Defence vs Position for NBA Props: A Practical Reading Guide to DvP
What DvP Tries to Capture and Where It Lies
I’ll start with a confession. For the first three or four years I bet props, I treated DvP rankings the way some people treat horoscopes — a little superstition, a little confirmation when things went my way. Then a quant friend of mine asked me what the metric actually measured. I started to answer, paused, and realised I didn’t really know. I knew it was “defence by position”. That was the whole answer.
DvP, short for Defence vs Position, attempts to summarise how a team’s defence performs against each lineup slot. The number you usually see on a fantasy or props site is points allowed per game to opposing point guards, shooting guards, small forwards, power forwards and centres — averaged across the season, sometimes weighted toward the last 10 or 15 games. It tries to give you a one-glance read on whether a centre is walking into a friendly matchup or a hostile one.
What it captures honestly is positional outcomes. What it captures sloppily is everything that produced those outcomes — schedule, opponent star quality, garbage-time minutes, blowout context. A team that’s faced four star centres in a row will have a “weak vs C” DvP that has more to do with the schedule than with their actual interior defence. The metric is the average of the question, not the answer.
How DvP Tables Are Constructed
Most DvP tables on free fantasy sites are built the same way under the hood. Take every player who logged at least one minute at a given position — usually defined by the official lineup data, sometimes by the position the player is listed at on the roster — and sum the relevant stat (points, rebounds, assists, threes) across the season. Divide by games played by that team’s opponents, normalise to per-36 minutes or per-game depending on the publisher, and you have a number.
The first issue, which everyone notices but few correct for, is that “points allowed to PGs” is a fiction in modern basketball. A player listed as a point guard might spend 15 of his 32 minutes guarded by a wing while running off-ball. The label says PG. The reality is something else. The DvP table ignores the reality and aggregates by label, which means a team with a defensive specialist guard but a soft wing rotation can show as bad against PGs, when actually their PG defence is fine and the wings are getting eaten alive on switches.
The second issue is the rolling window. A 15-game rolling DvP can swing meaningfully if a team played three of those games against the same star — a Doncic three-game homestand, for instance, will tank a defence’s “vs PG” rating for weeks afterward. Look at the underlying schedule before you trust the rolling number. If the recent fixture list contains two or three elite players at the position, the DvP rank is partly a comment on those players, not on the defence.
The third issue is more subtle. DvP is a per-game number, not a per-possession one. A slow team will allow fewer counting stats just because there are fewer possessions, and they’ll show up as “good” defenders by position even if their per-possession defence is mediocre. Pace adjustment fixes this in proper analytics tools. Most free DvP tables don’t bother.
DvP Reads Differently for Points, Rebounds, Assists
The rule I’ve ended up with is that DvP for points is the most reliable, DvP for assists is workable with caveats, and DvP for rebounds is borderline useless without context. The reasons are mechanical, not philosophical.
Points DvP works because scoring against a defence is a head-to-head event. A team that genuinely defends shooting guards well — say, with a switchable wing and a help-side rim protector — will produce a real depression on opposing SG points totals. The signal cuts through the noise reasonably well over a 30-game sample. If a team ranks top-five in DvP vs SG with no recent schedule distortion, that’s a meaningful read on a points prop.
Assists DvP is messier because assists are a function of teammate shot-making more than personal matchup. A point guard playing against a stingy defensive backcourt can still rack up 9-10 assists if his teammates hit shots. DvP for assists is best read as a tiebreaker — if everything else points to an over, a friendly assist DvP confirms the read. If everything else is neutral, the assist DvP alone isn’t enough.
Rebounds DvP is where I’ve stopped looking entirely. Rebounds depend on missed shots, lineup configuration, and box-out tendencies of multiple players on both teams. A “good” defensive team allows fewer points but creates more long rebounds, which can actually inflate the rebound counts of opposing bigs. The metric tries to summarise something it can’t, and the resulting number is too noisy to act on. For rebound props the better lookups are pace, opponent shooting volume and the centre’s individual rebound rate. DvP is decoration.
The Position Problem: Modern NBA Doesn’t Fit Five Slots
Here’s where the DvP construct really wobbles. The five-position model — PG, SG, SF, PF, C — was built for a basketball league that no longer fully exists. The 2025-26 NBA runs lineups with two ball-handlers and three wings, or one true big and four perimeter players, more often than it runs the classic 1-through-5. A 6’8″ point guard guards opposing 4s in switch coverage. A “centre” plays the four-spot when the team goes small. The position label is a uniform; the actual matchup is what’s on the floor.
This means DvP for the modern NBA is least reliable for the players you’d most want to use it on — the hybrid forwards, the big-guards, the playmaking 4s. A player listed as a SF might log 60% of his minutes against opposing PFs in switch-heavy lineups. The DvP-vs-SF column tells you nothing about his actual matchup that night. The DvP-vs-PF column does, but you wouldn’t know to look there unless you’d already watched the team’s recent lineups.
The fix is to ignore the metric for these cross-position players and look instead at the on-court matchup data on more analytics-leaning sites — who’s been guarded by whom, in what coverage. It’s slower, but it’s the only way to get a clean read on a hybrid player’s matchup. For traditional positions in traditional roles — a true PG, a true centre — DvP retains most of its usefulness. For everyone in between, treat the table as a starting point that probably needs revision.
Using DvP Without Overweighting It
The line that hangs above my prop spreadsheet, taken from a Wizard of Odds editorial last December, sums up the entire posture I take toward single-input metrics: “Uncertainty is part of the analysis. Don’t pretend you know true probability with precision you don’t have.” DvP is exactly the kind of input that invites overweighting because it produces a clean rank — top five, bottom five, middle of the pack. The cleanness is a trap. The metric is trying to compress dozens of variables into a single number, and the compression loses information.
The way I use DvP is as a binary filter, not as a continuous score. If the matchup is in the top five vs the player’s stat, I lean toward the over. If it’s in the bottom five, I lean toward the under. If it’s in the middle 20 ranks, I treat DvP as silent and let pace, usage and rotation do the talking. Trying to act on a “ranked 14th vs SG” DvP is a way to spend your bankroll on noise.
The other discipline is checking the schedule that produced the rank. A bottom-five DvP rank that’s been built off a string of star matchups is not actually a soft defence. It’s a defence that played a hard schedule. The same applies in reverse for “elite” defensive ranks built on a soft schedule. Five minutes of looking at the underlying fixtures gives you the context the table compresses out, and that’s the difference between using DvP and being used by it. The same kind of context-sensitivity matters even more once you start dealing with hybrid positions and switch-heavy lineups, which is the territory the deeper read on positionless basketball and prop models picks up.
DvP FAQ
Two questions on this topic come up almost every time I talk DvP with someone new to props, and they’re the questions worth answering carefully because the obvious answers are wrong.
Is DvP a leading or lagging indicator?
Lagging, with caveats. The rolling DvP table tells you what a defence has allowed up to today, not what it will allow tonight. Lineups change, players get traded, schemes get tweaked mid-season, and DvP only catches up after several games at the new normal. The shorter the rolling window the more recent the signal but the noisier the data. A 15-game rolling number is a defensible compromise. A season-long DvP that includes early-November games against a roster that no longer exists is mostly a historical document by March.
Where can UK bettors find DvP for free?
Several public fantasy and analytics sites publish positional defence tables that are accessible without paywalls — the standard public basketball-reference data, the major fantasy aggregators and a handful of analytics blogs. The free versions are usually season-long with no rolling cut, which is the main weakness. For free DvP that updates more responsively, the better route is checking points-allowed-by-position on team pages and doing the rolling cut yourself across the last 10-15 games.
DvP Is a Filter, Not a Verdict
If DvP enters your prop process at all, it should enter as one of several filters, not as the line that decides the bet. Use it to confirm a read you’ve already built from pace, usage, minutes projection and rotation news. Use it to flag a matchup that looks soft on paper. Don’t use it to override the rest of your projection.
The metric is honest about what it is — a season-long average of a positional outcome, with all the schedule and lineup baggage that average carries. Treat it that way, and it earns its place. Treat it as a verdict, and the noise will quietly bleed your bankroll across a season of “I had the matchup” bets that landed on the wrong side of the line.
Published by the nba Props Betting team.
