Last updated: January 16, 2026
Today I’m officially launching a new feature on OddsRX: NBA Player Stat Predictions + Performance Tracking.
This is a big step forward for the platform — not just because we’re projecting player stats (Points, Rebounds, Assists, 3PM, and Minutes)… but because we’re doing the most important part too:
✅ Tracking accuracy publicly.
✅ Showing where the model is strong (and where it’s not).
✅ Improving over time with real results and calibration.
If you’ve followed OddsRX so far, you know this has always been the mission:
Build tools that can help make smarter decisions — and back it up with real data.
What’s Included in the Player Stat Predictions
OddsRX now produces player-level projections across the most important box score categories:
- Points (PTS)
- Rebounds (REB)
- Assists (AST)
- 3-Pointers Made (3PM)
- Minutes (MIN)
These aren’t meant to be random “gut feel” guesses. They’re designed to be consistent, repeatable, and rooted in historical performance, role, usage, team context, and matchup factors.
And just as important — they’re designed to be evaluated.
The Bigger Feature: Accuracy Tracking That Updates Over Time
The projections are only half the story.
The real reason I’m excited about this launch is that OddsRX now includes a dedicated page that tracks performance in a transparent way — by stat category, by team, and by player.
This makes it easy to answer questions like:
- “How accurate are these projections overall?”
- “Which categories are the most predictable?”
- “Are we consistently too high or too low on certain stats?”
- “Which teams are hard to project because of rotations?”
- “Which players are stable vs volatile?”
That matters because in sports betting — and especially in player props — the edges come from:
understanding variance, not ignoring it.
What Makes This Different From Normal Projections
Most projection systems stop at the projection.
OddsRX isn’t doing that.
This dashboard doesn’t just show “the number.”
It shows:
✅ How far off the projection typically is (error size)
We track MAE (Mean Absolute Error) and RMSE, which help quantify the average miss and how often we get outlier games.
✅ Whether the model tends to run high or low (bias)
Bias matters. A projection can look accurate overall but still lean in one direction — and that can impact how you treat certain props.
✅ How often the projection is “close” (hit rates)
Hit rates show the percentage of projections that land within a range of the true result (like ±1, ±2, ±3, etc.).
This is one of the most useful “game day” measures because it reflects real-world volatility.
Why Minutes Might Be the Key to Everything
If you’ve ever bet player props, you already know this:
Minutes are everything.
Minutes drive:
- scoring opportunity
- rebounding opportunity
- assist opportunity
- shooting volume
- 3PT attempts
Minutes are also the hardest stat to predict because they can swing fast from:
- blowouts
- foul trouble
- injuries
- coaching adjustments
- rest situations
- rotation experimentation
One of the biggest goals of this project is tightening minutes accuracy, because when minutes are right, the rest of the stat line becomes easier to project too.
How You Can Use This as a Better Prop Tool
This isn’t a “lock button.”
It’s a data tool.
If you’re looking at player props, this helps you answer:
1) “Is this stat category reliable?”
Assists and rebounds often behave more consistently than points or threes.
2) “Is this player predictable?”
Some players are stable nightly roles. Others are matchup-dependent or rotation-sensitive.
3) “Is this team a rotation mess?”
The By Team section helps quickly identify teams that are difficult to project.
4) “How much error should I expect?”
If the typical miss is 5 points, a line edge of 0.5 points isn’t exactly a slam dunk.
Understanding the model’s natural error range helps frame confidence and sizing.
This Is Version 1 — and It’s Going to Improve
This launch is the first step, not the finish line.
Version 1 is focused on building:
- consistent daily player projections
- stable data delivery to the website
- clean evaluation logic
- a public tracking page that keeps us honest
From here, the improvements get exciting:
- stronger minute forecasting
- better handling of questionable/out tags
- improved rotation assumptions
- faster calibration feedback loops
- refining how projections behave for volatile bench roles
Where to Find It
If you want to explore the performance tracking dashboard, you can view it here:
(Stats by Category → by Team → by Player)
This page will continue updating as more games are played and the sample size grows.
Final Thoughts
Player projections are easy to publish.
Accountability is the hard part.
OddsRX is built around transparency — not just “calling picks,” but tracking what works, what doesn’t, and what’s improving.
If you’re using OddsRX already, I appreciate you being part of it.
And if you’re new here: welcome.
This is where the platform starts getting really fun.
—Dr. Cover