Loading ...
Sorry, an error occurred while loading the content.

Recording Indiv Defensive Stats

Expand Messages
  • Dean Oliver
    One of my big goals this year has been to track individual defense a bit more. I ve particularly done this for my local Warriors, who don t play much defense
    Message 1 of 1 , Feb 15, 2001
      One of my big goals this year has been to track individual defense a bit
      more. I've particularly done this for my local Warriors, who don't play
      much defense unfortunately. However, last night, I took the chance that
      TNT wouldn't break away from the Toronto-Miami game and scored the
      defense. The scoresheet from that is attached. Below is an explanation
      (written previously) to explain the base defensive method I use

      Individual defense is much more difficult than offense to evaluate.
      Conceptually, it is more difficult to decide _what_ to evaluate. What we
      watch in games is how one guy gets beat and we see what he doesn't do -
      stop his man. Or we see how a guy denies a scorer a good look. Neither
      of these cases adds to the defender's stat book. All that we have to
      evaluate individual defenders in official stats are

      1. steals
      2. blocks
      3. defensive rebounds
      4. team defensive ratings

      This really is insufficient. Within the framework I will outline below,
      it would be very nice to have the following stats also recorded:

      5. forced missed field goals (non-blocks)
      6. forced turnovers (non-steals)
      7. forced free throw misses
      8. field goals allowed
      9. free throw possessions allowed (or just free throws allowed)

      I have taken these stats before, using the Excel sheet I posted under the
      Files link. I definitely split up some of these stats into halves, like
      they split up sacks in football. Two players may be responsible for a
      forced turnover, so I give them both half. Sometimes, like on a fast
      break, no one player is responsible for the field goal allowed and I give
      it to the Team (which ultimately gets spread evenly among players).

      So what to do with all this information? Eventually, an estimate of each
      individual's number of points allowed per 100 possessions could be
      appropriately compared with his individual offensive rating. This
      individual defensive rating is what we're shooting for.

      The way to get this defensive rating is turn your thought of defense
      around. We typically frame defense in terms of how many points a defense
      _allows_. Now think of defenders actively stopping the offense. A steal
      stops the offense. A block plus a defensive rebound stops the offense.
      We can potentially count these "defensive stops" because they are active,
      not passive. If we can count the stops and estimate the possessions faced
      by a player, the difference between the two is the number of scoring
      possessions they allowed, which essentially gets us to points allowed.

      The team stops are easy:

      1. A turnover
      2. A missed field goal rebounded by the defense
      3. Missed free throws rebounded by the defense

      Distributing credit for them among teammates requires some theory.

      First of all, a turnover is pretty easily split. A steal is a stop for
      the individual who has the steal. Often, a steal is created by more than
      one person, but only one person gets the credit; that's a flaw in how the
      NBA records the stat, but accepting that flaw makes this equivalence easy.
      If we record forced turnovers (and split them into halves as necessary),
      they are also equivalent to stops.

      Stops associated with missed field goals require some distribution of
      credit. Namely, both the forced miss and the defensive rebound are
      required for the stop. As a (valid) first cut, it is fine to give 50%
      credit to the forced miss and 50% credit to the defensive rebound. But in
      the NBA, it's relatively easy to get a defensive rebound compared to
      forcing a miss. The defense rebounds the basketball roughly 70% of the
      time and forces a miss roughly 50% of the time. The player doing the
      harder job - forcing the miss - should get more credit for the defensive
      stop. The method for generating the weight on a forced miss is

      Weight on forced miss = w_FM = DFG%*(1 - OR%)/[DFG%*(1 - OR%) + (1 -

      where OR% is the offensive rebounding percentage of the opponents and DFG%
      is the field goal percentage allowed by the defense. The weight on a
      defensive rebound, w_DR, is then 1 minus the weight on the forced miss.

      These weights apply when there really is a stop, so the stop formula
      itself must account for when a miss is forced but the defense doesn't get
      the board (you can't give credit for a forced miss as a stop unless the
      stop is completed with the defensive board). The part of the formula for
      stops that accounts for missed field goals is then


      where FM is forced misses, FM_TM are team forced misses (not credited to
      any individual), OppOR is opponent's offensive rebounds, MIN are a
      player's total minutes played, TMMIN are the team's total minutes played,
      DR is an individual's defensive rebounds, and BLK is an individual's block

      The final piece is forced missed free throws. Since these are rare, and
      offensive rebounds off of missed free throws are also rare, forced missed
      free throws are taken for simplicity to be a full defensive stop.
      Theoretically, they should be weighted slightly less and credit given to
      defensive rebounds, but the complication isn't worth the accuracy.

      Then the full formula for calculating an individual's defensive stops is

      Stops = FM*w_FM + BLK*w_FM + DR*w_DR + FM_TM*MIN/TMMIN*w_FM -

      where STL is an individual's steals, FTO represents the player's forced
      turnovers, FTO_TM represents the team's forced turnovers not credited to
      any individuals, and MFTA are forced free throw misses. (As Bob said,
      more accurate formulas are necessary and these may not be simple. This
      formula ain't simple.)

      I'll get to more documentation of the actual defensive rating soon and to
      all the approximations available when the recorded stats aren't. For the
      next few days, though, I'll be traveling. Take a look at the attached
      scoresheet and the following recorded stats

      Forced Forced FTA
      Plyr Team Miss Blk TO Stl Miss FGM FTM
      Oakley TOR 4 2 2 3 0 5.5 3
      Petersn TOR 2 0 0 0 0.5 1 1.5
      Davis TOR 7.5 6 0 0 0.5 7.5 1.5
      Carter TOR 4.5 0 0 2 0 4.5 0.5
      Jackson TOR 6 0 0 4 0 5.5 0
      Clark TOR 0.5 0 0 0 0 3 2
      Willims TOR 7 1 0 4 1.5 4 1
      Curry TOR 3 0 0 0 0 2 0
      Murray TOR 1 0 0 0 0 0 0
      Team TOR 2.5 0 3 0 0 3 0.5
      Bowen MIA 6 0 1.5 0 1 7 1.5
      Mason MIA 8 1 2 0 0.5 3 0.5
      Grant MIA 10.5 0 1.5 0 0 4.5 0
      Hardawy MIA 4.5 0 0 0 0 2.5 0
      Jones MIA 3.5 1 2 0 0.5 3 1.5
      Green MIA 5 2 0 0 0.5 3 0.5
      Majerle MIA 3.5 3 0 1 0 2.5 0.5
      Carter MIA 2 0 1 1 0 2.5 0.5
      Team MIA 1 0 1 0 0 5 0.5

      A few things to note in the calculated stats

      1. Carter was not a horrible defender in the game. I actually have seen
      little to indicate that he is horrible. He does seem to get lost a bit
      and think that others will rotate to help him. I had his defensive rating
      at 94.0, about equivalent to the team rating. He did have very few stops,
      something Jordan always had a lot of. He wasn't guarding a big scorer the
      whole night, rotating between Jones and Bowen.
      2. The big men tended to be involved in a lot of plays. They both stopped
      a lot of possessions and allowed a lot. Both Antonio Davis and Brian
      Grant seemed like they were everywhere when I scored this game. Davis had
      6 blocks. But I also recorded them as giving up some scores.
      3. Oakley didn't look very good as a man defender. In a lot of one-on-one
      situations, he seemed to fail, giving up wide open jumpers and allowing
      people to shoot over him But he makes up for this with good help in traps
      and by dominating the defensive boards.
    Your message has been successfully submitted and would be delivered to recipients shortly.