Rim protection explorer
Blend the gates (your own weights)
This makes a number called "Your blend" that you can put on either axis. It is whatever you weight it to be. Players are not ranked by an all-in-one total, and the reasons are at the bottom.
Projected best defenders,
Rank by
Key
RedirectQuality of the shots opponents take away from the rim while he is on the floor. Points per 100 possessions, higher is better.
DeterHow rarely opponents take shots at the rim against him. Same units.
ContestHow much less often rim shots go in than that shot normally would. Same units.
Rim scoreDeter plus Contest. Blocks are not in it. Redirect is left out because it mostly mirrors Deter (correlation -0.69).
BackstopContext only, nothing is subtracted. Rates the best shot-blocker playing alongside him by block rate: strong, some or little.
CarryHis rim score counting only the minutes when no other top-fifth rim protector is on the floor with him. It asks how much of the rim defense is his own. It tracks the rim score closely (0.93), does not forecast next season any better, and is not steadier for players who change teams, so treat it as a description. Shown for players with 400 or more of those minutes.
Event gapHow many blocks, steals and charges drawn per 36 minutes he gets compared with what his total defense would predict. Zero means his defense accounts for them, positive means more than it explains, negative means fewer. It is a style description: players with a high gap did not defend worse the next season (they were slightly better), so it does not show that chasing blocks and steals costs a team anything.
DeterrerStrong at Deter only. He changes which shots the offense chooses to take (Valančiūnas, Green).
ContesterStrong at Contest only. The shots at the rim go in less often (Gafford, McDaniels).
RedirectorStrong at Redirect only. The offense is pushed into worse shots away from the rim (Suggs, Walter).
Two gatesStrong at two gates (or, for one player, all three), usually Contest and Deter (Wembanyama, Clingan) or Contest and Redirect (Gilgeous-Alexander).
BalancedNot strong or weak at any one gate. Most players land here.
LeakyNo strong gate and a clearly below-average total (Missi, Portis).
"Strong" means roughly the top fifth of the players shown on that gate.
The table
Click a header to sort, click a row to see that player. "Rim rank range" is where the player landed in 90% of resamples of the data. A wide range means the number is noisy, and almost everyone's is wider than you would guess.
What this does not do
- It does not pick the best defender. It is about rim protection. Perimeter defense, off-ball work, communication and effort are mostly outside what these three gates capture.
- There is no all-in-one total. Adding the three gates together did not hold up in pre-set tests. Guards who play next to elite shot-blockers made up too much of the top of the list, and players' totals were not stable when they changed teams. The three gates are shown separately and the only ranking is the rim score.
- Teammates matter. A guard whose team has a great rim protector behind him can look better on Redirect than he would alone. An adjustment was tested and did not move the scores in a way that holds up, so it is shown as context instead. The "Backstop" line on each player says how good the best shot-blocker playing alongside him was.
- One season is noisy. Rim score repeats from one season to the next at about 0.43 among rim protectors, and 0.28 for the ones who changed teams. That is why ranges are shown, not ranks. It does predict next season's team rim defense a little better than just using last year's team (0.60 against 0.57).
- Carry and Event gap are descriptions, not grades. Carry did not forecast next season better than the rim score or hold up better for players who change teams. Event gap is steady from year to year (0.72), but players with a high gap did not defend worse the next season, so it does not show that chasing blocks and steals hurts. Neither is used in the projection.
- The top is a cluster. The players at the top of the rim score list are separated by less than the uncertainty. The rank range column shows how much each player's rank moves when the data is resampled, so read that, not the rank.