The model likes a slice of value in Atlanta, a number in New Orleans, and Portland's number — with one total held back entirely.
These are morning reads on where the model sits relative to the market as of today, October 29. Nothing here is final. Each game gets an official locked pick roughly 60 minutes before tip, built off whatever line is actually on the board at that moment. If a number moves between now and then, the locked pick can differ from what's written below, or it can vanish entirely. The published record only counts picks that actually locked.
The model has a rule baked in: it only publishes a side when its own number sits within 3 points of the market. A gap bigger than that isn't treated as a bigger opportunity — it's treated as the model probably missing something, an injury or a lineup change it doesn't know about yet. So it stays quiet instead of guessing.
CLE at ATL (7:00 PM CT) is the one game with two live leans. The model projects Atlanta 120.78, Cleveland 117.62 — a margin of roughly 3.16 points for Atlanta against a market line of -1.5. That's a 1.7-point gap in Atlanta's favor, comfortably inside the publishing window, and it's enough for the model to lean ATL -1.5 at -105.
The total follows the same math. Model total sits at 238.41 versus a market number of 237.5, less than a point of separation. That's the lean toward the OVER 237.5 at -115 — not a loud disagreement, just a small lean the model is comfortable publishing.
SAC at NOP (7:00 PM CT) projects out to New Orleans 120.04, Sacramento 114.93, a margin near 5.11 for the Pelicans against a market spread of 6.5. That 1.4-point gap favors Sacramento getting the extra points, so the lean is SAC +6.5 at -118. The total side of that same game comes in tight: model total 234.97 against a market 235.5, half a point of daylight, pushing the lean to UNDER 235.5 at -105.
MEM at POR (9:00 PM CT) is a split case. On the spread, the model has Portland winning by 7.97 against a market line of -7.5 — a half-point gap, about as close as these get, and it supports the lean of POR -7.5 at -110. The total in that same game is a different story. Model total sits at 229.99 against a market of 233.5, a 3.5-point gap that clears the model's own 3-point cap. That total gets held back rather than published, because the model treats that size of disagreement as a red flag on its own information, not a green light.
The season record on locked picks is 28-27, sitting at -1.71 units, with spreads running 14-14 and totals 14-13. These ratings are built off 68 games so far this season.
Worth saying plainly: in the model's own backtesting, no particular band or slice of its picks has beaten the 52.4% break-even threshold that bettors need to clear the vig. That means this is being presented as a market-level model, one that tracks lines closely rather than consistently beating them, not as a demonstrated edge. Treat all of the above as one model's output for the morning, not financial advice, and remember the locked pick near tip is the one that actually counts.