Conviction-proportional sizing
Bet larger on the bets you have the most reason to believe in, smaller on the rest. The principle most retail bettors invert.
Bet sizing is the highest-leverage decision in any disciplined betting strategy. Most retail bettors flat-bet, ignoring how strongly they believe in each bet. The operators that consistently outperform size proportional to conviction. The framework is straightforward; the discipline is the hard part.
The flat-betting baseline
Flat betting: every bet is the same fraction of bankroll, regardless of how strong the bet is. A 1% bettor bets 1% on every signal. The strategy has the virtue of simplicity. It also leaves money on the table when some bets carry meaningfully more edge than others.
FLAT-BETTING BANKROLL EXAMPLE
$10,000 bankroll, 1% units
every bet = $100
60-bet sample mix:
20 bets at +1% EV
20 bets at +3% EV
20 bets at +5% EV
expected return on $6,000 of action
flat: ~ +$180 (avg 3% EV)
same bets, sizes proportional to EV:
20 × $50 at +1% = +$10
20 × $100 at +3% = +$60
20 × $200 at +5% = +$200
same handle, +$270 expected returnSized to conviction, the same handle delivers more expected profit. The math compounds across hundreds of bets and hundreds of opportunities to make sizing decisions correctly.
What conviction looks like
Conviction is a quantitative read on the strength of a bet, not a feeling. It can come from:
- Modeled probability deviation from market price (the larger the deviation, the higher the conviction).
- Confluence of signals (multiple independent indicators pointing the same direction).
- Historical performance of similar bets in the operator's data.
- Inputs the market is not pricing fast enough (lineup news, weather, late information).
Conviction is not 'I have a feeling about this game.' If your conviction cannot be expressed as a probability or a structured signal, it is decoration on a guess. Treat it as such, not as a sizing input.
Bands instead of continuous sizing
Most operators do not size continuously. They use bands. Three to five bands is plenty. Each band has a defined unit size. Each bet maps to one band based on its conviction score.
| Band | Conviction | Unit size |
|---|---|---|
| NO-BET | below threshold | 0% (no bet) |
| OPENER | low end of edge | 0.5% bankroll |
| STANDARD | moderate edge | 1.0% bankroll |
| ELITE | strong edge | 1.5% bankroll |
| GEM | highest conviction | 2.0% bankroll |
Why bands instead of a continuous function of conviction? Operationally cleaner. Easier to communicate. Easier to enforce. The data on whether the bands are calibrated correctly is observable in the bettor's ledger over time.
Why retail bettors invert this
Two cognitive failures combine. First, bettors confuse comfort with conviction. The bet that feels safest (heavy favorite, popular team, primetime game) gets the biggest bet. The bet that feels riskiest (underdog, niche market, less televised game) gets the smallest. Comfort and edge are nearly uncorrelated.
Second, bettors size up on tilt. After a losing streak they bet bigger to make it back. The size is uncorrelated with the conviction; it is correlated with the streak. This is the retail bettor's most common path to bankroll destruction.
Sizing is most of the edge
Two bettors. Both pick the same bets. One sizes proportional to conviction. The other flat-bets. Over a thousand bets, the first compounds bankroll significantly faster. They had no information advantage. They had a sizing advantage.
Most casual bettors think the work is in the picks. The work is in the sizing.
What to read next
Kelly criterion is the formal version of conviction-proportional sizing. Bankroll management covers the broader context. The institutional approach explains why this matters more than anything else.