1 unit flat at the best US price we logged. The same bets, one by one, are on the betting test.
BetLosing beyond lucksince Oct 13
Anytime goal Yes at the best US price from +400 to +600 when our goal model (g60-1) gives an edge of 15% or more at that price; 1 unit flat. Locked about 75 minutes before puck drop.
- Record
- 10–87
- Profit
- −39.45u (−40.7%)
- Likely range
- −75.9% to −5.4%
- Hit / expected (goal model)
- 10.3% / 21.2%
Toward the pre-registered read97 of 300
One read at 300 bets: passes only with profit above zero, a 95% lower bound above −5%, and still positive without DraftKings prices.
Research before the test: 2025-26 backtest, edge 15%+ (T-15): +25.2% ROI on 221 (CI −5% to +58%; −9.2% without DraftKings-best bets)
Anytime goal scorer track record →Every bet →
BetToo early: noisesince Oct 11
When one team is Fast and the other Slow (lineup speed index from prior-season EDGE bursts 20+ mph/60), back the Fast team on the moneyline, sized by the M2 speed coefficient.
- Record
- 3–9
- Profit
- −6.01u (−50.1%)
- Likely range
- −100.0% to +1.0%
- Hit / expected (model)
- 25.0% / 54.7%
Read at the end of the regular season (n ≥ 150 and a market-null p ≤ 0.10 to promote).
Research before the test: 2024-26 Fast vs Slow (closing ML): +3.2% ROI on 562 (won 59.6% vs 56.1% fair; ROI CI −3.8% to +10.5%)
Research behind it →Every bet →
BetToo early: noisesince Oct 11
Solve the structural margin model (M1) to the market no-vig moneyline and total. If M1 makes a ±1.5 side at least 3 points more likely than the no-vig consensus, bet that side (almost always the underdog +1.5).
- Record
- 4–6
- Profit
- −2.40u (−24.0%)
- Likely range
- −85.0% to +37.0%
- Hit / expected (model)
- 40.0% / 54.7%
Read at the end of the regular season. Nearly every bet is an underdog +1.5, so one bad week of blowouts moves it a lot.
Research before the test: 2024-25 discovery: +3.9% ROI on 302; 2025-26 holdout: +4.7% ROI on 179; Pooled win rate: on 481 (59.0% (Wilson 54.6–63.3) vs 56.6% break-even)
Research behind it →Every bet →
BetToo early: noisesince Oct 11
ML favourite played the previous calendar day and the underdog did not → underdog +1.5.
- Record
- 2–1
- Profit
- +0.17u (+5.6%)
- Likely range
- −100.0% to +111.3%
- Hit / expected (model)
- 66.7% / 66.5%
Rare: about one game in fifteen qualifies. Read at the end of the regular season.
Research before the test: 2024-25 discovery: +0.5% ROI on 96; 2025-26 holdout: +11.4% ROI on 81 (BH q = 0.07)
Research behind it →Every bet →
BetToo early: noisesince Oct 16
One team played the previous calendar day and its opponent did not → the rested team’s moneyline. Primary grade: games where the tired team’s starter did not start the day before.
- Record
- 5–3
- Profit
- +1.53u (+19.1%)
- Likely range
- −55.3% to +93.5%
- Hit / expected (model)
- 62.5% / -
Primary grade is the tired team’s starter being fresh (graded after the game). Read at the end of the regular season; harm stop if the 95% range is below zero at a monthly read with 100+ bets.
Research before the test: 2024-26 fade vs rested opponent, median close: +4.5% ROI on 530 (tired team won 40.6% vs 44.7% priced; ROI 90% −1.8% to +11.0%); 2024-26 any opponent (as first tested): +2.6% ROI on 764 (ROI 90% −3.2% to +8.5%)
Research behind it →Every bet →
BetToo early: noisesince Oct 16
At the props lock (~75 min before puck drop), when a same-team regular is out and Ascension projects the player into his L1/L2 spot, and the player’s usual EV rate is under 0.45 points a game: points and shots over, 1 unit flat at the best US price. Our model edge is ignored (it does not know about promotions).
- Record
- 23–23
- Profit
- +3.63u (+7.9%)
- Likely range
- −24.4% to +40.2%
- Hit / expected (market no-vig)
- 50.0% / 43.1%
Toward the pre-registered read46 of 300
Points and shots overs for the same player are bet together, so the bets are not independent and the real range is wider than shown. One read at 300 bets.
Research before the test: 2026-27 exploratory, ex-post promotions (points): +11.6% ROI on 57 (overs 52.6% vs 45.4% no-vig; z +1.1; pooled t +0.74); Same, looser definition (all markets pooled): on 106 (over − market −1.7 pts (t −0.53))
Research behind it →Every bet →
