Why Everything You Know About Break Points Is Wrong

Posted by Brodie under: Statistics

8 Jan 2014

The Championships - Wimbledon 2008 Day One

In team sports, the ultimate goal is to win games. We know this to be true. At the end of the season, your ranking in the standings or table is based solely on your number of wins or total points. In sports like basketball or baseball, there are no ties, and it is a simple win-loss record.

However, we know that the total number of wins or points a team accumulates over a period of time or entire season does not tell the tale of the whole season. Along with the rise of advanced player statistics, we have also seen a rise in advanced team statistics. Things such as corsi in hockey, total shots ratio in soccer, or advanced offensive and defensive statistics in basketball give us a better idea of what a team is good or bad at. There are even more basic measures such as goal differential or total runs scored and given up, and so on.

Ultimately, these statistics give us a better idea of how “good” or “bad” a team was over the course of an entire season. Let’s use basketball as an example. Obviously a team finishing the season with a +200 points total is likely to have many more wins than a team finishing the season with a -200 points total. However predictive or interesting these statistics might be, winning games remains the only thing that matters.

It does not matter if the Bulls beat the Lakers 2 by 90-89 or by 90-50, they still only receive a single win for the performance. We can use these advanced statistics to see if a team was largely unlucky or unlucky in accordance to their predictive statistics. (An excellent modern example would be the 2012 Baltimore Orioles who, despite not having the greatest overall numbers as a team, won an incredible amount of 1 run baseball games to essentially “cheat” the system and make the playoffs.)

Now, let’s talk tennis.

A similar bias exists in a single set of tennis. In our basketball example, the amount of points you score in a season does not matter – what matters is the amount of games you won, which are essentially subdivided events. A set of tennis works in the same way. It does not matter how many points you win in a set, but only that you win more games than your opponent. (It makes sense that they’re called games, no?)

Much like it doesn’t matter if the Bulls win by 90-89 or 90-50, it doesn’t matter if Roger Federer holds in 45 seconds with 4 aces, or after 17 deuces – a single game is a single game.

When we look at set (and eventually match) statistics, we are aware of this, but only partially. Break points are always an excellent example of how a player did in a set. Ideally, you want to create break points and take them and stay away from facing break points on your serve.

But there is a problem.

Break points are recorded in a peculiar way. We receive the amount of times a player won a break point against a slash line of the amount of break points played. While we will always know how many times a player broke, the second number of the amount of break points played largely tells us nothing.

As stated, winning games is the most important end goal, not necessarily just winning points. In other words, if a returner plays a long return game and needs a total of five break points to break, the outcome has the same effect on the match as if the returner had broken on the very first break point.

This means that one of the most important statistics to keep track of is the number of games in which a break point was created. In the previous example, even though the player needed five break points, they did not “waste” the other four, as they ended up winning that game anyway. Still, at the end of the set, a “1/5 Break Points” statistic would be displayed on the screen. This tells us the player was ineffective at taking break points, when in reality they ended up winning 100 percent of the games in which they earned a break point (in this case one).

It doesn’t matter if a team wins 90-89, or 90-50, and it doesn’t matter if a player needs 14 break points or one – what matters is if they broke in that game or not.

The solution?

Recording and telling the viewer how many games a player created one or more break points is important, and easy to do. Let’s dream up a new scenario.

Let’s say in the first set, Tommy goes up 0-40 in his first return game, and breaks on his third break point, and in his third return game, he creates two break points at 15-40, but fails to take them both. He creates no other break points.

His traditional number would be 1/5 on break points. This tells us that Tommy created break points from anywhere between one and five return games. However, if we put the number of games in which a break point was created, in this case two, we gain perfect context. In this case, it would look like 1/5(2). This tells us that Tommy went 50% in the games in which he created one or more break points, which isn’t bad. Winning games is what matters.

Let’s go a step further…

In our previous example, Tommy went up 0-40 and took the third break point, making him 1/3 on break points. However, if Tommy won the first break point, he would have been 1/1. As we discussed, it does not matter which of the three break points Tommy takes, the outcome is the same – Tommy receives a break. Oddly, when Tommy goes up 0-40, the screen might display “3 Break Points”, yet there is no recognition in the final statistics of the set that Tommy created these “3 Break Points”.

In other words, creating break points is more important than actually playing them, because even if he won the first point, his good returning is not reflected in the statistics.

What I recommend is a new slash line using break points won, break points created (not played) combined with the total number of games in which one or more break point was created.

Let’s finish with a broader example of what statistics might look like at the end of the match, using this scenario –

Tommy creates 3 break points at 0-40, wins the first one to break.

Tommy creates 2 break points at 15-40, fails to break.

Tommy creates 5 break points in a long deuce game and breaks.

Tommy creates 3 break points at 0-40, wins the second one to break.

Tommy creates 3 break points in a long deuce game, fails to break.

Amount of breaks earned – 3

Amount of games in which one or more break points were created – 5

Amount of break points played – 13

Amount of break points created – 16

A traditional break points breakdown of this match would look like 3/13. This tells us that Tommy broke three times – not bad – but it tells us that Tommy was wildly ineffective on break points.

My first revision would look like this 3/13(5). This tells us a bit more. We see that Tommy broke 3 times – not bad. It also tells us that he played 13 break points, which is quite good, but also importantly that he earned one or more break points in five games. This tells us immediately that he broke in 3 of the 5 games in which he created one or more break points, which is actually good.

Finally, a final slash line might look like this – 3/16(5) or 3/13/16(5). This tells us that Tommy broke in 3 of the five games in which he earned one or more break points. He earned a total of 16 break points, excellent, though only played 13.

We should start to interpret points played and won much like points are awarded in basketball. They can tell us about a performance of a team over time, but ultimately winning games is what matters. The same is true of tennis. The amount of break points that Tommy played is not important – what is important is whether or not he went on to win that game. You’re only wasting break points if you fail to win that specific game.

Let me know what you think!

The Changeover Podcast:

Episode #56 – Indian Wells Wrap