Data Journalism Other Articles

Featured

Hockey: Love When We Go OT? 

Soccer: The Rise of The Six, To Start or Sit

To Start or Sit: Squad Rotation in Soccer

Isaac Schmidt ◆ October 20, 2017

On June 1, 2013, Bayern Munich lifted the DFB-Pokal trophy, after beating VfB Stuttgart 3-2 in the final. This was the third major trophy Bayern won that year. One week earlier, Bayern had beat fellow Bundesliga team Borussia Dortmund 2-1 to win the UEFA Champions League. More than six weeks earlier, they clinched the Bundesliga title, constituting a treble season. Bayern’s navigated through a congested fixture list over the season to successfully clinch three titles. However, playing deep into multiple competitions often leads to struggling performances for many teams. Multiple games a week means players must be rested and starting the same team every game in every competition would be impossible, which is why squad rotation is necessary. The extent to which a team should rotate their players has always been a contentious topic amongst fans. Sitting a team’s best player on the bench can lead to harsh criticism, as shown when Arsenal recently lost 4-0 to Liverpool after sitting out star player Alexis Sánchez. Chelsea notably won the 2016-17 Premier League with a very consistent starting lineup, especially after a change in formation early in the season. In this article, I’ll examine whether squad rotation is really necessary, whether teams who make fewer changes win more games, and if consistency should be desired.

What we want to find out is if changes in a soccer team’s starting lineup from game to game have anything to do with any change in performance. Fortunately, there exists a nice numerical correspondence to a soccer team’s result—points. It is possible to look at all the games for a given team over the course of a season, and for each one, check how many changes they made to their starting eleven from the last game, and how many points they gained or lost compared to their last result. For example, if a team loses a game, makes three changes the next time out and wins, those three changes led to a difference of three points. For this test, along with all of the others, I’ve looked at all of the 98 different teams in the “Top 5” European domestic leagues over the 2016-17 season. The results are shown in the chart below.

The results don’t seem to point to much of a trend. Remember, we’re looking to see if lineup changes affect a result, not if they improve or worsen it, so moving from a draw to a win is measured the same as going from a win to a draw. As shown on the graph, the decreasing line of best fit means that conceivably, making more changes in a lineup leads to smaller differences in its results, which means more consistent performances—either good or bad. However, this “trend” is far from statistically significant. A T-test for slope can check to see if there is in fact a relationship between two variables—in this case, lineup changes and differences in results. The result of such a T-test is the p-value, and the smaller that p-value is, the more likely a trend actually exists. For this data, the p-value is .359, which means that there is no way we can say that changes in lineup lead to differences in result. The low r2 value of .106, where 1 would represent perfect correlation, also supports this notion. In short, it’s highly unlikely that the number of changes a team makes to its starting XI has anything to do with a change in performance.

The results don’t seem to point to much of a trend. Remember, we’re looking to see if lineup changes affect a result, not if they improve or worsen it, so moving from a draw to a win is measured the same as going from a win to a draw. As shown on the graph, the decreasing line of best fit means that conceivably, making more changes in a lineup leads to smaller differences in its results, which means more consistent performances—either good or bad. However, this “trend” is far from statistically significant. A T-test for slope can check to see if there is in fact a relationship between two variables—in this case, lineup changes and differences in results. The result of such a T-test is the p-value, and the smaller that p-value is, the more likely a trend actually exists. For this data, the p-value is .359, which means that there is no way we can say that changes in lineup lead to differences in result. The low r2 value of .106, where 1 would represent perfect correlation, also supports this notion. In short, it’s highly unlikely that the number of changes a team makes to its starting XI has anything to do with a change in performance.

We’ve seen that changing a team won’t affect a change in result, but we can take another approach. Over the course of a season, does a team that makes more changes score more points? Once again, we can take a look at all 98 teams in the Top 5 leagues, count how many changes each one made to its lineup from league game to league game, and count how many points they earned. The results are shown in the scatterplot to the right.

As one can see, there isn’t much of a correlation at all. Chelsea, who didn’t play in the Champions League last season and were able to focus on domestic competitions, scored 93 points and ran away with the Premier League, while making the second fewest changes per game out of any team in the set. Real Madrid, juggling not only the Champions League but the Club World Cup, made the most changes by far—and also won their league with 93 points. Celta Vigo also made more than four changes per game, but finished a measly 13th in La Liga with just 45 points. West Bromwich Albion also finished with just 45 points, but made only 1.29 changes from game to game. The r2 value below .01 means that there is not even a point to running a significance test—it is clear there is no correlation. This test, along with the previous one, might suggest that rotating a squad or not has absolutely no effect on team performance. Unfortunately, it’s not that simple. Looking only at domestic league games ignores vital context, namely, other competitions a team might be forced to juggle. Chelsea and Real Madrid’s situations have been made clear, and Celta Vigo’s case might be explained by a deep Europa League run. West Brom is coached by Tony Pulis, a manager with a known reputation of having a strict system that is drilled into his team. Changing the starting lineup too often might have an adverse effect on the system and thus, performance. Unfortunately, trying to eliminate or even just account for such context could easily lead to subjectivity, or a very small sample size, which would render any statistical analysis irrelevant. To conclude, it doesn’t seem like the extent of squad rotation has any general effect on a team’s performance or consistency level. Squad rotation shouldn’t be utilized by managers as a generic tool to increase point total, but should be relied upon on a case-by-case basis.

Never Much Love When We Go OT?

Eric Herrmann ◆ March 26, 2017

The Winners and Losers of 3-on-3 Overtime’s Sophomore Season

This current NHL season has marked just the second year of the league’s drastic new approach to reducing the number of games decided by shootouts. Since the start of the 2015 season, when a game goes to overtime in the NHL, the total number of skaters on the ice is cut down by four and the two teams play five minutes of 3-on-3 sudden death.

The aim of the rule change was to cut down on boring, unfair and unpopular shootouts and increase the amount of overtime scoring to make that part of the game more fast paced and exciting. Halfway through the second season of its implementation, the question remains, has it achieved these aims or not?

The answer? Pretty much a resounding “Yes.” According to statistics from Puckanalytics.com, in three versus three situations this season, the average NHL team is able to score over 279% more goals per that period, meaning that in overtime, the rate of scoring more than doubles. And this is despite the fact that the average NHL goalie is saving nearly of 91% of all shot attempts.

Like with any other rule change, the implementation of 3-on-3 overtime has forced coaches and players to react. Because of how challenging it is to play defense with only three skaters, coaches have gotten more and more conservative during overtime over the past two years. It doesn’t help much that the rule change basically discourages over aggression: if a team pulls their goaltender for an extra attacker during the overtime period, they’ll lose the point earned for the tie at the end of regulation if the opposing team scores into an empty net.

But even after just under two years with this rule change in effect, many teams clearly still haven’t figured it out. As with anything in the league some teams are better than others. So how is your team doing? The chart below maps each team’s regulation goals per period to their 3-on-3 overtime goals per period: *The New York Rangers have played in a league-low four overtime games to this point in the season.

I think this gives a pretty good feel for how well each team plays in overtime versus in regulation. What’s very clear is that some teams are very clearly better than others at scoring in overtime.

For example, let’s take a look at the Colorado Avalanche and the Los Angeles Kings. Neither of those teams seem to be fantastic at generating goals in regulation. Both teams are bottom ten in the NHL in terms of regulation goals per period, but somehow they manage to unleash nearly 5.91 and 6.12 goals per period in overtime respectively. Next, let’s compare that to the Pittsburgh Penguins. NHL’s leading goal scorers fail to score over the league average in overtime goals per period. As a result, the Pens just break even in games where they go to overtime. Overall, when a game goes to overtime or beyond for the Pens, their chance of winning it decreases by just over 23%. That’s likely not the biggest concern for the 5th ranked team in the league, but it certainly should be a thought in the back of Mike Sullivan’s mind when he looks up at the scoreboard and sees that the game is tied with two minutes to go in the third.

In the end, most of the league more or less behaves as expected. Bad teams also play badly in overtime. On the flip side, good teams generally manage to play at least somewhat competently in overtime, which leads to more overtime victories.

I’d hate to single out a team, but Detroit is far and away the clearest example of this. Detroit wins only 29% of their OT games partly because they only manage to score 0.73 goals per overtime period. Their regulation results aren’t much better; the Atlantic Division’s last place Red Wings have only won 25 games all season on 0.78 goals per regulation period. It makes sense that one of NHL’s worst teams should also be one of the worst 3-on-3 teams in the league.

And while we’re handing out superlatives, I’d like to give my “Most Average Team in the League” award to Boston. The Bruins have managed a perfect 0.500 record in overtime by managing to score nearly exactly the league average in goals per overtime period. That’s about as mediocre as it gets.

Ranking teams based off of goals per period is certainly interesting, but what does this all mean in terms of the only statistic that matters? How does this equate into wins and losses? Simply comparing each team’s Overtime Strength Metrics (OTSM = Goals Per OT Period / Goals Per Regulation Period) to their overtime win percentage gives a pretty good insight into how important playing aggressively in overtime is.

Based off of this graph, it’s fairly obvious that OTSM by itself isn’t the most perfect of metrics. But putting quality of defense and goaltending aside, it’s abundantly clear that simply scoring goals in overtime in high volumes has a huge impact on which team walks away with the victory. Alright, so that might just be the most obvious statement of the year in sports journalism. But the key takeaway here is that maybe the trend of going super conservative in overtime could be costing some teams wins and seems to be taking the game in the wrong direction. The thing that sets overtime winners apart from overtime losers is their ability to shift into that extra gear for the final deciding period. In the end, some teams who were able to score prolifically in regulation often seemed to run out of energy in overtime and ended up falling flat on their way to defeat. The teams who managed to dial up their scoring abilities and their aggression in overtime more often than not were able to go the distance and earn the two points. *Minimum of five overtime games played

Citations

Edited by Derek Topper, Sports Analytics at Berkeley

The Rise Of The Six

Kairav Sheth ◆ March 18, 2017

Something happened last May that would’ve put the writers of Moneyball to shame. The Cleveland Cavaliers overturning a 3-1 deficit in the NBA finals seemed extremely insignificant to what had been achieved by a few soccer players across the Atlantic. Claudio “The Tinkerman” Ranieri did something everyone thought impossible. He led Leicester City to the league title against all odds; suddenly, a team that had fought a serious relegation battle a season ago, was lifting the coveted trophy. Where the hotshots of the league like Eden Hazard and Sergio Aguero flopped, lesser known names like Jamie Vardy, Riyadh Mahrez and N'Golo Kanté rose to occasion.

As captain Wes Morgan laid his giant paws on the premier league trophy, a shift in the balance of power was felt. The “Big 6” of English football, Manchester United, Chelsea, Arsenal, Liverpool, Manchester City and Tottenham Hotspur, could only watch as a team that cost less than $70 million to assemble beat odds greater than 5000/1 to lift this trophy. Just to put things in perspective, the squads of the next 3 teams in the league, i.e. Arsenal, Tottenham and City cost a mammoth total of over a billion dollars to assemble, with Manchester City’s Kevin De Bruyne costing the blues $70 million alone.

It was time to sit down and try to accept what had happened. That summer was going to be a busy one in terms of money being splashed around for transfers if parity was to be restored. A 10th place finish for the previous year’s champions Chelsea was more than a wakeup call. In a masterstroke of all sorts, they appointed Antonio Conte (seasonal winner at Juventus and fairly successful with the Italian National Team) as manager. They spent a total of over $150 million in transfers with their most significant signing being Leicester’s midfield maestro, N'Golo Kanté. Younger talents like Michy Batshuayi and Marcos Alonso along with a resigning of David Luiz completed their core transfer recruits.

While Manchester City were the highest spenders in the window, their most accomplished signing had to be the appointment of Pep Guardiola as manager, despite spending around $220 million on key players like John Stones and Ilkay Gundogan.

Manchester United appointed Jose “The Special One” Mourinho as the successor of Louis Van Gaal. Spending a whopping $185 million on transfers, the world saw Paul Pogba come back to Manchester United as the most expensive player. The highlight of United’s transfer window, however, was the signing of Zlatan Ibrahomvic. Arsenal, Tottenham and Liverpool too spent on key signings like Granit Xhaka, Moussa Sissokho and Sadio Mane respectively. Six months later, the effects of these massive investments can be clearly seen. Where Leicester still seems hungover from their exploits of the previous season, the top 6 clubs have significantly improved and have pulled away from the rest of the league. The so-called Big 6 have not only improved when it comes to average league position, but they have accumulated a lot more points this season as compared to this time last year.

Arsenal being the most consistent of the 7 teams shown to the right, we can see that 6 of them have improved vastly. Leicester are in a dogfight on the wrong end of the table and they seem to have lost the drive and thirst that led them to a very unlikely title a year ago. Many would say that equilibrium has been restored as the Top 6 in the table now reads Chelsea, Tottenham, Liverpool, Arsenal, Manchester City and Manchester United. But it is not just these ridiculous investments that have made this possible. Credit must go to these managers who are proving to be masters of the game time and again. The Top 6 are a good 9 points away from the next best team, Everton (the difference was only a point at this time last year) and it doesn’t seem unlikely that each of them could go on and win the league at this moment.

As it stands, each of the 6 teams could better their points tally from last season and put to rest the age long debate of whether money buys you success. 2 of the 6 teams will be gifted with Europa League spots (a consolation no team would be satisfied with) and 4 teams will enter the prestigious UEFA Champions League. A Europa League consolation would be unacceptable to any of the 6 teams but that is what you get when you have 6 great teams vying for the top 4 positions. Regardless of what happens, someone will be left in the dirt.

Edited by Derek Topper, Sports Analytics Group at Berkeley