NBA Midseason Report: Which Prop Markets Are Most Profitable in 2026?

We are roughly 55 games into the 2025-26 NBA season, the All-Star break is here, and we finally have enough data to answer the question that every prop bettor should be asking: which markets are actually profitable this season, and which ones are a mirage?

At Turtle +EV, we have graded thousands of NBA player prop picks since October. Every single one — win or loss — is tracked, timestamped, and available for review. No hiding the losers, no cherry-picking the winners. That transparency lets us do something that most prop betting services cannot: honest midseason analysis of what is working, what is not, and where the edges are for the second half of the season.

The Midseason Data Advantage

Early in the season, our model operated with limited current-season data. Player sigmas — the standard deviation of each player's performance for each stat type — were estimated from a combination of last season's data and a small number of 2025-26 games. That meant wider confidence intervals, more conservative probability estimates, and fewer picks clearing our 6% EV threshold.

Now, with 50+ games per team in the books, the picture is dramatically different. Player sigmas are calibrated against a robust sample of current-season performance. Our weighted averages — Last 3 games (50% weight), Last 5 (30%), and Season (20%) — are now anchored by meaningful season-long data rather than noisy early returns. The model is more confident, more accurate, and surfacing more picks per night.

This is also the point in the season where sportsbook lines become sharper. Books have the same data we do, and they have been adjusting all season. The easy edges from October — mispriced role-change players, anchored preseason lines — are mostly gone. What remains are the structural edges: stat types where books systematically misprice probability, players whose consistency the market underestimates, and directional biases that persist because of how the public bets.

Which Stat Types Are Hitting?

Breaking down our graded NBA picks by stat type reveals clear winners and losers at the midseason mark. Here is what we are seeing across the major prop markets.

Combo Props: The Clear Winner

Pts+Reb, Pts+Ast, and Pts+Reb+Ast continue to be our strongest NBA stat types by a significant margin. These combo markets carry higher win rates and stronger ROI than standalone stat props.

The reason is structural. Sportsbooks often price combo props by approximating — they add the individual lines together or use a simple formula to estimate the combined total. But stats are correlated within a game. A player who is having a big scoring night is often also accumulating assists (because his team is in the flow of the offense) or rebounds (because more possessions and more shots mean more rebound opportunities). That intra-game correlation means the true probability of hitting a combo OVER is slightly higher than the simple addition would suggest.

Our model captures these correlations through our probability calibration, which uses sport-specific slopes for combo stat types. The result is that combo props clear our EV threshold more frequently and win at a higher clip.

Rebounds: Quietly Strong

Rebounds have been one of our most consistent prop markets this season. Rebounding is one of the more stable stat types game to game — a player who averages 10 rebounds per game rarely drops below 6 or spikes above 16. That tighter distribution means our probability estimates are more accurate, and when we identify an edge, it tends to hold.

Rebounds also benefit from being a "boring" stat for casual bettors. The public bets points and threes. Rebounds get less action, which means less public money distorting the lines. The lines stay closer to where they should be most nights, but when they drift — because of a pace matchup, a rest day for a teammate, or a subtle rotation change — the edge is real and underexploited.

Assists: Matchup-Dependent But Profitable

Assists are more volatile than rebounds but still profitable overall. The key with assists is that they are heavily influenced by game flow. A point guard in a close game that goes to overtime will rack up assists. The same player in a 25-point blowout where he sits the entire fourth quarter might finish with 4. That variability makes assists harder to model precisely, but it also means the market misprices them more often.

The best assist edges this season have come from targeting specific matchups where a team's defensive scheme funnels the ball through the opposing point guard. Teams that switch heavily on defense tend to give up more assists because the ball handler can consistently create advantages off the switch.

Points: The Most Efficient Market

Points props are the most liquid and most efficiently priced market in NBA player props. That does not mean there are no edges — there are — but they are smaller and harder to find. Points props are the first thing casual bettors look at and the first thing sportsbooks optimize. The result is a market where the average edge per pick is lower, but the volume is high enough that you can still build a profitable portfolio.

Interestingly, the directional split on points props matters. UNDER picks on points have performed better than OVER picks across our graded sample. This is consistent with a broader pattern we see across all NBA stat types: the public tends to bet OVER, which pushes lines up, which creates UNDER value. Our model accounts for this directional bias through a calibration adjustment that compresses OVER probabilities and is more conservative on high-side predictions.

Three-Pointers Made: High Variance, Moderate Edges

Three-Pointers Made is the most volatile stat type in our NBA model. A player might hit 6 threes one night and 0 the next. That variance makes 3PM props inherently noisier, which means you need a larger sample before you can say with confidence whether your edge is real or just luck.

At the midseason mark, 3PM is roughly breakeven-to-slightly-profitable for us. The edges exist but they are harder to size correctly because the probability distribution for threes is not as clean as rebounds or assists. We continue to generate 3PM picks but with the understanding that this stat type requires patience and a longer runway to show its full value.

Steals, Blocks, and Fantasy Points

Steals and Blocks are low-volume prop markets that rarely clear our EV threshold. The lines are typically set at 0.5 or 1.5, which creates a binary outcome that our continuous probability model handles less precisely than higher-count stats. When we do surface a steals or blocks pick, it tends to be on a player with an unusually high defensive rate against a turnover-prone opponent.

Fantasy Points has been a solid performer. Fantasy scoring aggregates points, rebounds, assists, steals, blocks, and turnovers into a single composite number. The aggregation smooths out the variance of any individual stat, which makes fantasy props more predictable than standalone stats for high-usage players. Books sometimes price fantasy lines based on simple point projections without fully accounting for the defensive stats that can push a player's fantasy total well above his scoring output suggests.

The OVER vs. UNDER Split

One of the most important findings from our midseason analysis is the persistent directional bias in the market. Across all NBA stat types, UNDER picks have outperformed OVER picks in our graded results.

This is not a coincidence — it is a structural feature of the prop betting market. The public overwhelmingly bets OVER on player props. They want to root for big performances, and betting UNDER on a star player feels counterintuitive. That one-sided action pushes OVER lines higher than they should be, which creates systematic UNDER value.

Our model handles this through direction-specific calibration. OVER probabilities are compressed by a factor that reflects this market bias, while UNDER probabilities receive a smaller adjustment. The result is that our UNDER picks carry, on average, higher conviction than our OVER picks — and the graded results confirm that the adjustment is working.

This does not mean every UNDER pick wins or every OVER pick loses. It means that over a large sample, the market's OVER bias creates a structural edge for UNDER bettors who have an accurate probability model. If you have been skipping UNDER picks because they feel less exciting, the data says you are leaving money on the table.

What to Expect in the Second Half

The second half of the NBA season brings several dynamics that affect prop markets.

The trade deadline (February 8) will create short-term chaos.Players moving to new teams will have mispriced props for the first few games in their new role. We will cover this in depth in a separate post.

Rest and load management increase. As teams clinch playoff spots or fall out of contention, stars will sit more games. This creates backup-player prop edges similar to what we saw in the early season — books scramble to set lines for players who do not normally start, and those lines are often soft.

Playoff implications sharpen performance. Players on teams fighting for seeding tend to play harder and more minutes in March and April. That uptick in intensity can push counting stats above their season averages, making late-season OVER picks on playoff-hopeful teams more attractive than the season-long data suggests.

The midseason sample tells a clear story: combo props and rebounds are the strongest markets, UNDER picks outperform OVER picks, and a disciplined approach to +EV betting works. The second half of the season is where those edges compound into real profit. Stay patient, trust the math, and let the sample do its work.