the emotion of pride.

Sports Analytics Use cases

Visualization to deliver insights

Alex Ferguson and Arsene Wenger fight each other in their football interface tablets much before Man U and Arsenal dole out in the ground. Using raw data in tabular format, they wouldn’t have received clear and quick insights about when Thiery Henry’s left foot will strike or where Ryan Giggs is going to pass. Presenting the data in a graphical format enables the management to grasp difficult concepts or identify new insights. Next step to the graphical representation is the interactive visualization which has become child’s play with the likes of viz apps like PowerBI, Tableau and R Shiny.

Imagine a cricket team checking a matrix of how a batsman has performed against all bowlers, all left arm bowlers, all left arm pace bowlers, all left arm pacer bowlers in Australia. Such drill downs are much deeper than Saudi oil wells. And the performance is not restricted to just averages, but % dots, boundary rate, strike rate, gloves changing rate, bat rotation rate. Believe us, the last 2 metrics also impact performance!

Imagine a tennis coach analysing his player’s match and skill metrics – volleying winning rate vs baseline winning rate, deuce court vs ad court scores, break points conversion. The dashboards can show a pattern of a player’s backhand’s landing positions, unforced errors, net play. Agassi did his own analytics and found out which way Becker is going to serve by looking at where his tongue stuck out!

Dashboards can be of many forms

Teams can utilise to know about the rival teams’ strategy and tactics, individual players’ strengths and weaknesses, their movements. Players just want to execute plans and don’t want to think about many things in the field. Give me plan A, if not plan B!

Broadcasters can show the match events that have happened and also the consequences if they proceed in different scenarios of best, normal, worse cases. Viewers already know the events, they want to be reassured by vivid but understandable graphics!

Content providers can provide insights to the audience through the team and player level stats in a picturesque layout to reduce bounces and exits. Keep in mind, that the audience is not reading they are just scanning the content. You have 1 second to retain them!

ML/ AI Predictive Models

The primary use case of predictive models is for the teams – to deliver insights on the team strategy for different conditions such as grounds/ courts, opponents, tournaments. This in turn, increases the team’s win probability by advising on dynamic tactics as per the match phase, strokes to be played, players to be targeted, equipment to be used, mode of play – attack or defence.

Using the Machine learning models, we can predict which player performs better at which position, on the match day. The model will be built on the player’s stats as the base, how well he performed against the rival team, match conditions like home or away, etc. So, we can predict which players fit into which position, given the game conditions and opponents.

Player analysis

  • On the pitch, the player’s game can be modelled to determine where he is faltering and performing.
  • Off the pitch his fitness level can be monitored by simulating his training pattern and diet chart.
  • Her injury prone areas can be predicted in advance based on her play pattern
  • Neural networks and SVM models can be built using the team stats, to figure the winning combinations with their probabilities.
  • Operations Research models like Linear programming can be used to find the optimal player substitutions and positions
  • Recommendation systems’ innovative usage can help predict an opponent’s play strategies

Team analysis

Fans management analysis

  • With the social media data, we can find the fan-segments using clustering algorithms, list out the targets to run campaigns, customise the offers for the targeted segments.
  • Knowing factors which attracts the fans most, team management can focus on leveraging that aspect. It leads to new fan acquisition and retention of existing base.
ML- AI Predictive Models

Understanding the fan-base network

Every sports team holds its passionate fan base, it needs a way to better connect with them particularly when they are distributed all across the world. Our reactive dashboards allow the team to get the heatmaps of fans on a geographical map and engage one-on-one with them. Use the data gathered on their activity and analyse fan behaviour to create targeted promotional campaigns. This way the management knows, what factors drive their fans to go crazy and what puts them off. Further action can be to fine-tune those factors to increase the engagement rate.

In this regard, social media data comes handy. Tweets, on the official handles of teams, on the trending hashtags generate information to perform sentiment analysis. Unstructured data in the form of reviews, comments, conversations can be picked up using text mining. To understand the trending topics, techniques like topic modelling, summarization, classification come to the fore.

Use cases in gambling can be

  • To forecast the revenue, the activity of the gamers can be used to forecast the number of users and in turn the in-flow of money.
  • To optimise the points for players. Players are given points in events like “pick your team”, instead of giving points based on rank, optimisation algorithms can help in arriving at the right value for the players
  • To give tips to the gamers. Gamers expect some basic information on the players via stats and also some derived metrics like odds for different players and teams.
  • To design the subscription rates. Different add-ons in the betting games are available, for example, x points to bought for bidding on y players. The plans can be designed after predicting the gamer’s threshold for payment.

Leave a Reply

Your email address will not be published.

Shares