Tonight wraps up the NBA's last slate before the All-Star break, and then something unusual happens. Everything goes quiet. No NBA, no NHL, nothing until Thursday, February 19th. A full week with zero pro sports on the board.
We're not sitting around waiting.
The Second Half Starts Here
The All-Star break is one of those natural reset points in the NBA season. Before the break, the league is still figuring itself out. Rotations are in flux. Teams are experimenting. Players are getting load-managed in ways that make modeling a nightmare.
After the break, things tighten up. Playoff positioning starts to matter. Coaches shorten their rotations. The data gets more predictable, and more predictable data means sharper edges for the model to find.
That shift alone should work in our favor, but we're not banking on it.
Where Things Stand
The NBA model is sitting at 92-97-2 with a 48.7% win rate and -3.00 units on the season. Not where we want to be, but also not far off. A few games swing differently and we're having a very different conversation right now. That's the nature of betting on razor-thin margins. The model is close. Close enough that the right adjustments during this break could make a real difference in the second half.
What Lab Week Looks Like
This week is all about refinement. Not scrapping anything, not starting over. Tuning. The foundation is solid, but there are specific areas where we think the calibration can get tighter.
Here's what we're digging into:
Rest and fatigue modeling. Back-to-backs, road stretches, schedule density. The NBA is the sport where these factors hit hardest, and we want to make sure our weighting reflects that accurately.
Injury processing. Not just who's out, but how the model adjusts when key players return after extended absences and teams need time to re-integrate them. That transition period is tricky to capture and we think there's an edge in getting it right.
Totals calibration. Pace of play shifts throughout the season, and the model needs to stay locked in on how teams are actually playing right now, not how they played in November.
And it's not just the NBA getting attention. The NHL model has been performing well this season, but "performing well" doesn't mean "done." We're using this week to sharpen that up too. Good is cool. Better is the goal.
A Richer Dataset Ahead
Here's the thing about the second half of the NBA season that gets overlooked: the data is just better. You've got 50+ games of sample size for every team. The early-season noise (roster turnover, new coaching systems, players finding their rhythm) has mostly settled. What you're left with is a cleaner picture of who these teams actually are.
Better inputs, better outputs. Combine that with targeted refinements this week and we like where this is headed going into the stretch run.
Back at It February 19th
The break gives us something we rarely get mid-season: time to step back, look at the big picture, and make real improvements without games flying at us every night. Think of it like a bye week. Reviewing film, installing new plays, getting ready for the second half.
See you on the 19th.