Forecasts against the market
On past NBA, NFL, NHL and Premier League seasons, no model we tested, ours included, beat the market’s prices with confidence. Our live record keeps scoring forecasts saved before each game.
EPL
Premier League
Your matchups
Pick a fixture, run the simulator, and explore the pitch in detail.
Research
What our testing shows so far. Our testing has not yet established a reliable forecasting advantage over market prices. Here is what we found, including the results that don’t favour us.
Findings so far
Price gaps
81 of 3,380 games met our price-gap filters in our 2025–26 historical study. The study did not verify that bets could be placed or profits realized. Gaps are rare. We are testing whether they last long enough to use before a game.
NHL
Our NHL forecasts currently show limited predictive evidence. NHL model calls on Scoutics carry a “Limited evidence” label.
Simulated games
Seven sportsbook models on simulated games
We ran seven sportsbook models through the same simulated games. These are synthetic results, not real prices or real bets.
-$250$0$250$500$750$1,000$1,250Cyrus+$1,002Mitra
+$745
+$745
Cyrus: block upCyrus: block downMitra (holds every ticket)Other five models
Cyrus finished first out of seven models Simulated games, not real prices: the same 848 tickets of $10, spread over 1,000 simulated games, for every model. Mitra keeps every ticket to the end; Cyrus and the five others are research models that try to cash out early. Cyrus’s edge over just holding every ticket is within the noise, and we haven’t tested it on real price history yet.
| Games | Open | High | Low | Close | Mitra close |
|---|---|---|---|---|---|
| 1–20 | 0 | 0 | -67 | -26 | 42 |
| 21–40 | -26 | 41 | -70 | 41 | 71 |
| 41–60 | 41 | 152 | 39 | 81 | 95 |
| 61–80 | 81 | 81 | 23 | 48 | 92 |
| 81–100 | 48 | 57 | 1 | 57 | 72 |
| 101–120 | 57 | 61 | -30 | 61 | 31 |
| 121–140 | 61 | 121 | 61 | 82 | 55 |
| 141–160 | 82 | 82 | 20 | 40 | -3 |
| 161–180 | 40 | 137 | 30 | 86 | 42 |
| 181–200 | 86 | 110 | 59 | 97 | 27 |
| 201–220 | 97 | 97 | 31 | 31 | -56 |
| 221–240 | 31 | 57 | -4 | 0 | -106 |
| 241–260 | 0 | 71 | -10 | 35 | -91 |
| 261–280 | 35 | 66 | 14 | 62 | -107 |
| 281–300 | 62 | 237 | 62 | 219 | 136 |
| 301–320 | 219 | 256 | 184 | 247 | 138 |
| 321–340 | 247 | 247 | 162 | 162 | 91 |
| 341–360 | 162 | 196 | 150 | 154 | 128 |
| 361–380 | 154 | 172 | 112 | 143 | 97 |
| 381–400 | 143 | 160 | 118 | 128 | 68 |
| 401–420 | 128 | 160 | 112 | 140 | 84 |
| 421–440 | 140 | 235 | 118 | 230 | 185 |
| 441–460 | 230 | 519 | 214 | 519 | 434 |
| 461–480 | 519 | 519 | 384 | 458 | 369 |
| 481–500 | 458 | 633 | 433 | 631 | 490 |
| 501–520 | 631 | 713 | 631 | 674 | 573 |
| 521–540 | 674 | 729 | 638 | 657 | 546 |
| 541–560 | 657 | 759 | 657 | 759 | 607 |
| 561–580 | 759 | 795 | 707 | 707 | 642 |
| 581–600 | 707 | 707 | 663 | 663 | 664 |
| 601–620 | 663 | 666 | 615 | 666 | 650 |
| 621–640 | 666 | 739 | 635 | 720 | 680 |
| 641–660 | 720 | 763 | 720 | 763 | 679 |
| 661–680 | 763 | 839 | 736 | 839 | 716 |
| 681–700 | 839 | 847 | 817 | 817 | 648 |
| 701–720 | 817 | 1033 | 788 | 1008 | 823 |
| 721–740 | 1008 | 1067 | 994 | 1064 | 847 |
| 741–760 | 1064 | 1076 | 1028 | 1040 | 790 |
| 761–780 | 1040 | 1121 | 1040 | 1080 | 817 |
| 781–800 | 1080 | 1115 | 1045 | 1045 | 771 |
| 801–820 | 1045 | 1045 | 935 | 946 | 649 |
| 821–840 | 946 | 965 | 888 | 950 | 634 |
| 841–860 | 950 | 1043 | 928 | 1030 | 755 |
| 861–880 | 1030 | 1105 | 993 | 1074 | 814 |
| 881–900 | 1074 | 1089 | 978 | 978 | 727 |
| 901–920 | 978 | 978 | 891 | 891 | 596 |
| 921–940 | 891 | 966 | 884 | 938 | 610 |
| 941–960 | 938 | 1011 | 938 | 972 | 671 |
| 961–980 | 972 | 1035 | 942 | 1035 | 772 |
| 981–1000 | 1035 | 1063 | 966 | 1002 | 745 |
Past results don’t guarantee future ones.
What we’re still testing
Some ideas look promising but aren’t proven. They stay labelled as research until the evidence is in, and we’ll publish the results either way.
Price gaps before games
How often a gap between sportsbooks lasts long enough to use in the hours before a game, when most people are looking. On Scoutics they are marked “Research preview”.
Sportsbook price patterns
Whether a sportsbook whose price is out of line with the rest tends to move back toward them, measured on our saved price history.
New information
Whether confirmed lineups and injuries, close to the start, are something prices take time to absorb.
Our forecasts against the market
Scored only on forecasts saved before each game. No verdict for a sport until enough games have finished (200 per sport) for the result to mean something.
Read more
Forecast recordEvery forecast saved before kickoff and scored after the final, next to the sportsbooks.Read more
Worked lineup exampleA projected lineup and a three-player change, run over the same 100 simulated games.Read more
How it worksWhere the prices come from, what each simulator covers and how the Champions League forecast was tested.Read more
Get a Second Opinion Before Game Time
Compare every sportsbook’s price, simulate matchups and check our forecast record. Start free and upgrade when you need more. Every number is a model estimate, never a guarantee or betting advice.