You’ve got to show me some stuff, and I’ll know it when I see it.

Steve Jobs, yelling at a copywriter about an iPad commercial.

By now, we all agree that numbers don’t tell us enough about how basketball players defend. Keeping track of blocked shots, steals and defensive rebounds provides us with only a fraction of the information we need in order to evaluate a player’s defensive performance. How many fouls does a player commit going for these blocks? How many lay ups does he give up going for a steal? And how many of those rebounds come at the expense of proper defensive rotations? The same is true for team defense. Different strategies and game plans are not taken into account by ‘mainstream’ or even ‘advanced’ statistical categories. And of course, statistical information about that gray area between defense and offense is sketchy at best. For example, points off turnovers would be a more useful tool if we knew more about what exactly led to each turnover.

You might say that the answers to these questions are simple if one who actually watches any given game. That’s partly true. After all, keeping track of advanced defensive stats requires video analysis. But I don’t see any contradiction between video and statistical breakdown. Numbers provide added incentive to watch the game closely and learn more about it every time. Isn’t that what makes basketball great?  However, if we want to learn more, he have to know what we are looking for. Each statistical category must incorporate a wide range of plays with certain common features, without being so abstract that it fails to exhibit meaningful individual and team tendencies. This is not as simple as it sounds. Luke Winn and Rob Mahoney initiated this debate with a detailed case study and a proposal for ‘new’ statistical categories respectively. Here’s my idea of how defense should be measured, based on their work, my defensive chart of the Olympiakos – Bilbao game and a rant from the coach of the team I support.

What’s not included

Winn’s article put a lot of emphasis on turnover percentage, i.e. the percentage of a player’s defensive possessions which lead to turnovers, and free throw rate, meaning the free throw – to – field goal attempts ratio allowed by a defender. I find both of them to be somewhat impractical. It is not always possible to define a turnover forced by individual effort, as opposed to poor decision making by the ball handler, or effective team defense. Let’s watch some tape:

Both of these turnovers partly result from defensive effort. Kalin Lucas keeps up with Janis Blums off the screen. Aaron Jackson does a great job pressuring the ball. But can the impact of this effort be measured and translated into a forced turnover? How about the help defense of Lazos Papadopoulos or the anticipation of D’Or Fischer? I don’t think there is a definite answer, which is why I prefer a combination of traditional and advanced numbers in order to capture a defender’s contribution in forced turnovers: charges taken (already available via play by play at euroleague.net) and steals on the ball, as proposed by Mahoney – how many steals a defender records against a player handling the ball. These two categories tell us enough about individual ability on the defensive end, without crediting team work and strategy (cutting off passing lanes, being in position to provide help) to a specific player.

Free throw rate, on the other hand, does not take into account most of the fouls committed by a player before his team is in the penalty (i.e. not in the act of shooting) and reveals very little about the player responsible for each pair of free throws: if Milos Teodosic loses his man, Krstitc rotates over to help and Khryapa fouls the opposing center in order to prevent a dunk off the assist, what does free throw rate tell us about this play? Fouls per possesion (Or at least per minute) are a much more useful indicator. Sometimes, the more advanced statistical category is not necessarily the most accurate one.

Hedge rate vs ppp and fpp on switches

After a game against PAOK last December, Olympiakos head coach Dusan Ivkovic publicly blasted his players in what is now considered a classic press conference. Duda provided a lot of memorable quotes that evening: he talked about developing Victor Khryapa and Sergey Monya at CSKA Moscow, alleged that some of his players think about the NBA too much and warned that ‘the moment of truth is approaching’. It was great drama, but something else he said was even more interesting. Ivkovic revealed that his assistants keep track of points conceded in mismatches created by switching on screens – a great example of advanced stats meeting game strategy. Mahoney proposes hedge rate, even though he admits that,

Assessing pick-and-roll defense in statistical form is difficult because of systemic considerations, but quantifying how often — or how far — a player hedges to counter a screen would provide an added level of defensive specificity.

Well, it would certainly help coaches make more sound decisions, but it is doubtful that it would make the game more accessible to fans. First of all, hedge rate indicates not only individual ability, but also a team’s defensive strategy. Kyle Hines showed his prowess as a hedge defender at Bamberg last season, but Ivkovic’s philosophy has kept him in the paint during his first games with Olympiakos. Also, using numbers to decide whether hedging out works or not is impossible without a clear definition of success. Hedge is only a small part of defensive strategy, as it sets off a series of rotations. If the defender manages to stop the ball handler coming off the screen but his teammates give up an open three on the weak side, does hedge rate provide us with meaningful information?

I am not saying that Mahoney, or Winn for that matter, are wrong; without their brilliant work you wouldn’t be reading this article (now you know who to blame). It is just that certain statistical information may be very useful to a coaching staff but complicate matters for fans – and not just casual ones. In other words, we have to agree that we know exactly what we see. This is why I find points and fouls per possession on switches (or mismatches) to be a more universal category.

The formula is simple: points allowed and fouls committed by a defender on every possession where he switches to a player that he has not been assigned to. Keeping track of fouls provides extra information about the efficiency of each defender in mismatches; keeping track of assists allowed in these plays raises once again the less – than – obvious distinction between individual tendencies and team defense. Advanced player stats on offense turn into team numbers without much trouble. This is not the case with defense. Perhaps there is a need for team-only categories: assists or offensive rebounds allowed on switches, ppp on hedge outs, ppp when going into a zone etc. The more detailed the chart of defensive performance, the more stats keep popping up.

Defining steals and deflections

Nonetheless, when certain numbers of each defender are added up, they tell us a lot – only about team tendencies, but the pace of the game as well. Take steals-plus and blocks-plus, meaning steals and blocks which lead to transition points. In the first half of the game I charted, Olympiakos had two steals to Bilbao’s three. However, those three steals by the visitors led to six points and an extra possession, after Lazos committed a flagrant foul. Olympiakos, on the other hand, managed just two points on their steals. In the second half, Bilbao recorded six steals, which produced only four points, the same as the hosts on four pick ups. This is a great way to assess the impact of defense on the other end of the floor and determine which team dictated the pace of the game.

Speaking of steals, Mahoney also proposes keeping track of gambling ratio – how many times a players gives up defensive position for each steal he makes. This is a very interesting concept, but is there a concrete definition of a gamble? Sometimes, there is a very fine line between, say, gambling and going over a screen as part of the game plan. Deflections, on the other hand, is a category we can all agree on – except for Euroleague.net, apparently. Each player should be credited with a deflection every time he tips the ball loose and his team secures possession, either via an opponent throwing it out of bounds or because a teammate got his hands on it. Here’s an example:

Euroleague credited Mavroeidis with the steal, but I think it would be more accurate to give him a deflection and Vasileiadis a steal-plus. Mavroeidis did not secure possession and while Vasileiadis did not have a lot of work to do on this play, the addition of another category in traditional Euroleague boxscores would allow a much better understanding of a player’s defensive performance.

You may disagree with some of my suggestions. You may think that hedge rate is more useful than ppp on switches or that Mavroeidis actually stole the ball. And you could be right. The point of this post is not to convince you about what we should measure. It’s all about starting the conversation. Fans can learn a lot more about basketball by simply looking more closely. If you think that you are up for it, contact us and you could be charting the defense of your favourite Euroleague team. Trust me, there is a lot to be discovered.