Flat staking ignores the size of your edge. Betting 2% every time means the same exposure on a coin-flip as on a strong +EV bet. Kelly Criterion fixes this — it scales your stake to your edge. This guide shows you exactly how to calculate it, use the KiqIQ Kelly Calculator, and apply it to your betting workflow.
The Kelly Formula
f* = (bp − q) / b
f*
Kelly fraction — the percentage of your bank to bet
b
Net decimal odds minus 1. At odds of 2.50, b = 1.50
p
Your estimated probability of winning
q
1 − p (probability of losing)
Three bets at different edge levels. See how Kelly responds.
Example 1 — Strong Edge
+21% EVBet
Over 2.5 goals
Odds
1.75
Your probability (Poisson)
59%
b = 1.75 − 1
0.75
bp = 0.75 × 0.59
0.4425
q = 1 − 0.59
0.41
f* = (0.4425 − 0.41) / 0.75 = 0.0325 / 0.75 = 4.3%
Full Kelly: 4.3% of bank. Quarter Kelly: 1.1% of bank → £5.50 on a £500 bank
Example 2 — Marginal Edge
+4% EVBet
Man City Win
Odds
1.40
Your probability
74%
b = 1.40 − 1
0.40
bp = 0.40 × 0.74
0.296
q = 1 − 0.74
0.26
f* = (0.296 − 0.26) / 0.40 = 0.036 / 0.40 = 9.0%
Full Kelly: 9.0% — risky at short odds. Quarter Kelly: 2.25% → £11.25 on a £500 bank
Note: short-price favourites can recommend large Kelly stakes. Always cap at quarter Kelly.
Example 3 — No Edge (Negative EV)
−5% EVBet
BTTS Yes
Odds
1.80
Your probability
51%
b = 1.80 − 1
0.80
bp = 0.80 × 0.51
0.408
q = 1 − 0.51
0.49
f* = (0.408 − 0.49) / 0.80 = −0.082 / 0.80 = −10.3%
Negative Kelly = do not bet. You have no edge on this selection.
This is the most powerful Kelly output — it prevents you from placing bets where the maths is against you.
The recommended approach for most bettors is quarter Kelly — here is why.
| Strategy | Stake (4.3% Kelly) | Growth rate | Variance | Best for |
|---|---|---|---|---|
| Full Kelly | 4.3% of bank | Maximum | Extreme — 50%+ drawdowns common | Perfect model confidence only |
| Half Kelly | 2.15% | 75% of max | High but manageable | Experienced model bettors |
| Quarter Kelly | 1.07% | ~56% of max | Low — recommended | Most bettors |
| Eighth Kelly | 0.54% | ~36% of max | Very low | Conservative or new bettors |
| Flat 1% | 1% | Variable | Low but ignores edge size | Beginners without a model |
Recommendation: Use quarter Kelly as your default. You will retain more than half the long-term growth advantage while protecting your bankroll from the variance that causes most bettors to abandon a profitable system.
The calculator does the maths. Your job is to supply an accurate probability estimate.
Get your probability from Poisson
Open the Poisson Calculator, enter each team's home/away xG average, and note the probability for the market you want to bet (Over 2.5, BTTS, or match result). This becomes your "p" input.
→ Poisson CalculatorFind the best odds via line shopping
Use 2–3 bookmakers or an odds comparison site to find the best available price. Higher odds means lower implied probability — making it easier for your edge to be positive.
Open the Kelly Calculator
Enter the decimal odds and your Poisson probability. The calculator outputs the full Kelly percentage and shows quarter and half Kelly recommendations automatically.
→ Kelly CalculatorApply quarter Kelly to your current bank
Multiply quarter Kelly by your current bankroll total (not your starting total — Kelly should always use your current bank). If your bank has grown, your stakes grow with it. If it has shrunk after a losing run, your stakes shrink automatically.
Log the bet with opening odds
Record the bet in your tracker with the odds at time of placement. You will need this later to calculate Closing Line Value — the best long-term indicator of whether your probability model is accurate.
→ Bet Tracking GuideUsing gut-feel probability
Kelly only works if your probability estimate is accurate. A Poisson model from xG data is dramatically more reliable than "I think this team wins 70% of the time." Garbage in, garbage out.
Using your starting bank, not your current bank
Kelly should always be calculated against your current bankroll. If you started with £500 and are now at £600, use £600. Kelly scales dynamically with your bankroll.
Using full Kelly without accounting for model error
Your probability estimate will sometimes be wrong. Full Kelly bet sizing on an incorrect estimate risks significant drawdown. Quarter Kelly provides a buffer for model uncertainty.
Applying Kelly to every bet regardless of market type
Kelly assumes your probability is accurate. In high-margin markets (correct score, first goalscorer), your model is less reliable. Stick to xG-modelable, lower-margin markets for Kelly sizing.
What is the Kelly Criterion?
The Kelly Criterion is a mathematical formula for sizing bets to maximise long-term bankroll growth. The formula — f* = (bp − q) / b — tells you what fraction of your bankroll to stake based on your edge over the bookmaker.
Should I use full Kelly or fractional Kelly?
Quarter Kelly (25% of the full Kelly recommendation) is the standard for most sports bettors. It reduces variance substantially while retaining the majority of the long-term growth advantage. Full Kelly is theoretically optimal but produces drawdowns that are psychologically very difficult to manage.
What happens if my probability estimate is wrong?
If you overestimate your probability, Kelly recommends a stake that is too large. This is the key risk — the formula is only as good as your probability estimate. Using a Poisson model from xG data rather than gut feel significantly reduces this risk. Fractional Kelly provides additional protection.
Can I use Kelly for accumulators?
Yes. Enter the combined acca odds and your combined true probability (multiply each leg's probability together). Always use quarter Kelly or less for accas due to higher variance and compounding probability uncertainty across legs.
How to Research a Football Bet
The complete pre-match workflow ending in Kelly stake sizing.
Expected Value Explained
Understand EV before sizing — the two concepts work together.
Variance in Betting
Why your Kelly-sized +EV bets still lose sometimes — and how many to expect.
Track Betting Performance
Log your stakes and measure CLV to validate your probability model.
Accumulator Value with Data
How to apply Kelly to multi-leg bets built with Poisson.
Kelly Criterion Explained
The underlying mathematics and theory behind the formula.
