Wednesday, February 6, 2013

The Corner Three

Of all the potential shot attempts, a three-pointer from the corner is one of the best shots an offense can get. Across the league, players shoot at a higher percentage from the corner than from other spots around the three-point arc. One obvious reason is that it is the shortest distance three-pointer available. But what about other factors that can affect shooting percentages, such as how the defense reacts to the shot or whether the shooters' feet are set? Using Vantage’s dataset from this season and last season, let’s take a look at how shot defense and shooter movement change from different spots around the three-point arc.

Shot Defense

Because of some teams’ defensive rotations and their willingness to leave certain shooters alone in the corner, it is worth investigating whether the average corner three is met with less defensive resistance than other threes. If corner three point shooters are getting wide open looks while shooters from other locations are getting their shots contested, then that could help explain why players hit corner threes at such a high rate. The corner three might not be that easy, it’s just that defenses are giving up open shots in the corner and guarding the threes around the wings and top of the key.

The chart below shows how often shooters from the different locations see various levels of shot defense.  "Contest+" includes when defenders block the shot, alter the shot, or are within three feet of the shooter with their hand up. "Pressured" indicates when the defender is within three feet of the shooter without a hand up, "Guarded" is in between three and five feet of the shooter, and "Open" is when a defender is outside five feet. 

As the chart shows, defenses are challenging shots in almost the exact same ways from each of the locations. Regardless of location, roughly half of three point attempts are strongly contested, and the other half are split about evenly among the remaining categories. So we know that defenses aren’t forgetting to contest three point shots, even when they are being launched from the corner. 

To further explore the numbers, here are the shooting percentages by location and level of shot defense. 

When defenses are strongly contesting the shot, the corner three loses some of its appeal. Shooters are hitting 33 percent of their strongly contested corner threes, about the same as a three from the wing. However, once the level of defense drops off, the corner three gets much more dangerous. When a defender simply can’t get a hand in the face of a corner shooter, the percentages jump up to 38 percent, and continue rising above 40 percent as the defense gets worse. On threes from the wing and top of the key, the percentages don’t start looking good until the defense gets really lax.

Number of Dribbles

Another reason corner threes may be generally easier than threes from other locations is that a player is highly likely to have his feet set prior to shooting when he is in the corner. Most players are much more comfortable taking jumpers with their feet set, as opposed to off the dribble. The chart below shows how often players are using their dribbles when taking jumpers from the three-point locations. 

Less than five percent of corner threes are shot off the dribble. That number pales in comparison to threes from the top of the key and wing, where 37 percent and 22 percent of threes are taken off the dribble, respectively. Obviously, there is more room to move on the wings and top of the key than in the corner, so we are more likely to see shots off the dribble, with players using ball screens, crossovers, etc.

When we look at the three point percentages by number of dribbles, we can see that when players don’t need to dribble, the corner three is still made at the highest rate (38 percent).

The wing three is close, at 36 percent, but the corner three still reigns supreme in terms of efficiency. Once a shooter starts dribbling, those percentages drop. While shooting off the dribble decreases percentages in any of the locations, nowhere is it more drastic than from the corner. On the rare occasion that a corner three is taken off the dribble, they are only going in 25 percent of the time. Unless it’s Kobe Bryant in his prime, dribbling into the corner is usually a bad idea.

If dribbling is eliminated from the equation and we only look at the difficulty of zero-dribble threes, the chart looks like this:


Even after accounting for shooter movement and the strength of the defense, the corner three still comes out looking like the easiest three. The big difference with the short distance corner three comes when defenders aren’t playing strong defense. If a shot is launched from the wing or top of the key, the defense might be able to get away with not getting a hand up. But if that three-point attempt comes from the corner, anything but great defense can turn an average shooter into a 40+ percent shooter.


  1. Great blog post. From my lengthy response, you can safely assume that it got me thinking about this specific shot/situation/play quite a bit. Apologies in advance for my verbosity.

    The discussion of corner threes has been ongoing for a while. I love the additional spin of putting the defense and feet set to tease out additional insights into shot location. Shane Battier is often cited as being an efficient shooter because of his propensity to shoot from the corner. I’ve always thought that it might be because he’s taking “open” shots but was never sure if that was the case or not.

    From your definitions (Contest+, Pressured, etc.) it’s unclear if that is from the point that the player receives the ball on a pass, the greatest amount of space created off a dribble/screen or if it’s from the point that the player is releasing the ball on a shot attempt. This is a subtle but important difference. A player catching the ball with no one within 10 feet and a defender closes out to within 3 feet at the point he releases the ball on a shot attempt is a very different situation then a player being guarded by someone within 3 feet the entire time he has the ball. It seems that there should be a notion of “defensive closeouts” that takes this data (distance of the closest defensive player at the point of release as well as the point at which the player catches the ball). It would be great to see the conclusions that come from that analysis.

    One of the initial statements you make is that the corner 3 is a great shot because “it is the shortest distance three-pointer available”. Along with the “feet set” analysis, it would be interesting to see the data with regards to “distance from the basket”. I’m curious if there’s a large discrepancy in that data by distance and location. My guess is probably not. The conclusion I’d expect is that a player has just a good of a chance of making a shot from any distance regardless of the location on the court. It just so happens that you get 3 points on some locations that you don’t get in others. I’d love to see a metric built off of distance to the basket, points if made and shot percentage from that distance. Something like points per chance (shooting_percentage_at_distance_in_feet_and_angle_to_basket * point_value_at_that_distance_and_angle). We should be able to tease out the best places to shoot for maximizing point output by charting distance and points per chance for shots with corresponding angles to the corner, wing and top of the key.

    What should a coach or player take from this analysis? Should there be more offensive plays that look to get the ball into the corner? Honestly, I would think not. One shot doesn’t make up a successful offensive trip down the court. I would think there’s a larger percentage of non-scoring opportunities that occur when the ball goes into the corner versus other locations (although I don’t have data to support that claim). The original analysis leaves this point out and ends at a single shot attempt. It would be awesome to see some further exploration of the “plays to the corner” concept to determine what the odds of actually contributing to a scoring opportunity are from the ball being in the corner. A couple data points to think about:
    • Rebounding percentages (offensive rebounds vs. defensive rebounds) for *missed* shots by 3 point location
    o If the defense rebounds corner shots at a significantly greater percentage than other locations, perhaps shooting from the corner isn’t such a good idea.
    • Assists to turnovers that come from the corner versus other locations (as a percentage of the time that the ball is in one location versus another)
    o If the assist to turnover ratio for “in the corner” plays is out of whack from other locations, then maybe we should actually stay away from the corners.

    Thanks for the blog post and making me think about this concept a bit more. I’m looking forward to seeing what additional analysis comes from the Vantage data set and team. Again, good stuff.


    1. Glad you enjoyed the post, Dan. Very thoughtful comments.

      The defensive pressure is at shot release, not prior to the shot. I should have clarified that better.

      As for the rebounding, Jordan Sperber did some research on NCAA rebounding by shot location on Ken Pomeroy's blog and Kirk Goldsberry has done similar analysis for the NBA.

      You pose a lot of interesting questions that I don't have answers for right now, but we can try to address them as we continue to work with the data. I like the idea of looking at corner turnovers, it's a small space on the floor where you have no passing angles and can be easily trapped. I'd guess the A/T ratio is out of whack, but it's worth a look.

    2. Would like to see analysis within players. This would be difficult because of samples, but it is incorrect to assume the population of shooters taking corner 3's is the same as those taking other 3's; I hypothesize that one-dimensional players known for their shooting abilities (i.e. very good shooters) are more likely to end up in the corners for floor spacing, which results in the corner 3 having a higher field goal percentage.

  2. Thanks for your great post.You share your post various graph and three corner.
    point to point