Receipts: Mako Bets had a winning week, but one big model lost money
Four of five models were up for the 7 days since 2026-10-20, and the combined result was +5.74 units. The biggest sample in the book finished in the red.
The week in one line
All models combined finished +5.74 units over the 7 days since 2026-10-20. That is a winning week. But the total hides a split. Four models made money. The largest one by volume, USE-NBA, lost. A good week is not proof of an edge. Most of these samples are small.
| Model | This week | Week units | Season W-L | Season units | Season return | Graded | Status |
|---|---|---|---|---|---|---|---|
| L3 rushing and receiving yards usage models | 7-3 | +3.32 | 7-3 | +3.32 | +33.2% | 10 | Failed design test; shown by choice |
| L2 receptions usage model | 5-1 | +2.96 | 5-1 | +2.96 | +49.4% | 6 | Passed holdout read 2 (design version); live forward test |
| L1 rushing and receiving unders | 3-2 | +2.25 | 24-21 | +8.95 | +19.9% | 45 | Forward test in progress |
| L4 passing yards usage model | 2-0 | +1.77 | 2-0 | +1.77 | +88.5% | 2 | Failed design test; shown by choice |
| USE-NBA usage model | 309-293 | -4.57 | 309-293 | -4.57 | -0.8% | 602 | Failed its tests in every market (holdout or design); shown by choice |
Best models this week
L3 rushing and receiving yards usage models made the most units, at +3.32 on a 7-3 record. L2 receptions usage model was close behind at +2.96 on 5-1. Neither sample is big enough to judge. L3 has 10 graded picks and L2 has 6, well below the 150-pick line we use for a small sample. Also, L3 failed its design test and is shown by choice, so treat its hot start with caution. L4 passing yards usage model went 2-0 for +1.77 units. Two picks tell you nothing.
The worst model this week
USE-NBA usage model was the only loser, at -4.57 units on a 309-293 record. That is also its season line, so it has no winning stretch to offset the week. Its 602 graded picks are a real sample, and the season return is -0.8%. It failed its tests in every market, and we show it anyway. Read it as a model with no demonstrated edge.
The one with a real sample
L1 rushing and receiving unders went 3-2 this week for +2.25 units. For the season it is 24-21, +8.95 units, a +19.9% return over 45 graded picks. That is a good start, but 45 picks is still under the 150 threshold. The model is in a forward test, which means the result is not settled. A few losing weeks could erase much of that profit.
Questions
Did Mako Bets models win this week?
Yes, combined they were up 5.74 units for the 7 days since 2026-10-20. Four of the five models made money and USE-NBA lost 4.57 units.
Which Mako Bets model has the biggest sample?
USE-NBA usage model, with 602 graded picks and a 309-293 record. It is down 4.57 units, a -0.8% return.
Are these results enough to bet on?
Not on their own. Every model except USE-NBA has fewer than 150 graded picks, which we treat as too small to judge.