Receipts: a losing week for Mako Bets, down 313.86 units across every model
Most of the loss came from a handful of +EV scanner and Fair Odds runs, several of which were later stopped. Some models had good weeks, but most samples are still too small to judge.
The honest headline: a losing week
Across every model, the seven days since 2026-11-03 netted -313.86 units. That is a losing week, and the table below shows where it came from. A few models were profitable. Most of the table was red, and the biggest losses were large enough to outweigh every gain.
| Model | This week | Week units | Season W-L | Season units | Season return | Graded | Status |
|---|---|---|---|---|---|---|---|
| +EV scanner, no long shots (EV1-XC) | 32-11 | +26.07 | 32-11 | +26.07 | +60.6% | 43 | Shadow of EV1-X from Nov 6; read at 300 bets |
| +EV scanner, power de-vig (EV1-XP) | 56-48 | +18.85 | 110-104 | +17.72 | +8.3% | 214 | Shadow of EV1-X from Oct 24; read with it |
| L1 rushing and receiving unders | 24-16 | +12.55 | 89-79 | +23.33 | +13.9% | 168 | Forward test in progress |
| College basketball first half | 53-41-2 | +8.93 | 127-87 | +31.28 | +14.5% | 216 | Forward test; picks at 5%+ EV are public on /model-predictions |
| L2 receptions usage model | 59-38 | +8.45 | 159-96 | +27.18 | +10.7% | 255 | Passed holdout read 2 (design version); live forward test |
| USE-NBA usage model | 68-66 | +4.66 | 528-501 | +4.09 | +0.4% | 1029 | Failed its tests in every market (holdout or design); shown by choice |
| L3 rushing and receiving yards usage models | 93-79 | +4.10 | 262-223 | +11.92 | +2.5% | 485 | Failed design test; shown by choice |
| College basketball first-half totals near the market (AGREE1) | 10-6 | +3.18 | 10-6 | +3.18 | +19.9% | 16 | Shadow of the first-half model from Nov 4; read at 300 close picks, or Mar 15 |
| QB1 passing yards | 1-0 | +1.12 | 4-9 | -4.85 | -37.3% | 13 | Experimental: untested model |
| L4 passing yards usage model | 11-10 | -0.15 | 28-31 | -5.99 | -10.2% | 59 | Failed design test; shown by choice |
| Fair Odds: books above Kalshi (FP1-X) | 59-76 | -1.90 | 107-142 | -14.50 | -5.8% | 249 | Forward test from Oct 24; read at 300 bets |
| +EV scanner, no long shots (EV1-XC) | 1-3 | -1.94 | 1-3 | -1.94 | -48.5% | 4 | Shadow of EV1-X from Nov 6; read at 300 bets |
| Following sharp Polymarket wallets (PM2) | 1-3 | -2.13 | 7-9 | -4.38 | -27.4% | 16 | Forward test from Oct 26; read at 300 sharp bets from 5+ wallets, or Feb 15 |
| Fair Odds: books above Kalshi (FP1-X) | 24-24 | -3.21 | 37-57 | -20.12 | -21.4% | 94 | Forward test from Oct 24; read at 300 bets |
| Fair Odds: books above Kalshi (FP1-X) | 1-5 | -4.00 | 6-5 | +1.22 | +11.1% | 11 | Forward test from Oct 24; read at 300 bets |
| +EV scanner, power de-vig (EV1-XP) | 4-8 | -4.94 | 19-10 | +8.77 | +30.2% | 29 | Shadow of EV1-X from Oct 24; read with it |
| College football picks not shown on the site | 15-19 | -5.16 | 161-174 | -2.81 | -0.8% | 336 | Logged, not shown on the site |
| +EV scanner (EV1-X) | 29-68-1 | -5.44 | 63-191 | +7.62 | +3.0% | 255 | Forward test from Oct 24; read at 500 bets |
| Arb Under leg (LEG1) | 39-43 | -5.51 | 39-43 | -5.51 | -6.7% | 82 | Forward test from Nov 6, after passing its held-back 2025-26 season; read at 300 bets, or Mar 15 |
| +EV scanner, no long shots (EV1-XC) | 7-13-1 | -7.50 | 7-13 | -7.50 | -35.7% | 21 | Shadow of EV1-X from Nov 6; read at 300 bets |
| +EV scanner (EV1-X) | 4-11 | -7.94 | 22-18 | +7.32 | +18.3% | 40 | Forward test from Oct 24; read at 500 bets |
| College football picks shown on the site | 22-34 | -11.63 | 88-117 | -32.78 | -16.0% | 205 | Live forward test |
| +EV scanner, power de-vig (EV1-XP) | 20-30-1 | -12.38 | 38-56 | -20.97 | -22.1% | 95 | Shadow of EV1-X from Oct 24; read with it |
| +EV scanner, no long shots (EV1-XC) | 22-37 | -15.05 | 22-37 | -15.05 | -25.5% | 59 | Stopped by the user on Nov 9 after losing; every pick it made stays in this record |
| Fair Odds: books above Kalshi (FP1-X) | 14-28 | -15.15 | 23-38 | -11.32 | -18.6% | 61 | Stopped by the user on Nov 9 after losing; every pick it made stays in this record |
| Fair Odds: books above the exchanges (FP1) | 1-19 | -17.80 | 12-30 | -11.46 | -27.3% | 42 | Forward test from Oct 23; read at 300 bets |
| +EV scanner (EV1-X) | 52-85 | -19.83 | 106-206 | -67.65 | -21.7% | 312 | Forward test from Oct 24; read at 500 bets |
| +EV scanner, power de-vig (EV1-XP) | 46-77 | -26.56 | 65-101 | -25.90 | -15.6% | 166 | Stopped by the user on Nov 9 after losing; every pick it made stays in this record |
| Parlay Builder suggestions (PB1) | 18-58 | -32.41 | 33-97 | -47.50 | -36.5% | 130 | Forward test from Oct 28; read at 300 parlays, or Feb 15 |
| +EV scanner, power de-vig (EV1-P) | 44-82 | -44.60 | 131-145 | -4.37 | -1.6% | 276 | Shadow of EV1 from Oct 24; read with it |
| +EV scanner (EV1) | 32-95 | -61.89 | 115-171 | -7.78 | -2.7% | 286 | Forward test from Oct 23; read at 500 bets |
| +EV scanner (EV1-X) | 58-200 | -94.66 | 92-252 | -79.06 | -23.0% | 344 | Stopped by the user on Nov 9 after losing; every pick it made stays in this record |
Best models this week
The top of the table is led by a +EV scanner, no long shots (EV1-XC) run at +26.07 units, 32-11. It has only 43 graded bets, well under the 150 threshold we use for a small sample, and it is labeled to be read at 300 bets. Treat it as an early result, not proof. Next came +EV scanner, power de-vig (EV1-XP) at +18.85 units on a 56-48 week, and L1 rushing and receiving unders at +12.55 units. The EV1-XP run has 214 graded bets and a season return of +8.3%, so it is past the small-sample line, but its status says to read it alongside EV1-X. The college basketball first-half model (+8.93 units this week, +31.28 for the season on 216 graded) and the L2 receptions usage model (+8.45 this week, +27.18 for the season on 255 graded) both have decent samples and positive seasons. That is encouraging, and it still isn't a guarantee.
Worst models this week
The worst result was the stopped +EV scanner (EV1-X) run at -94.66 units on 58-200, with a season mark of -79.06 units. Its status notes it was stopped by the user on Nov 9 after losing, and every pick stays in the record. The +EV scanner (EV1) lost 61.89 units at 32-95 this week and is down 7.78 for the season. The +EV scanner, power de-vig (EV1-P) lost 44.60 units at 44-82. Parlay Builder suggestions (PB1) went 18-58 for -32.41 units and sits at -47.50 for the season on 130 graded. That is below 150 and still small, but the direction is bad. Among the Fair Odds runs, FP1 (books above the exchanges) went 1-19 for -17.80 units, but on only 42 season bets.
What the sample sizes mean
Many rows here have fewer than 150 graded bets, including EV1-XC (43, 4, 21 and 59 across its runs), PM2 (16) and QB1 passing yards (13). Their results, good or bad, say very little. Even the larger samples are labeled forward tests. USE-NBA usage model is flat at +0.4% over 1,029 graded, and its status says it failed its tests in every market. It is shown by choice. The college football picks shown on the site are at -16.0% for the season on 205 graded, a live forward test that has not worked so far. A good week or a bad week does not prove a model works or doesn't.
Questions
Did Mako Bets models win or lose this week?
They lost. All models combined were down 313.86 units over the seven days since 2026-11-03.
Which model had the best week?
The +EV scanner, no long shots (EV1-XC) run went 32-11 for +26.07 units, but it has only 43 graded bets. That is too small a sample to judge.
When is a model's record big enough to trust?
We flag samples below 150 graded bets as too small. Many models also list their own read point, such as 300 or 500 bets.