NBA DvP Betting Guide: Defence vs Position
Available in US
Why DvP Is the Most Mis-Used Stat in Prop Betting
DvP — defence versus position — is the stat most often quoted incorrectly by people who think they have an edge on NBA props. I have lost count of how many times I have seen a tipster post “rank 27 in DvP” as if that alone justifies a play. Rank 27 in DvP over 50 games might mean a real defensive weakness. It might mean a team that played five high-scoring opponents in a row and has not corrected for matchup. It might mean nothing at all.
The stat itself is useful. The way it is commonly used is not. A defensive ranking against a specific position aggregates outcomes from a wide range of opponents, schemes, and game states. Without filtering for scheme similarity and matchup quality, DvP is a noisy proxy for a player’s expected production against a specific team.
What DvP Actually Measures
DvP is the average production a team allows to players at a specific position. The most common formulation: points allowed per game to opposing point guards, shooting guards, small forwards, power forwards, and centres. Some versions extend to PRA-style combined production. Others track specific stats like three-pointers made or rebounds allowed.
The position label is the first problem. Modern NBA rosters routinely play players out of nominal position. A 6’7″ creator listed as a point guard plays many minutes at shooting guard. A traditional power forward is increasingly a centre in small-ball lineups. DvP attribution by listed position can mean a centre’s production gets bucketed under power forward DvP depending on how the data provider classifies him.
The cleaner approach is to look at production by physical archetype. A primary creator facing a team that struggles against primary creators — regardless of how the data labels him — is the matchup that matters. The label is administrative; the matchup is real.
The Sample-Size Problem with DvP
A team has played 20 games against opposing point guards in a typical mid-season cross-section. That sample is small enough that two or three outlier games can move the team’s DvP ranking by 10 spots. A team that gave up 35 to a star point guard in one matchup, when the team was without its primary defender, can look like a permanent weakness at the position when really it was a one-game anomaly.
Filtering helps. I exclude games where a team’s primary defender at the relevant position was out. I exclude games against the league’s top three or four players at the position — those games skew the average upward without telling me anything about the team’s defensive scheme.
What remains is a smaller sample, perhaps 12-15 games. Within that sample, the variance is still substantial. A 90% confidence interval on a team’s DvP after 12 games can span 4-5 points of production. The point estimate at rank 27 might overlap meaningfully with the point estimate at rank 12. The ranking is a lie of precision.
Switch-Heavy Defences and the DvP Lie
A switch-heavy defence is the classic example of DvP failing as a model input. The numbers say the team gives up X to opposing point guards. The reality is the team rarely lets a point guard run pick-and-rolls against his nominal defender; the switch happens, and the production gets reattributed to whoever ends up guarding the ball.
If you bet a points over on a primary creator against a switch-heavy team because the season DvP looks generous, you are betting against the scheme. The scheme exists precisely to suppress the production that DvP would predict. The closing line on these matchups often moves later in the week as sharper money reads the scheme, but mid-week openers can sit at lazy DvP-based prices.
The opposite pattern is also real: a switch-heavy team facing a player whose game thrives against switches — a downhill driver, a post player, an isolation specialist — has its DvP overrated. The scheme that suppresses pick-and-rolls invites isolation, and isolation specialists feast against it.
Isolating the Real Matchup, Possession by Possession
The next level beyond DvP is possession-level matchup data. Tools that track which defenders guarded which players for how many possessions tell you the actual matchup outcomes, not the team-average. A primary creator’s points per possession when guarded by a specific defender, averaged over the last 200 possessions of that matchup, is far more predictive than the team’s season DvP.
Most retail bettors do not have access to possession-level data without paying for it. The alternative is to watch enough of the team’s recent games to know which defender is likely to draw the assignment and how that defender has fared in similar matchups. It is slower than scanning DvP, but the conclusions are more reliable.
For the bettor who wants to combine matchup reads with broader strategic frameworks, see my walkthrough of usage rate prop betting – the matchup question feeds directly into the usage question, and the two together produce better projections than either alone.
DvP Applied to Rebound and Assist Lines
DvP for rebounds and assists is more useful than DvP for points, because the sample sizes are larger and the matchup effects are sharper.
On rebounds, DvP tracks the rebounds a team allows to opposing players at each position. A team that fails to clear its defensive glass — say their starting power forward is a poor rebounder — gives up extra rebounds to the opposing power forward and centre. This is more deterministic than scoring DvP because rebounding is a high-event stat and team-level patterns persist.
On assists, DvP measures the playmaking allowed against teams. A team with a poor on-ball defender at the point of attack lets opposing primary creators rack up assists, because they get into the lane easily and create kickouts. The 12-possession differential between fast and slow games matters here too — assist DvP scales linearly with pace, and a slow-pace team can look poor at assists DvP without actually being poor defensively.
The cleanest use of DvP for rebounds and assists is in conjunction with pace and minutes projections. The combination of pace lift, favourable opponent DvP at the right position, and stable minutes produces the strongest single-bet edge I have logged repeatedly.
One more wrinkle worth raising: DvP for three-pointers allowed by position is its own animal. A team that gives up lots of threes to opposing shooting guards is often a team with a slow closeout from a specific defender — a one-player problem rather than a scheme problem. The bet that follows is on a sharp-shooting opposing shooting guard specifically, not on any wing. Generic “team gives up threes” reads miss the player-specific texture that drives the production.
Articles
Written by the editors at HoopMargin.
