|
The analytics boom in the NBA has predominantly been focused on the offensive end and it makes sense as these stats are easy to track and have been readily available for years. You can physically see who shot a ball, who passed the ball, who turned the ball over and we can see a tangible outcome. ie The player scored, got an assist or turned it over. We can account for what a player does on the offensive end pretty well, as it’s easy to measure. It has spawned the majority of the advanced stats and ways in which we evaluate players and teams today. What has always lagged behind are the defensive stats. These really don’t show up in the box score, the best we have publically to evaluate an individual player or 5 man line-up are variations of +/- stats. +/- tries to account for everything that is done on the defensive end, spitting out a number that shows how many points a player or line-up is impacting on the defensive end. These +/- stats are fairly useful for sure but there are certainly plenty of holes in the data and gaps in what it captures. So why are defensive stats so hard to attain? Well how do you measure defensive instincts and positioning accurately? Yes we have tracking data that shows the on court movements of players, but the tracking data doesn’t know where Marcus Smart’s hands are on a defensive close out for example. It also doesn’t know when Rudy Gobert is reading the screening actions happening on both strong and weak sides, and then anticipating when to help on defense or when to stick closer to his man. Gobert being an elite defender would have half a dozen things going through his mind at once when defending. How do you account for that with a stat? These instincts are very tough to measure, yes we can measure every pick and roll that Gobert is involved in and yes we can measure every loose ball Danny Green dives on and these certainly all help, but there are so many intangibles on defense that make it very hard to measure everything that is going on. There is no doubt that the highly efficient offences we are seeing today (Houston, Golden State) have been largely built around the analytics boom on the offensive end. Even the casual fan can see the offensive juggernauts at the top end of the league are leaving defenses in their wake. The gap between a great offense and a great defense seems larger than ever. Defensive analytics have been left behind and there is no doubt in my mind it’s the biggest driver between the offensive and defensive gulf we see today. In this post I start diving into how teams can get more efficient defensively through the use of situational fouling and hopefully start bridging the gap with elite offenses. Fouling? Seems counter intuitive? What I will demonstrate below is how teams can control the shot quality of offenses through a certain type of foul and what benefit it can have. Let’s first try an answer the question, do teams who foul less often have a better defensive rating? 
The graph tells us there is some correlation (slightly linear) between low fouls and a good defensive rating, however it’s certainly not the be all and end all. This makes sense as a good defense relies on what are known as the “4 factors” as created by Dean Oliver (Can be seen in the Book of Basketball). These being limiting an offenses eFG%, defensive rebounding %, limiting the amount of free throws you give up and causing turnovers. Fouling mainly impacts how many times you send a team to the free throw line but it also factors into the eFG% of an offense, as by fouling we can dictate the quality of shot a team gets (more on this later). How much weight does each factor hold on the defensive end? The 2 factors where fouling contributes are eFG% (46% weight) and FT Rate (7% weight), so by controlling how and when we foul we can contribute to factors that account for over 50% of what impacts a team’s defense. A few findings from the graph:
The Jazz are 4th on the defensive glass, 5th in causing turnovers, 7th in opposition eFG%, in summary they are a good all round defensive team. The 76ers rank highly on defense as they are 2nd in defensive eFG%, yet don’t rank in the top 10 in the other 3 factors, they really rely heavily on limiting the opponent’s eFG%. The Raptors are 4th in opposition eFG%, 11th in causing turnovers and don’t rank highly in the other 2 factors. Again they rely heavily on limiting the opponent’s eFG%. The Celtics are 1st in opponent eFG%, yet don’t rank highly in the other 3 factors. Like the 76ers and Raptors, the focus defensively for the Celtics is limiting the opponents eFG%. The Spurs are 6th in opponent eFG%, 2nd in opposition FT rate and 7th in rebounding. Similar to the Jazz, the Spurs are a defensive all-rounder with a high focus on keeping teams off the foul line. We could stop there and say, well the Raptors and 76ers seem to be doing just fine by fouling a lot? But let’s dig a little deeper and bring in a fouling scenario that many teams utilize today already, but we can make the whole process a lot more efficient and improve teams defensively. Take a look at this video (Keep an eye out for the Jazz turning the ball over and Joe Ingles committing a quick foul at half court) You may or may not recognize the scenario, a team turns the ball over through a lost ball or bad pass and then commits a quick foul to stop the fast break. This has crept into the NBA more and more, with Jeff Van Gundy questioning the strategy on air this week. Van Gundy’s argument is if you commit a turn-over, your team should be doing all it can to hustle back on defense rather than giving up a cheap foul. Let’s see if he’s right. The data I have pulled from play by play logs is every time a team has turned the ball over through a lost ball or bad pass and then commits a foul within 3 seconds of the turnover (a quick foul to stop the break) First, here is which teams are using it more than others: 
On average, teams are utilizing the quick foul 5% of the time when they commit a turnover with the Jazz far and away utilizing this the most at 11%. Comparing the Jazz (who commit fouls at a league average rate) they have nearly double the examples as the Spurs (lowest fouling team in the league). There are then teams like Charlotte, Indiana and OKC who are at the opposite end of the spectrum and hardly ever commit this quick foul, therefore taking Van Gundy’s side of the argument. So what is the better strategy? Well it depends :-) When a team turns the ball over like this, the opposition on average scores at a rate of 0.92 points per possession (ppp) on the fast break. Obviously there are times when it’s a wide open fast break and it’s almost a guaranteed 2 points vs times when 3-4 defenders can get back on defense and potentially get a stop. On average though, teams score 0.92 ppp. When a team chooses to commit the early foul, the very next play (opposition will get a sideline inbound) teams are averaging 0.67 ppp. Some very basic math would suggest 0.92-0.67 = 0.25 points advantage is committing a foul vs trying to get back on defense. That’s a very basic approach and assumes everything else in the game remains unchanged. In reality, the game does change when you foul. By committing just 1 extra foul on average, you are nearly always likely to be sending a team to the free throw line later in the game. (You will be in the bonus, which means 4 or more fouls in a quarter have been committed) I have included a table later on which shows nearly every team on average by adding an extra foul in any quarter would in most cases give up 2 extra free throws later on. Let’s take the scenario again, where a team chooses to commit on average 1 extra foul per game to stop a fast break after a turnover, but it means there will be a play later on where if they foul they give up 2 free throws that we wouldn’t have previously. In this scenario teams are actually .58 ppp worse off, on average. If however they chose to foul only in a situation where they were giving up a guaranteed 2 points (wide open layup, as they don’t have anyone back on defense) then they would actually be .5 ppp better off. So this is where “it depends” comes into play for the majority of the league, and also a couple of other factors a team needs to take into account if they choose to foul. We have established:
Does it make sense if Kevin Durant is on 2 fouls in the first quarter and there is a situation where he can stop a guaranteed fast break but pick up a 3rd foul? That’s for a team to decide, but the impact of having Kevin Durant on court rather than on the bench in foul trouble, seems like a pretty easy decision not to foul. If it’s a rotational player who might get in foul trouble, then it’s a risk the team should be probably be happy to make. The last factor being, is the team already in the bonus? If you are already in the bonus, then by fouling you are giving up an automatic 2 free throws. Again it’s situational, if a player believes he is going to give up a wide open 2 points, then sending the player to the free throw line is the higher percentage play. As you can see, there are a lot of decision a team needs to make. Based on what I have found, the decision process and easy rules for a team & players to follow, should look like this: 
Based on the above, Van Gundy is right in saying not to foul where you turn the ball over below the foul line. You should trust your players to get back on defense and make it a contest which is better than fouling. When turning the ball over below the foul line, your defense has a greater chance of getting back. However, if the turnover is above the foul line then it makes sense to commit the foul as your defenders are highly likely to be out of position, thus leading to an easy layup for the opposition. Teams will have more accurate tracking data on this and may determine that the top of the key or 3 point line is where you make a decision to foul or not to foul, but for my example I have used the free throw line as the decision point for when a turnover is made. There is 1 more scenario where we can get even more benefits and it ties back into teams that don’t foul as much (The Spurs). Below is a table of fouls on average per quarter. Keeping in mind when teams commit a 5th foul they are sending the opposition to the foul line for 2 bonus free throws. 
What you can see is, the Spurs in 1st quarters are only averaging 3 fouls. In this quick foul scenario off a turnover, the Spurs could foul 1 extra time on average and have no impact in terms of giving away extra free throws later in the quarter. They have a foul up their sleeve. This means, they could foul in any situation where they turn the ball over through a bad pass or loose ball and be at minimum .83 points per game better off. If they happened to add that extra foul only in situations where they were giving up a guaranteed 2 points, then they would be 1.33 points per game better off. Doing this for the whole season would have them leap frog the Celtics to be the best defensive team in the league. This is where the power of being a low foul team can start opening up many different situations, where a team can start controlling a game and shot quality of the opposition through situational fouling. Right now you can see the Spurs can really only utilize this in the 1st quarter of games if nothing else changed defensively for them. If a team like the Spurs can become even more disciplined on defense and cut out further silly fouls, it will open many more opportunities and scenarios where it makes sense to commit a foul. Putting the opposition in a position where they are less likely to score and start dictating what shots the offense gets. So when I said the 76ers and Raptors look pretty good defensively despite committing a lot of fouls, Is there an opportunity for them to now get even better? We need to keep in mind that fouling also represents a form of aggressiveness that maybe works for teams defensively in the long run, and if the 76ers and Raptors remove some of this aggressiveness it may have more adverse effects. For example they might not generate as many turnovers if they are less aggressive. However, there certainly seems to be some opportunity here for cutting out as many silly fouls as possible and committing situational fouls in their place. A team like the Spurs who are great defensively already by not fouling, can they take it to another level? I have already found an opportunity for them to save between .83 and 1.3 points every game, just by looking at 1 example of situational fouling. Is it a way to perhaps contain the highly efficient Rockets and Warriors offense by employing a full blown situational fouling strategy? To pull this off, teams are obviously going to need to be highly disciplined, fundamentally sound and have defenders with High IQ’s but there certainly seems to be some opportunity here to bridge the gap between offense and defense.
0 Comments
Previously I took my machine learning algorithm and found line-up specific insights for the Dallas Mavericks.
Continuing on with the machine learning theme of the last few posts, where previously I had looked at how a high level scouting report can be created for teams. I now wanted to get into some nitty-grity detail and see what insights machine learning can discover in terms of specific 5 man line-ups.
My last few posts have been based around producing a scouting report from a machine learning algorithm I have put together.
Where is the future for NBA analytics going? The buzz words you will hear more and more over the next few years will be “artificial intelligence”, “machine learning” and “deep learning”.
This is one of the hardest measures in basketball, how good or bad of a job is a head coach doing?
NBA Summer League is pretty early to call “bust” isn’t it? Yeah it sounds crazy and it might be rash, but hang in there with me.
For more than 25% of NBA teams, the first offensive set of the game looks to be predictable.
We have hit a milestone in the NBA schedule as we move through Christmas. So what better time to check in and analyze which players are hurting their teams, with poor on court decisions and simple skill errors which I tally up in my Coach Killer metric.
“Coach Killer” as defined by urbandictionary.com – "Any dumb or accidental action that causes an enormous let down, especially if it occurs at a critical time ".
The humble on-ball screen has been around since the beginning of basketball. For decades the formula has been simple, send one of your team's big guys to set an on-ball screen for your guard and that in itself will create a heap of options offensively. Whether it be allowing the guard to drive to the basket, the big guy to roll to the basket or the big popping to the perimeter to shoot a 3 to name a few.
The 6 point swing. One of the biggest perceived momentum changing plays in basketball, where a team misses a 3 point shot and the opposing team drills a 3 pointer on the very next play, therefore causing a 6 point swing...
The latest reports have Stephen Curry coming back from injury sometime in the 2nd round series vs the Portland Trailblazers. The Blazers will be looking to steal early wins in the series whilst Curry sits, but how are they going to do it? And what adjustments are the Warriors making without Curry?
Here is the scenario:
Well maybe not totally forget about them, but regardless of what schemes a team tries to implement to stop the duo they are going to score whether the Thunder win or lose. More about that later.
Limit elbow touches, paint touches, post touches and passes out of the post. Simple right?
Make him do a left hand layup. That sounds pretty basic and logical right? Basketball 101 stuff? It is that basic.
Again.
|
RSS Feed
