Two overs and a Lakers spread clear the model's 3-point gate this morning, while a stack of games get held back for sitting too far off the market.
Everything below is a morning lean, generated off this morning's board on 2026-10-27. The official pick for each game locks roughly 60 minutes before tip, at whatever line is live then. If a number moves between now and lock, the locked pick can differ from what's written here, or it can disappear from the slate entirely. Only locked picks count toward the published record.
The model only speaks up when its own number sits within 3 points of the market. Anything wider gets treated as the model missing something real — an injury, a lineup change, whatever — rather than a bigger edge worth chasing. That gap is a kill switch, not a confidence booster.
DET at NYK (7:00 PM CT): the lean is OVER 218.5 at -112. Model total sits at 219.33 against the market's 218.5 — a tight 0.8-point gap, well inside the window.
SAC at SAS (7:00 PM CT): lean is OVER 231.5 at -110. Model total is 231.84 versus 231.5 on the board, a 0.3-point gap — about as close as this model ever gets to agreeing with the market outright.
POR at LAL (10:00 PM CT): lean is LAL -2.5 at -108. The model's margin favors LAL by 0.7 points more than the market number, projecting LAL 116.32 to POR 113.15 — a 3.17-point margin on a projected 229.47 total.
Five numbers got shelved this morning, all for blowing past the 3-point cap. BKN at BOS saw both sides held — the spread disagreed by 5.0 points, and the total missed by 3.1 (model 220.37 vs. market 223.5). DET at NYK's spread was also held, with an 8.3-point gap between model and market even though the total for that same game cleared. SAC at SAS saw its spread held on a 5.0-point disagreement, again while the total passed. And POR at LAL's total didn't make it — the model's 229.47 sat 4.0 points under the market's 233.5, even as the spread for that game made the cut.
None of that is the model hedging. It's the rule working as designed: big daylight between model and market means something's probably being missed, and the system would rather say nothing than guess.
Locked picks stand at 19-18, -0.87 units on the season — spreads 9-10, totals 10-8. That's off 52 games of data. In the model's own backtesting, no slice of its picks has cleared the 52.4% break-even most books require. This is a market-level model producing market-level results, not a demonstrated edge, and nothing here should be read as financial advice. Check the board again closer to tip — these are morning reads, and the lock is where it counts.