Major sports betting sites (such as William Hill, Bet365, Bwin) dynamically alter their pre-set betting odds to reflect the probability of the instantaneous… · More match outcome as well as to optimize their overall profits.
We developed and trained a machine learning algorithm that dynamically incorporates match data from a live feed and instantaneously re-calculates the match odds every 30 seconds (given that the average reasonable time frame for any of the most significant game defining actions such as scored goals, yellow cards, red cards, fouls, freekicks, corner kicks can happen within that time frame).
We optimized the algorithm to overall maximize the profit levels based on the re-calculated odds.
Currently, live (play-by-play) sports data is generated merely by a bunch of sports data entry agents and data scientists experiencing the match/game live in… · More the stadium (or turf, pitch, field, court, rink, track etc) and then entering the statistical data by hand into a laptop/mobile device which is then connected to some remote servers elsewhere. This manually entered data forms the basis of all sports analyses done by professional clubs, scouts, federations, agents, sports betting agencies/companies, pundits, journalists, media houses, federations etc. OPTA, an industry leader in the generation of live sports data licenses this data to various companies and agencies on a tiered, business model. Given our discontent with the antiquated system of live data generation in the sports industry (given the availability of supportive technologies), we're prototyping an automated, play-by-play sports data scraping algorithm (using association football as a paradigm) that autonomously detects, scrapes and analyzes variables such as player position in 3-D space, number of completed passes, speed, acceleration, cumulative amount of time a player spends with the ball (in milliseconds), possession, etc. This algorithm is built on computer vision (a subset of machine learning). Based on the inputs obtained, we're able to generate better insights on team formations, tactical set-ups throughout the course of the 90 minutes, effects of substitutions, key players at crucial stages of the match and also players that ought to be substituted
Using data supplied by an institutional sports client, we helped provide insightful fan analytics on ticket, merchandise and food/beverage sales prior to… · More each match/game day to enable the associated sport teams and clubs increase the profitability and further enhance their brand loyalty and fan engagement levels. The strategically placed beacons obtained data from opted-in fan smartphones within specially designated fan zones. Analytics on beacon data are a fantasic way to foster fan loyalty and digital engagement in an era where fans heavily rely on their smart phones on the latest updates regarding their favourite teams and/or players.
Using historical as well as live data obtained from wearable trackers on athletes training for some collegiate athletic events, we attempted to provide some… · More useful, predictive insights on who the best performing athletes were for their upcoming collegiate athletic events. We assessed crucial variables such as acceleration, speed, displacement, muscle tension, lactic acid build rates, breathing rates, heart rate, etc to construct a machine learning, predictive algorithmic model trained on historical data to make some meaningful predictions on who the best runners were for the various 100m, 200m and 400m races.
Working in consortium with a university AI team preparing for the forthcoming 2018 edition of the RoboCup (Robot Soccer World Cup), we helped design an… · More algorithmic model based on the Finite State Machine (FSM) with several iterations that defined not only how these robotic automatons would play but their formation and tactical set-ups as well
We developed a pre match, machine learning prediction algorithm which scored a 63.4% accuracy rate in predicting the standard WIN (W), DRAW(D) and LOSS(L)… · More outcomes of English Premier League (EPL) matches in general over the 2016/2017 EPL season.
We addressed the problem of deciphering the standard W/D/L outcomes in the EPL by tackling the problem as a multi-label classification one. Via the agency of a comprehensive team and player training data set & feature vectors, we successively trained a modified support vector machine (SVM) algorithm before deploying it on line-up information roughly an hour preceeding to each match kickoff.
We understand that the different team combinations of 11 mean different winning probabilities. It is therefore no surprise that all teams deploy their best 11 possible players (10 outfield + 1 goalkeeper) each match day. Out of the standard 25-man EPL squad size, only 10 outfield players + 1 goal keeper can begin on the pitch each game.
Given the huge potential player combinations that can be generated out of each 25-man squad size (Outfield players: 22C11=705,432 combinations; Goalkeeper position: 3C1=3; Total of 705,435 combination), it makes less sense to attempt to predict matches before team line-ups are announced(approx. an hour before the game) given the inability to account for stochastic factors/occurences.
Various stochastic factors such as player fitness levels/injuries, disciplinary problems, coach decisions, unexpected events etc affect the certainty of each successive match day line up so we endeavoured to make predictions as soon as the line-ups where announced an hour before the game.
We automated our system such that we relied on instant line-up information from the Twitter API so we essentially had the algorithm automatically make predictions without having to manually enter in the player line-up details.
We are currently focused on error optimization, improving our feature vector set and ensuring the algorithm has the best possible fit to establish an appreciable percentage point difference (+20PP-->70%+ accuracy rate) compared to the mainstream betting industry (averaging ~50% accuracy rate).
Using the EPL as a paradigm (arguably due to its massive global appeal, availability of comprehensive data and higher reputability/officiating levels), we envisage standardizing the pre-match algorithm and applying it to other leagues such as La Liga, Serie A, Bundesliga, Ligue 1 etc).
Our pre-match algorithm will be of greater utility to;
1. Professional sports bettors 2. Quantitative traders 3. Sports pundits 4. News media agencies
What I Do
Galvanizing teams and providing direction to ensure the best solutions are hacked out to address the needs of the consumer