NBA value betting means backing a price above the fair odds for that selection, and in the NBA the gap is most often real in the player prop market. The head to head is priced by everyone and lands within a tick or two across the board. A player points line is priced by fewer operators, carries more margin, and can sit on a different number altogether at the next operator.
It is also the easiest place to get the method wrong, for reasons that have nothing to do with basketball. Two prop prices are frequently not two prices for the same bet, and the news that decides a prop lands late. A scratching before tip-off can move that player's markets further than anything moves the team head to head, and on an Australian clock it lands before breakfast.
The prop market is where the NBA margin sits
Measure the margin rather than assume it. Implied probability = 1 / decimal odds, so add one over every price in a market: the amount over 100% is what you pay before you have an opinion.
| Market | Outcomes | Why the margin lands there |
|---|---|---|
| Head to head | 2 | Everyone prices it and the money concentrates there |
| Line and total | 2 per line | Two-way, but operators hold different lines |
| Player points, rebounds or assists | 2 per rung | Fewer operators, a ladder of rungs, moves on news |
| Points, rebounds and assists combined | 2 per rung | Three stats, so three sources of error in one bet |
| First basket, first team to 20 | 10 or more | Margin split so wide nobody counts it |
Worked example: an over and under on a points line
Illustrative prices on a 24.5 points line, to show the arithmetic rather than a real game.
| Side | Priced at | Implies | Fair odds |
|---|---|---|---|
| Over 24.5 | $1.80 | 55.56% | $1.92 |
| Under 24.5 | $1.95 | 51.28% | $2.08 |
1 / 1.80 is 55.56% and 1 / 1.95 is 51.28%, so the book totals 106.84%. Scale both by that factor: 55.56 / 1.0684 = 52.00% and 51.28 / 1.0684 = 48.00%, which are fair odds of $1.92 and $2.08. Now say the best over 24.5 anywhere is $2.00, and the best under is still $1.95.
Example: over at $2.00 against a fair $1.9231 is 2.00 / 1.9231 - 1, an edge of 4.00%. Under at $1.95 against a fair $2.0833 is minus 6.40%. One is a bet and the other is the margin.
Kelly stake = edge / (decimal odds - 1), so 4.00% / 1.00 = 4.00% of bankroll, and a quarter Kelly on a $2,000 bankroll is about $20.
A 4.00% edge on a two-way prop is large, and a large number is a prompt to check rather than to bet. A stale line at the operator you de-vigged, two prices on different rungs, or news one operator has not caught up with each explain it more often than a mispricing does. These fair odds also carry one operator's opinion, because they came out of one operator's market. Build them from several.
A team total, a player total and an alternate line
| The bet | What settles it |
|---|---|
| Team total over 112.5 | The team's final score |
| Player total over 24.5 points | One player's points |
| Alternate line, over 29.5 points | The same player's points, at a different threshold |
The third row is the expensive one: the player and the stat match and only the number differs, so it lines up in a table as though it were the same bet. A 27-point game pays over 24.5 and loses over 29.5, and no de-vig turns a price on one rung into a price on another, because the two markets share no outcomes. Comparing rungs needs a distribution of that player's points, which is a model rather than arithmetic.
Note: check the stat as well as the number. Points, rebounds and assists combined is one market, and three separate markets added together is a different bet that settles differently, see points, rebounds and assists and correlated bets.
Why an even de-vig flatters the longest price
Scaling every implied chance by the same factor assumes the margin is spread evenly. On a two-way prop that is close enough. On a first basket market it is not.
Ten illustrative first basket prices: $3.00, $5.00, $7.00, $9.00, $11.00, $13.00, $15.00, $21.00, $26.00 and $31.00. One over each adds to 114.01%, so 14.01 points of margin have to come out.
| Method | Fair odds on the $3.00 | Fair odds on the $31.00 |
|---|---|---|
| Scale every chance by the same factor | $3.42 | $35.34 |
| Take the same number of points off each chance | $3.13 | $54.81 |
Dividing by 1.1401 turns 33.33% into 29.24% and 3.23% into 2.83%. Taking a tenth of the margin, 1.4014 points, off each chance instead gives 31.93% and 1.82%. The two answers are 9% apart on the favourite and 55% apart on the longest price.
Example: $36.00 offered on that longest outcome is an edge of plus 1.86% against the even method and minus 34.32% against the other. Same price, same market, 36 percentage points of disagreement.
Neither method is right, which is why the power method and the Shin method exist. For the NBA that is a boundary rather than a warning: on a two-way prop the methods agree closely enough to work with, and on first basket or first team to 20 they disagree by more than any edge you would claim, so a wide NBA market is not a place to price an edge at all.
The news breaks while Australia is asleep
NBA games tip off in the US evening, which is the Australian morning, so the reports that decide a prop land here overnight. The NBA publishes an injury report and runs a player participation policy covering rest, and the beat reporters follow both; NBA betting strategy sets out those published deadlines converted to Australian time.
What matters for a value bet is what the news does to the arithmetic, because the edge is a ratio to the fair price, and the fair price moves while the offered price sits still. Keep the illustrative $2.00 on offer and move only the fair number.
| What a rotation change does to the over's true chance | Fair odds | Edge on $2.00 |
|---|---|---|
| 55%, a teammate is out and the minutes go up | $1.8182 | plus 10.00% |
| 52%, the number you de-vigged last night | $1.9231 | plus 4.00% |
| 49%, a return to the rotation takes minutes away | $2.0408 | minus 2.00% |
Three points of probability either way is an ordinary response to one change in a rotation, and it moves a 4.00% edge to 10.00% or to a losing bet. Nothing about the price you were watching moved at all.
Risk: Betting involves risk. A prop price that looked like an edge last night may be priced on a player since ruled out or on a rotation that has changed, so a stale fair price is the explanation to rule out first. See responsible gambling for limits and support.
Rules written in advance beat watching a screen
If the decisive information arrives while you are asleep, you either stay awake for it or decide beforehand what you would do. Written down, the second is a price floor below which the bet is refused, a maximum stake, an odds range, and a list of markets you will touch beside one you will not.
The arithmetic gives you the floor. If the fair odds on over 24.5 are $1.9231 and you want 2% of edge to cover the risk of a late change, the floor is 1.9231 x 1.02 = $1.96, so $1.95 is a pass rather than a near miss.
A floor is also the only defence against a price that was right when it was read and wrong by the time you clicked, because prices on any comparison board are read on a repeating cycle rather than tick by tick. Confirm the price in your own account before you bet.
Tip: write the floor down as a number before the news, not after. A rule you can renegotiate at 2:00am is not a rule.
What to check before you trust an NBA prop edge
The rung. Over 24.5 and over 25.5 are different bets, and a row that mixes them shows the biggest edge on the board.
Participation. A did-not-play and a late scratch are not always settled the same way, and the difference is whether your stake comes back: did not play bet rules.
Overtime. Player and team totals usually include it, so a four-quarter estimate understates the line: does overtime count.
The sample. One game can hand you a long list of prop bets, and that is not a long list of independent results: they share the same minutes, the same pace and the same overtime, so one bad read on the rotation shows up across all of them at once. Count games rather than legs, and read closing line value price by price rather than waiting on profit.
Where B337 fits
The Terminal lines up racing and sports prices from 40+ bookmakers side by side, including player props where bookmakers offer them. The +EV screener ranks the bookmaker prices sitting above the fair price by edge, filtered by sport, market and bookmaker. That fair price is built by taking the margin out of prices across the market, including exchanges and global market references, and combining them, and the de-vig is adjustable, so you can change which prices count and how much margin comes off each.
Three honest limits. The fair price is an estimate rather than a known probability, so a price above it can still lose. Prices are read on a repeating cycle, so a price on screen can trail the bookmaker's own. And B337 places no NBA prop bet on its own: the screener finds the price and you place it into an account you hold, by hand on Terminal Pro or from your own code through the betting API, and a bet placed through B337 uses credits, a per-bet usage charge. A free account opens a limited view of the Terminal with live odds, plus the live count of prices above fair, with the rows locked. For free and confidential support call 1800 858 858 or visit gamblinghelponline.org.au.
Risk: Betting involves risk. A screener row points at a price above an estimated fair price, not at a winner, a prop priced above fair can still lose, and there is no guarantee of profit. See responsible gambling for limits and support.
For the method with no sport attached, see value betting.