Expected Value of Each-Way Bets: The Maths Behind Profitable Place Betting
Available in US
Expected Value Is the Only Metric That Matters Long-Term
I spent my first three years of serious horse racing betting chasing winners. My strike rate was respectable, my selections were sound, and I still lost money. The problem was not my handicapping – it was my ignorance of expected value. I was backing horses at prices that did not compensate for the probability of losing, and no amount of skill at reading form can overcome that mathematical deficit.
Expected value – EV for short – is the average amount you expect to win or lose per bet if you placed it infinitely many times. A positive expected value (+EV) means the bet is profitable over the long run. A negative expected value (-EV) means it is not, regardless of how many individual winners you pick. A typical bookmaker overround for an average horse race sits between 110% and 125%, which means the average bet in the average race is -EV by design. The bookmaker’s margin ensures that the collective pool of bets generates a loss for bettors in aggregate. Beating that margin on individual bets – finding +EV spots – is the entire game for serious place bettors.
The EV Formula Applied to Each-Way Bets
The EV formula for a simple win bet is straightforward: EV = (probability of winning x profit if you win) – (probability of losing x stake lost). If a horse has a 20% chance of winning and you back it at 6/1 for £10, the EV is: (0.20 x £60) – (0.80 x £10) = £12 – £8 = +£4. That bet has a positive expected value of £4 per play.
Each-way bets are more complex because they have three possible outcomes, each contributing to the overall EV. Outcome one: the horse wins (both win and place legs pay). Outcome two: the horse places but does not win (only the place leg pays; the win leg is lost). Outcome three: the horse finishes outside the places (both legs lose).
The each-way EV formula combines all three: EV = (P(win) x total profit if win) + (P(place only) x profit on place minus win stake lost) – (P(unplaced) x total stake). You need three inputs: the probability of winning, the probability of placing but not winning, and the probability of finishing outside the places. Favourites in British horse racing win approximately 30 to 35% of the time, but the place probability is always higher – a horse with a 30% win probability might have a 55% to 65% place probability depending on the field size and the number of paid places.
Let me work through the formula with a concrete example. A horse at 10/1 each-way, £5 each way (£10 total stake), in a 12-runner race paying three places at 1/4 odds. You estimate the horse has a 10% chance of winning and a 30% chance of placing (which includes the win probability, so the chance of placing but not winning is 20%).
If the horse wins: win leg pays £5 x 10 = £50 profit. Place leg pays £5 x 2.5 = £12.50 profit. Total profit: £62.50. If the horse places but does not win: win leg loses £5. Place leg pays £5 x 2.5 = £12.50 profit. Net on this outcome: £7.50 profit. If the horse is unplaced: total loss of £10.
EV = (0.10 x £62.50) + (0.20 x £7.50) – (0.70 x £10) = £6.25 + £1.50 – £7.00 = +£0.75. This each-way bet has a positive expected value of 75p per play. Over 100 identical bets, you would expect to profit roughly £75. That is a +EV selection.
Worked Example: Finding a +EV Each-Way Selection
Theory is useful; application is what pays the bills. Here is how I identify +EV each-way bets in practice.
Step one: estimate the true win probability. I do this by comparing the horse’s form against the field, adjusting for conditions (going, distance, draw), and cross-referencing with tissue prices from multiple bookmakers. If three independent bookmakers price a horse at 8/1, 9/1, and 10/1, the consensus implied win probability is roughly 10 to 12%. I take the midpoint as my starting estimate – say 11%.
Step two: estimate the place probability. This is harder than the win probability because it depends on field size, number of paid places, and the competitive profile of the field. As a rule of thumb, place probability is roughly 2.5 to 3 times the win probability in a race with three paid places and 12+ runners. For my 11% win probability horse, I estimate a place probability of around 30%.
Step three: calculate the EV. The horse is available at 12/1 each-way with 1/4 place terms. For a £5 each-way bet (£10 total): if it wins, total profit = £5 x 12 + £5 x 3 = £75. If it places only, net = £5 x 3 – £5 = £10. If unplaced, loss = £10. EV = (0.11 x £75) + (0.19 x £10) – (0.70 x £10) = £8.25 + £1.90 – £7.00 = +£3.15. That is a strongly +EV each-way bet – the price is generous relative to my probability estimates.
Step four: sanity-check. Is my probability estimate plausible? Am I overestimating the horse’s chance because I like it? Have I accounted for the overround in the market? I revisit the exchange place market as a cross-reference: if the exchange place odds imply a 28% place probability and my estimate is 30%, I am close enough to trust the figure. If the exchange implies 20% and I am at 30%, something is off and I need to reconsider.
EV Limitations: Sample Size, Variance, and Real-World Noise
Expected value is a long-run concept, and the long run in horse racing betting is genuinely long. Variance – the natural fluctuation between expected and actual results – dominates the short term. A +EV strategy can lose money for weeks or even months before the edge manifests in the results.
I typically need 200 to 300 bets at similar odds ranges before I can statistically distinguish a +EV strategy from random noise. Below that threshold, a winning streak might be luck and a losing streak might obscure genuine value. This is why I track every bet, record the estimated probabilities and the actual outcomes, and review the data quarterly rather than daily. Short-term results are meaningless for evaluating an EV-based approach.
The other limitation is estimation error. The entire EV calculation rests on your probability estimates, and those estimates are inherently imprecise. A 2% error in your win probability estimate can flip a bet from +EV to -EV. The antidote is humility: I build in a margin of error by only betting when the EV is clearly positive, not marginally so. If the calculated EV is +£0.20 on a £10 stake, the margin of error swamps the edge and I pass. If the EV is +£3 or more, the edge is robust enough to withstand estimation imprecision.
Discipline, patience, and honest self-assessment are the non-mathematical ingredients that make an EV-based approach to place betting strategy work in practice. The formula gives you the framework; your character determines whether you can stick to it long enough for the numbers to prove themselves.
How many bets do I need to place before expected value smooths out variance?
As a practical guideline, 200 to 300 bets at similar odds ranges are needed before you can reliably distinguish a +EV strategy from random fluctuation. Below that sample size, short-term winning or losing streaks are likely driven by variance rather than genuine edge. Tracking every bet with estimated probabilities and reviewing the data over quarters rather than weeks gives you the clearest picture of whether your approach is genuinely profitable.
Can the place part of an each-way bet be +EV even when the win part is -EV?
Yes, absolutely. The win and place legs of an each-way bet are independent bets with separate expected values. A horse at 10/1 might be -EV to win (your estimated win probability is lower than the implied probability at those odds) but +EV to place (your estimated place probability is higher than the implied place probability at the each-way fraction). In that scenario, a standalone place bet or an exchange place bet would be the more efficient choice, since the each-way structure forces you to take the -EV win leg alongside the +EV place leg.
Created by the ”win Place bet Horse Racing” editorial team.
