Who Will Win It? An In-game Win Probability Model for Football - - PowerPoint PPT Presentation

Who Will Win It?

An In-game Win Probability Model for Football

Pieter Robberechts, 
 Jan Van Haaren and Jesse Davis

Belgium stuns Japan with exceptional comeback

"

NyTimes

How exceptional was this comeback?

We need an in-game win probability model!

Japan Belgium Time remaining 38 min (+ stoppage time) Score 2 Team strength 1699 Elo points 2009 Elo points Red cards Shots on goal 2 3 Possession 42% 58% ....

A win-probability model provides the likelihood that a particular team will 
 win/draw/lose a game based upon a specific game state

Popular in other sports

⚾ Baseball 🏉 Football 🏁 Basketball

Popular in other sports

A number of relevant use cases.

• 1. Win Probability Added (WPA) metric

• Rate a player's contribution to his team's performance
• Measure the risk-reward balance of coaching decisions
• Evaluate in-game decision making
• 2. Improve the fan experience
• 3. In-game betting
• 4. Identify "clutch" players

... but not (yet) in football

Why? Football is a lot harder to model!

Challenges

• 1. Dealing with stoppage time

You do not know how much time is left


• 2. ...

• 3. ...

• 4. ...

T = 93? 95?

Dealing with stoppage time

Capture the game state in each frame T = 100

Challenges

• 1. Dealing with stoppage time

You do not know how much time is left


• 2. Describing the game state

Find a minimal set of features that describe the current state
 and have the most impact on the game outcome


• 3. ...

• 4. ...

Describing the game state

• 1. Base features
• Game Time
• Score Differential
• 2. Team strength features
• Elo Rating Differential

• 3. Contextual features
• Number of goals scored so far
• Number of yellow cards received
• The difference in number of red


cards received

• Attacking passes
• Duel strength

Event-stream data

xt :

P(Y = win | xt)

that the home team will win the game

P(Y = tie | xt) P(Y = loss | xt)

that the game will end in a tie that the away team will win the game the game state at time t Given: Do: Estimate probabilities

Idea 1: Directly model these probabilities


Idea 1: Directly model these probabilities


Logistic regression classifier

All game states in test set where the model 
 predicts a win probability of 60% About 60% of these games should actually end in a win Well calibrated =

Idea 1: Directly model these probabilities


Logistic regression classifier

Poorly calibrated win / loss 
 probabilities Struggle to predict ties Tie Win Loss

Struggle to predict ties

Idea 1: Directly model these probabilities


Multiple logistic regression classifiers per time frame

Tie Win Loss

Breaks down in late game
 situations

Idea 1: Directly model these probabilities


Random forest classifier

Tie Win Loss

Challenges

• 1. Describing the game state

Find a minimal set of features that describe the current state
 and have the most impact on the game outcome


• 2. Dealing with stoppage time

You do not know how much time is left


• 3. The frequent occurrence of ties

Football games are often very close, with a margin ≤ 1 goal


• 4. Changes in momentum

Goals often shift the tone of a game

Expected number of goals scored after time t

y>t,h ∼ Bin(T − t, θt,h) y>t,a ∼ Bin(T − t, θt,a)

Time remaining Scoring intensities estimated 
 from the game state

Reds Duel Strength Time

Idea 2: Modelling the number of goals scored 
 between now and the end of the game


Scoring intensities estimated from the game state

αt ∼ 𝒪(αt−1, σ2)

θt, home = invlogit(αt xt, home + β + Ha) θt, away = invlogit(αt xt, away + β)

regression coefficients change over time

Idea 2: Modelling the number of goals scored 
 between now and the end of the game


Results

Proper probability
 calibration! 😄 Tie Win Loss

Results

Proposed

Better

How accurately can the model predict the final match outcome?

Fan Engagement


A great "story stat"

Quantifying performance under mental pressure


An Added Goal Value metric

Hypothesis

Scoring / conceding a goal has
 a large impact on win probability High mental pressure Example: Identifying "clutch" goal scorers

AGVp90i = ∑Ki

k=1 3 * ΔP(win|xtk) + ΔP(tie|xtk)

Mi * 90

The total added value that occurred from each of player i’s goals, 
 averaged over the number of games played

Quantifying performance under mental pressure


An Added Goal Value metric

Conclusion

In-game win probability models for football ...

• ...are not straightforward to implement

• ...have useful applications
• fan engagement
• football analytics

• ...will appear everywhere soon