College Basketball: How the DSS is created and some early SEC Results

The last time I wrote a DSS post I asked if people would find it useful to see a more in-depth description of the process and the response was in favor of such an explanation. So this will be mostly dedicated to describing how I take the action on the court and translate it into the defensive results posted in the DSS. I have the results for UK's first 4 SEC games to offer by way of example and they should be of interest even if you don't want to read the rest of the piece.

The concept of the Defensive Score Sheet began with Dean Oliver and his book Basketball On Paper. Oliver at the time was working in WNBA and organized an effort to provide a more complete description of the defensive results from a game.

In a regular box score you can see who attempted every shot, every free throw, grabbed every rebound, and committed every turnover. The other side of those plays is incomplete at the individual level however - the players who forced misses are only partially available through blocks and the players who force turnovers is only partially available through steals. The players who committed the fouls that led to free throws is not available in the box score, but can be found in the play-by-play record of the game. beyond those problems, NO information is recorded regarding who (if anyone) allowed made field goals.

The goal of the DSS is to fill in this missing information.

Plays Recorded

There are 4 events that I determine which players were engaged defensively: free throws, turnovers, missed field goals, and made field goals. For the latter three, the defender can be a single player, multiple players, or no players if the shot was undefended or the turnover unforced. When multiple players are involved the event is evenly distributed to all of them. So for example, if 2 players challenge a shot (such as in the above picture) and it misses, they each get 0.5 of a forced miss.

1. Free Throws: These are easy as the player who committed the foul is listed in the play-by-play record.
   
   
   
2. Forced Turnovers: These are usually pretty straightforward to determine as it is typically easy to determine who applied the pressure or took the action which resulted in the turnover occurring. In order for a player to be credited with forcing a turnover they have to do something more active than simply be in the right place at the right time.
   
   
   
3. Forced Missed Field Goal: These are also typically straightforward. I determine this by which player(s) were actively guarding the shot when it was taken - by getting a hand up, forcing the shooter to move to get the shot away, or some other action that hindered the shooter.
   
   
4. Made Field Goal: These are sometimes difficult to determine exactly which players were engaged defensively on the shot. I have adopted the rule of "Who did the scorer have to beat in order to score" in order to determine which players should be legitimately charged with a made field goal allowed.

It is not always obvious which players should be considered engaged defensively on a given play. For example, if Joe Pointguard blows past Donny Defender to get to the rim and score, is Donny a defender on the shot even though he wasn't actually at the basket when it was attempted? I think he was and so he gets charged with a made field goal allowed.

What if Joe, after getting past Donny also drives past Sammy Smallforward who is guarding his own man - is Sammy a "defender" on this play since he was in physical proximity to Joe? In this case I say no, unless Sammy takes some substantial action that Joe has to avoid, such as trying to draw a charge or trying to block the shot.

This gets to a subtle point of what the DSS is trying to do: as I fill in the blanks I am *not* trying to determine if a particular player did the "right thing" or the "wrong thing" on a play. I am only trying to determine "who was engaged defensively when the play occurred." So I am not trying to judge what Sammy "should" have done - only the coach can do that - but only what he "did" do. This is similar to how the regular box score will not tell you if a player took a "good" shot or if the offensive set was run properly - only whether the shot went in or not.

Having said that, there are places where I have to make a judgment call that sometimes drift into what "should" happen.

Some Specific Situations

• Layups: As mentioned already, when a defender gets beat for a layup, he is charged with a made field goal even if he wasn't physically present at the basket when the shot was made.
• Offensive putbacks: To the best of my ability I try to figure out who had the block-out responsibility on the player who attempted the putback. This is usually the initial defender when the play started. If this can't be easily figured out - for example if the initial defender comes over to give help defense - then the outcome is listed as an "Undefended" play.
• Fast breaks: One of the principle rules for defending a fast break is to first stop the ball. So in a 2-on-1 or 3-on-2 type situation, if the defenders force the initial ball handler to give up the ball then they have done their job and a made basket will be listed as an "Undefended" play.
• Closing out on a 3 point shooter: This is by far the most difficult play to determine - it is hard to know if a defender was able to close out quickly enough to actually affect a shot attempt or not. In general I tend to attribute a forced miss when the defender is taller/longer than the shooter. Attributing made shots is harder, but I typically (though not always) list them as "Undefended" under the assumption that playing off a shooter is part of the overall defensive plan rather than a lapse by the defender.

As you can tell, there is quite a bit of subjectivity to this exercise. I think that most people would agree on roughly 85% of the defensive plays that occur in a game though. Back in November when David Hess and I each charted the UK-Kansas game, we arrived at very similar results.

Defense Box Score

After watching the game and filling in the missing information, I create a defense version of the box score. Here is one for UK's game against Arkansas.

Each column corresponds to its offensive equivalent. Using Marquis Teague's line as an example:

• Teague defended on 9.84 of Arkansas's 57 field goal attempts. The fractional part comes from when Teague and other UK players defended the same shot attempt. Of these 9.84 shot attempts, 2.34 made baskets were scored against Teague.
• Of the 9.84 attempts, four were 3 point shots and no Arkansas player was able to hit a long distance shot over Teague.
• Teague committed 2 fouls that resulted in 2 free throws, both of which were made.
• Teague grabbed 4 defensive rebounds, didn't block any shots, and had 3 steals. Overall he forced 3 turnovers.
• Teague's steal count doesn't match up with the official box score because he had a clear steal that was somehow attributed to Doron Lamb.
• Overall, Arkansas players scored 6.68 points against Teague.

Defensive Ratings

From the defensive box score I calculate some defensive metrics that help put some context around an individual player's defensive results. Below are the composite results for UK's first 4 SEC games.

• The first 3 columns are based on the 4 Factors. Staying with Marquis Teague, opponents are shooting 26.3 eFG% against him, he is forcing turnovers on 24.0% of his individual defensive possessions, and sends people to the line at a rate of 29.5 times for every 100 field goals they attempt.
• Overall, Teague's plays have contributed to 17.2 stops and 7.7 scores allowed. Fractional values reflect that credit for stops is usually distributed among several players - one forces a miss and another grabs the rebound for instance.
• Plays are just the sum of scores and stops, this represents the number of defensive possessions an individual was directly involved in.
• Stop% is the percentage of time that an individual prevents the opponent from scoring any points on a possession. Teague has prevented his man from scoring and gotten the ball back for UK on 69% of his plays in SEC games.
• Defensive Possession % (DPOSS%) is the percentage of the team's defensive possessions that a player was engaged in while he was on the floor. Teague is directly engaged on 11% of the defensive possessions the team faces when he is on the floor.
  
• Defensive Rating (Def Rtg) takes an individual's total defensive effort and places it in context of what their teammates did on average. You can think of it like this: If Marquis Teague is on the floor with 4 players who are playing "UK-average" defense, this is what the overall Team Defensive Efficiency would look like if Teague was involved on 11% of all defensive possessions and stopped his man 69% of the time.
• Opponent Efficiency: Unlike Def Rtg, this is a purely individual rating. It is simply the points allowed by the player divided by the number of defensive possessions they are engaged in, so it is like the Team Efficiency found at KenPom or StatSheet.

If you made it through all that then good for you! There are number of other people taking an interest in charting defense this way and I've included links below for some who are doing so.

• David Hess at TeamRankings.com
• Luke Winn at Sports Illustrated
• Chris Mackinder at Deuce 2 Sports (Michigan St blog)