Our Methodology
Quick answer
DawBets calculates expected value by devigging sharp sportsbook odds to derive fair probabilities, then comparing those fair prices against retail book odds. Bets priced better than fair are flagged as +EV and sized using fractional Kelly criterion.
How we calculate expected value
Expected value (EV) is the mathematical edge on a bet. We derive it by comparing the offered odds at a sportsbook against our estimate of the true fair probability.
Our fair probability estimates come from devigging odds at sharp sportsbooks — books that are known for tighter lines and faster market corrections. We support four devig methods:
- Multiplicative — divides each side's implied probability by the overround total
- Additive — subtracts an equal portion of the margin from each side
- Power — finds an exponent that normalizes implied probabilities to 100%, better for lopsided markets
- Most Conservative (default) — runs all three methods and picks whichever implies the smallest edge, ensuring we never overstate the opportunity
We derive fair prices from a consensus of sharp book lines rather than relying on a single source. This reduces noise and makes our fair price estimates more robust.
Badge tier system
Every bet surfaced in the DawBets feed is assigned a badge tier based on its expected value:
| Badge | EV Threshold | Meaning |
|---|---|---|
| Must Bet | ≥ 7% EV | Exceptional edge — bet immediately before the line moves |
| Strong Bet | ≥ 3.5% EV | Clear positive expectation worth acting on |
| Good Bet | > 0% EV | Positive expected value — mathematically favorable |
| Fan Bet | ≥ -2% EV | Slightly negative but within entertainment range |
| Caution | < -2% EV | Significantly overpriced — avoid |
Kelly criterion bet sizing
For each +EV bet, we calculate a recommended wager using the Kelly criterion formula: f = p - (1-p) / b, where p is the fair win probability and b is the decimal odds minus 1.
We default to fractional Kelly to account for estimation uncertainty:
- High risk tolerance — ½ Kelly
- Medium risk tolerance (default) — ¼ Kelly
- Low risk tolerance — ⅛ Kelly
Player prop markets receive an additional 50% reduction multiplier to account for higher variance and potential model uncertainty in individual-level outcomes.
Data freshness and market coverage
Odds are sourced in real-time from 20+ US sportsbooks via licensed data feeds. Our staleness detection system flags bets where the underlying odds may have moved since the last refresh, prompting users to verify before placing.
We cover moneylines, spreads, totals, player props, and futures across NFL, NBA, MLB, NHL, NCAAF, NCAAB, MMA, and tennis.
Limitations and assumptions
- Fair price estimates are derived from market-based devigging, not predictive models. They reflect the market's current opinion, not a guaranteed true probability.
- EV is a long-run concept. Individual bets can and will lose regardless of edge size.
- Kelly sizing assumes bet outcomes are independent. Correlated bets (same game, same player) should be sized more conservatively.
- Odds can move between the time we surface a bet and the time you place it. Always verify the current price before wagering.
- Sharp book consensus is our best proxy for true probability, but sharp books are not infallible — they reflect the best available market information, not ground truth.