Moneyline
Who wins the game, at the price the book posted.
-0.4u
- Bets
- 11
- Record
- 5 · 6 · 0
- ROI
- -3.7%
- Avg CLV
- +1.09%
- Win rate against promised
- 45.5% vs 59.3%
The record
Every pick we call, locked the moment we call it at that price and line, and graded against the final score and the closing line at flat one unit stakes. Losses included. The database refuses edits and deletes.
Method
Four rules, so the numbers above mean what they say.
The engine calls a pick when it finds an edge of at least 2 percent over the vig free market price, the expected value is positive, and the certainty tier is A, B or C. The first time a pick clears that bar before its game, it is locked as a call at that price and line. If the edge disappears later, the call stays on the record and is marked as pulled.
Every call is staked at one flat unit at the price we called, with American price math. A winner at +150 returns 1.5 units, a winner at minus 120 returns 0.83, a loser costs one unit, a push costs nothing. ROI is units won per unit staked.
CLV is the move from the price we called to the price the market closed at, measured in probability. Beating the close over hundreds of bets is the earliest honest sign that an edge is real, long before the win rate settles.
Each call is stamped by the database itself with the time it was made. The database refuses a call for a game that has started, any change to a call once it is made, and any delete, for every account including the engine's own. The engine can only add the grade once the final score lands.
-1.3u
10 · 11 · 0won, lost, pushed
units at flat stakes across 21 graded calls · 27 more locked before kickoff, shown once graded
Equity
Flat one unit stakes, one point per day of games, from the first final to the newest.
Oct 29 to Oct 30
Where it comes from
The promised rate is what our probabilities said should happen. The gap between it and the real win rate is the honesty check, and tier A is the first place to look if the record ever turns.
17 bets graded
Record
10 · 7 · 0Win rate against promised
58.8% vs 62.8%
Units
+2.7u
ROI
+16.1%
Avg CLV
+1.61%
4 bets graded
Record
0 · 4 · 0Win rate against promised
0.0% vs 46.2%
Units
-4.0u
ROI
-100.0%
Avg CLV
+1.33%
By market
Who wins the game, at the price the book posted.
-0.4u
Who covers the number, at the price the book posted.
-1.7u
Every call
Newest first. The teams, the pick at the price and line we called it, when we called it, the tier, the result, the closing line value and the units it moved.
| Game | Pick | Tier | Result | Units |
|---|---|---|---|---|
| LAL @ GS u234.5 -106Called Fri, Oct 30, 2:22 PM · the line moved our way before the game |
Before launch
Before the first live pick, the engine replayed each season in order, training only on games that had already been played. The closing line is the hardest benchmark in sports. The blend sits within a hair of it, and the live record above is the test that matters.
2021 to 2026 · 1,532 games, walk forward
Log loss, lower is better
Over or under the posted total. The market the close guards best.
+0.8u
By tier
-1.0u
0 · 1 · 0 over 1 bet, ROI -100.0%
-0.3u
2 · 2 · 0 over 4 bets, ROI -6.3%
0.0u
8 · 8 · 0 over 16 bets, ROI -0.1%
| C49 |
| won |
| +0.94 |
| PHX +4.5 -110Called Fri, Oct 30, 2:43 AM · the line moved our way before the game | C50 | lost | -1.00 |
| PHX ML +143Called Fri, Oct 30, 2:43 AM · no longer a pick by game time, graded as called | C51 | lost | -1.00 |
| GS ML +114Called Fri, Oct 30, 2:43 AM · no longer a pick by game time, graded as called | C48 | won | +1.14 |
| GS +3.5 -108Called Fri, Oct 30, 2:22 PM · the line moved our way before the game | C51 | won | +0.93 |
| MIA -3.5 -157Called Fri, Oct 30, 2:43 AM · the line moved our way before the game | C47 | won | +0.64 |
| DAL ML +180Called Fri, Oct 30, 2:43 AM · the line moved our way before the game | C48 | won | +1.80 |
| MIA ML -245Called Fri, Oct 30, 2:43 AM · the line moved our way before the game | C48 | won | +0.41 |
| SEA ML -115Called Fri, Oct 30, 2:43 AM · no longer a pick by game time, graded as called | C53 | lost | -1.00 |
| NY -0.5 -112Called Fri, Oct 30, 2:43 AM · the line moved our way before the game | C47 | lost | -1.00 |
| TOR ML -200Called Thu, Oct 29, 10:45 PM · the line moved our way before the game | C51 | lost | -1.00 |
| BKN ML +270Called Fri, Oct 30, 2:22 PM · the line moved our way before the game | C47 | lost | -1.00 |
| NY @ PHI u230.5 -114Called Thu, Oct 29, 10:45 PM · the line moved our way before the game | B60 | won | +0.88 |
| PHI ML +138Called Fri, Oct 30, 3:10 PM · the line moved our way before the game | C49 | won | +1.38 |
| TOR -5.5 -108Called Fri, Oct 30, 2:43 AM · the line moved our way before the game | C45 | lost | -1.00 |
| WSH +0.5 -132Called Fri, Oct 30, 2:43 AM · no longer a pick by game time, graded as called | C52 | won | +0.76 |
| CHA @ WSH o220.5 -110Called Thu, Oct 29, 10:45 PM | A79 | lost | -1.00 |
| WSH ML -115Called Fri, Oct 30, 2:43 AM · no longer a pick by game time, graded as called | B75 | won | +0.87 |
| WPG ML +205Called Thu, Oct 29, 10:45 PM · the line moved our way before the game | B69 | lost | -1.00 |
| WPG +1.5 -122Called Fri, Oct 30, 2:43 AM | C58 | lost | -1.00 |
| CBJ ML +224Called Thu, Oct 29, 3:24 PM | B68 | lost | -1.00 |
Showing 21 of 21 graded calls in the last year, newest first. Times in Pacific.
The dashed line is the closing market. Winner accuracy 66.0%, the close 66.3%.
| Flat stakes vs the close | Bets | Record | Units | ROI |
|---|---|---|---|---|
| Moneyline edge 3% or more | 17 | 8 · 9 · 0 | +2.9u | +17.3% |
| Spread edge 3% or more | 27 | 13 · 14 · 0 | -1.9u | -7.0% |
| Spread edge 5% or more | 5 | 5 · 0 · 0 | +4.8u | +95.0% |
| Total edge 3% or more | 17 | 9 · 8 · 0 | +0.2u | +1.3% |
2025 to 2027 · 2,729 games, walk forward
Log loss, lower is better
The dashed line is the closing market. Winner accuracy 68.9%, the close 69.0%.
| Flat stakes vs the close | Bets | Record | Units | ROI |
|---|---|---|---|---|
| Moneyline edge 3% or more | 14 | 7 · 7 · 0 | -1.2u | -8.4% |
| Moneyline edge 5% or more | 2 | 0 · 2 · 0 | -2.0u | -100.0% |
| Spread edge 3% or more | 10 | 5 · 5 · 0 | -0.4u | -3.6% |
| Spread edge 5% or more | 1 | 1 · 0 · 0 | +1.0u | +98.0% |
| Total edge 3% or more | 15 | 8 · 7 · 0 | +0.4u | +2.4% |
| Total edge 5% or more | 2 | 1 · 1 · 0 | -0.2u | -8.3% |
2026 to 2027 · 1,602 games, walk forward
Log loss, lower is better
The dashed line is the closing market. Winner accuracy 55.7%, the close 56.1%.
| Flat stakes vs the close | Bets | Record | Units | ROI |
|---|---|---|---|---|
| Moneyline edge 3% or more | 21 | 5 · 16 · 0 | -9.3u | -44.3% |
| Moneyline edge 5% or more | 3 | 2 · 1 · 0 | +2.3u | +78.0% |
| Spread edge 3% or more | 70 | 46 · 24 · 0 | +1.7u | +2.4% |
| Spread edge 5% or more | 29 | 20 · 9 · 0 | +3.6u | +12.4% |
2024 to 2026 · 2,497 games, walk forward
Log loss, lower is better
The dashed line is the closing market. Winner accuracy 76.6%, the close 73.9%.
| Flat stakes vs the close | Bets | Record | Units | ROI |
|---|---|---|---|---|
| Moneyline edge 3% or more | 159 | 108 · 51 · 0 | +7.3u | +4.6% |
| Moneyline edge 5% or more | 33 | 21 · 12 · 0 | +2.5u | +7.7% |
| Spread edge 3% or more | 82 | 47 · 35 · 0 | +9.1u | +11.1% |
| Spread edge 5% or more | 23 | 13 · 10 · 0 | +2.0u | +8.5% |
| Total edge 3% or more | 104 | 48 · 56 · 0 | -11.6u | -11.1% |
| Total edge 5% or more | 14 | 5 · 9 · 0 | -4.3u | -30.5% |
Log loss is the penalty for confident wrong probabilities; a difference of a few thousandths over a season is the whole game, so the bars share a floor just below the best number. The flat stake rows show what betting every pick above an edge threshold would have returned against the closing price, the losing seasons included. Totals have not beaten the close in the backtest, and that is why they carry less weight on the live board.