NBA Double-Double Props: How the Historical-Rate Model Works
The simplest model on the site, one of the best graded, and a market most books price lazily. What the rate is, what it hit, and what it cannot see.
The short version
A double-double is a threshold event — 10 or more in two of points, rebounds and assists — so the natural model is not a projection but a rate: how often has this player done it? Our board publishes that rate for every candidate on the slate, badges a play at 60%, and last season the plays went 93-53 (63.7%). More usefully, the outcome tracked the rate: players the model put at 80%+ recorded one 85.7% of the time, players at 60-70% only 55.8% — under their stated rate. That ordering is the whole edge, and it decides what price is worth paying.
The method
1. Build the rate from the box scores
For every player the model reads the current season's box scores, keeps games with 20 or more minutes, and needs at least ten of them before it will say anything. The rate is the share of those games in which he reached ten in two categories. A player with 47 double-doubles in 61 qualifying games is 77.0%, and that number is what the card shows. Triple-doubles are the same calculation with three categories.
2. Show the averages behind it
A rate alone hides the mechanism. The card also shows the player's last-eight averages in points, rebounds and assists, so you can see which two categories carry the double-double and how much margin each has. 15.8 points and 11.4 rebounds is a double-double built on two comfortable categories; 22.1 points and 9.8 rebounds is one that lives on a coin flip in the second, whatever the season rate says.
3. Threshold and grade
Candidates below a 5% rate are not listed. A play is a rate of 60% or more, the same threshold the grader uses, so the record below counts exactly the plays that were badged. The next morning each play is settled against the official box score: a double-double recorded is a win, anything else a loss. Players who were ruled out or did not play are excluded rather than counted as losses.
What it hit, graded — and why the split matters more than the total
Every play from March 12, 2026 to June 13, 2026 (79 game days, playoffs included). The break-even at -110 is 52.4%; double-double props on high-rate players are usually priced heavier than that, which is why the tier column is the one to read.
| Season rate on the card | Record | Actually recorded one | Break-even price |
|---|---|---|---|
| 60–70% rate | 53-42 | 55.8% | -126 |
| 70–80% rate | 22-9 | 71.0% | -244 |
| 80%+ rate | 18-3 | 85.7% | -600 |
| All plays (60%+) | 93-53 | 63.7% | -176 |
Tier rows are a replay of the grader over the archived pick files (93-54 in the replay; the tracker's per-day count is 93-53). The total is the tracker on the NBA performance page. Break-even price = the American odds at which that hit rate returns exactly zero.
The last column is the point of this article. The rate on the card orders the outcomes — each tier hit more often than the one below it, and the top tier hit about its stated rate — so it can be put against a price, with the caution that the biggest tier came in under its rate. A 65%-rate player at -200 is a losing bet even though he records a double-double two nights in three: -200 needs 66.7% and the 60-70% tier delivered 55.8%. An 85%-rate player at -250 is a winning bet with room to spare. In this market the rate tells you what to pay, and most of the value is in refusing the middle tier at heavy prices.
The 80%+ row is 21 plays; treat its 85.7% as “high, and consistent with the rate,” not as a number to quote to two decimals. The 60-70% row is the one with a real sample, and it is the one that says be careful.
What the model does not see
Tonight
The rate is a season average. It carries no matchup, no projected minutes and no blowout risk; a home game against the slowest team in the league and a road back-to-back in Denver count the same. Rebound-driven double-doubles are sensitive to pace and the opponent's shot profile — long three-point misses go to wings, not centres — and none of that is in the number.
Who else is on the floor
A starting centre's rebound share rises when his backup is out and falls when a rebounding forward returns. The board hides players who are ruled out, but it does not re-project the rest of the rotation around them. This is the most common reason a 75% player misses, and it lives in the injury report an hour before tip-off.
The second category's margin
Two players at the same rate can have very different risk. Read the three averages: when the second category averages 11.5 or more the double-double is structural; when it averages 10.2 it is a nightly coin flip that happened to land often. A season rate cannot tell those apart. The card can.
How to use it
Start from the rate, then do three checks the model cannot: the price against the tier's break-even, the injury report for the player's own team, and the margin in his second category. When all three line up — an 80%+ rate, a comfortable second category, a price short of -300 — the market is offering what the season says it should offer, and there are usually only one or two of those on a slate. The board sorts by rate so they are at the top.
Tonight's double-double board
Every candidate on the slate with his rate, the three averages and the games count. The top five are free; the full board, with triple-doubles, is for subscribers. First board of the season: opening night, October 20.
Frequently asked questions
How do you predict NBA double-doubles?
With a rate: the share of a player's games this season (20+ minutes) in which he reached 10 or more in two of points, rebounds and assists. That rate is the probability on the card, alongside his recent averages in the three categories. A play is a rate of 60% or higher.
What is your double-double hit rate?
93-53 (63.7%) over 146 plays at the 60% threshold, graded March 12 to June 13, 2026. The outcome rose with the rate: 60-70% rates hit 55.8% (below the rate), 70-80% rates hit 71.0%, 80%+ rates hit 85.7%.
What price do you need on a double-double prop?
Whatever the rate says minus a margin. At -150 you need 60%; at -200, 66.7%; at -250, 71.4%. A player with a 65% rate at -200 is a losing bet even though he will record a double-double most nights. The high-rate tier is where the number and the price can both be on your side.
What does the model not account for?
Tonight. It uses no matchup, no minutes projection and no blowout risk — every game counts the same. A teammate's absence that changes a big's rebound share is not in the rate either. That is why the card shows the three averages: the margin over 10 in the second category is the thing to check yourself.
Do you predict triple-doubles?
Yes, with the same rate method and a 30% play threshold; last season's record was 11-10 (52.4%) over 21 plays, too few to say much. It is on the subscriber board with the double-doubles.
The week ahead, once a week
Last week's graded record — wins and losses — the plays for the week ahead, and a subscriber discount code. One email, no daily noise, unsubscribe in a click.