Don’t Waste Your Data: Using Tracking Data to Find Key Moments in Soccer with Open Space

Our research and devel­op­ment team was select­ed to present at the 2017 OptaPro & MIT Sloan Analytics Conferences on their recent work ana­lyz­ing pass­ing prob­a­bil­i­ties in foot­ball (soc­cer).

Don’t Waste Your Data: Using Tracking Data to Find Key Moments in Soccer with Open Space

Our research and devel­op­ment team was select­ed to present at the 2017 OptaPro & MIT Sloan Analytics Conferences on their recent work ana­lyz­ing pass­ing prob­a­bil­i­ties in foot­ball (soc­cer).

At Hudl, we believe more than any­one that under­stand­ing match­es is always tied back to video. Accordingly, when we build ana­lyt­i­cal mod­els, our intent is to help our ana­lyst col­leagues iden­ti­fy and high­light moments in video that can be used to pre­pare for the next match or learn from a pre­vi­ous one.

Moment Discovery

This year at the MIT Sloan Sports Analytics Conference, I was hon­ored to present a research paper about pass­ing prob­a­bil­i­ties in foot­ball. The pass­ing mod­el allows us to com­pute the prob­a­bil­i­ty that a pass will suc­ceed or fail and which play­er is best sit­u­at­ed to receive the pass. This answers many ques­tions when tied back to video. 

For exam­ple, if we’re inter­est­ed in know­ing how the oppo­nent was able to cre­ate dan­ger­ous attacks, we can look for sit­u­a­tions in which a high prob­a­bil­i­ty pass was com­plet­ed to their strik­er and/​or winger in the attack­ing third of the pitch. By high­light­ing these moments, we can ana­lyze the video to deter­mine how such a high prob­a­bil­i­ty pass was allowed to occur in the first place.

Adding Context

In addi­tion to find­ing moments in the video, we can add con­text to sequences of events. This could be used to aug­ment our abil­i­ty to find inter­est­ing moments, or to high­light infor­ma­tion about a sequence of events. Let’s take a look at a spe­cif­ic exam­ple to illus­trate how know­ing the prob­a­bil­i­ty of a pass can add impor­tant con­text to a sequence of events. 

This data comes from a pro­fes­sion­al men’s league. We have anonymized the the data and will refer to play­ers from the Red Team and the Blue Team by their ran­dom­ized jer­sey” num­bers. The Blue Team is attack­ing right to left and has pos­ses­sion of the ball. Its first pass is from defend­er Blue #26 who directs a pass to mid­field­er Blue #17.

Pass from Blue #26 to Blue #17. The probability that the pass succeeds is 43%.

This is a fair­ly easy pass but the pres­ence of Red #1, who could inter­cept, and Red #11 and Red #12, who could chal­lenge, mean that the receiv­er must quick­ly con­trol the ball if he’s going to keep pos­ses­sion. Blue #17 runs back and chips a short pass to Blue #24. This is an easy pass and has an 85% chance of success.

Pass from Blue #17 to Blue #24. The probability that the pass succeeds is 43%.

As Red #8 clos­es, Blue #24 piv­ots and notices that Blue #28 is in excel­lent scor­ing posi­tion. Blue #24 sends a pre­cise ball deep to Blue #28. This is a high­ly valu­able pass because it could result in a strong scor­ing chance, even though it is unlike­ly to be com­plet­ed due to the pres­ence of defend­ers (Red #2 and Red #7) and the goalkeeper.

Pass from Blue #24 to Blue #28. The probability that the pass succeeds is 26%.

Blue #28 suc­cess­ful­ly receives the pass and man­ages to shoot and score. These prob­a­bil­i­ties are for the aver­age play­er in the league, and we would expect them to be high­er or low­er depend­ing on the receiver’s skill level.

Open Space

Using the same ball con­trol mod­el that is used to com­pute the pass prob­a­bil­i­ties, we can quan­ti­fy how the space on the pitch is con­trolled by the Red and Blue teams by sim­ply cal­cu­lat­ing the con­trol each team exerts on every loca­tion on the pitch. At the moment of the third pass, we can see just how large the region open to Blue #28 is.

Of course, by the time the ball reaches him, defenders Red #2 and Red #7 have closed the distance which is why the pass has a low probability (26%).

Open Space Analysis

Using this open space mod­el, we can visu­al­ize how the pass to Blue #28 became pos­si­ble in the first place. Looking at the open space mod­el, we can see that Blue #22 is mov­ing down the right side of the pitch. This draws away defend­er Red #7, which opens up space to the right of strik­er Blue #28. Defender Red #2 could be posi­tioned clos­er to the goal, which would cut off Blue #28 from hav­ing an open lane, but such posi­tion­ing would give Blue #18 and Blue #27 a 2-on-1 matchup on the left side of the pitch.

Open space during the passing sequence leading to a Blue goal.

The last pass in the pass-chain lead­ing up to the score had a mere 26% chance of suc­cess, but even if a goal had not been scored, this would be a sequence of events that could be ana­lyzed after the match. Because we can com­pute the pass prob­a­bil­i­ties, we can iden­ti­fy sim­i­lar dan­ger­ous oppor­tu­ni­ties to pass into the box that rep­re­sent a scor­ing chance even if no actu­al pass or shot takes place.

Conclusions

By pair­ing video with track­ing data, we can assist ana­lysts in find­ing and eval­u­at­ing crit­i­cal moments dur­ing a match. 

These mod­els not only add a new dimen­sion of analy­sis to the data, but also make it faster and eas­i­er than ever to find key moments in video for ana­lysts to review. Teams are no longer lim­it­ed to on-ball events, as we can use these mod­els to find oppor­tu­ni­ties even if they were not exploit­ed dur­ing the match.

We’re excit­ed to announce our research in pass­ing and spa­tial analy­sis, and we look for­ward to using these tools to stream­line how you can engage with your data. If you would like to incor­po­rate these tools into your work­flow, please let us know. 

If you’re inter­est­ed in learn­ing more about our research efforts in foot­ball, fol­low me on Twitter or reach out to our prod­uct man­ag­er, Armen Badeer.