value betting: find the edge, then size it correctly

what value betting actually is

A bet is a price on a probability. The bookmaker offers you odds, and odds are just probability in disguise — convert any price and you get the implied probability, the chance the price is priced for. If you think the true chance of the event is higher than the probability the price implies, the bet has positive expected value. That is the whole definition. Value betting is not about finding winners; it is about finding prices that pay more than the chance deserves. Winners at skinny prices can lose you money over time, and losers at big prices can make you money over time. The result of any single bet is noise. The price relative to the chance is the signal.

This is the premise mro.tips works from — tips from the price, not the headline. A horse that "looks well" or a team that "needs the win" is not a reason to bet. A price that beats your probability estimate is.

step 1 — read the price as a probability

First, get the price into one format and read what it implies. The odds converter and value finder handles fractional, decimal and American odds in either direction, and shows the implied probability for whatever you type in. Implied probability is simply 1 ÷ decimal odds. A 2/1 shot is decimal 3.00 and implies a 33.3% chance. A 9/2 shot is 5.50 and implies 18.2%.

One check worth doing before anything else: what the market as a whole implies. Add up the implied probabilities of every runner in a race, or every outcome in a match, and you will almost always get a total above 100% — the overround, the bookmaker's margin. A football match priced home 2.10, draw 3.40, away 3.60 implies 47.6% + 29.4% + 27.8% = 104.8%, an overround of 4.8%. Every bet placed through that book pays a slice of the margin, so your edge has to beat the margin as well as the chance.

step 2 — put a number on the chance

The other side of the comparison is your own estimate of the win probability. It can come from a ratings model, from pace and form analysis, from market movements you trust, or from a simple "that price implies X, and I disagree because…". It just has to be a number, settled before you look at the price you are offered — deciding after you have seen the price is how we kid ourselves. If you cannot write down a probability and a reason, there is no bet.

For the rest of this guide we will use one running example: you rate a horse at 40% to win, and the best price you can find is 2/1 — decimal 3.00, which implies 33.3%. That gap is the edge.

step 3 — check the value, in numbers

Edge = your probability − implied probability. In the example: 40% − 33.3% = +6.7%. A positive edge means the bet is value. But edge is only half the story — the size of the edge relative to the price decides how much to stake, which is what step 4 handles.

The value check inside the odds converter does the arithmetic for you: enter your probability and the best decimal price, and it returns the edge, the Kelly fraction and the expected value per £100 staked. Expected value is the average profit per bet if the same price and probability repeated many times: EV = (probability × profit if it wins) − (probability of losing × stake). For our example, win 40% of the time at +£200 profit, lose 60% of the time at −£100: EV = (0.40 × £200) − (0.60 × £100) = £80 − £60 = +£20 per £100 staked. That is a genuinely good bet — most value bets are far thinner.

Notice what the numbers do not say. +£20 EV does not mean this horse wins, and it does not mean you are up £20 after one bet. It means that over many similar bets the arithmetic is on your side. Betting is a sequence problem, not a single-event problem — which is why the next step, sizing, matters more than most tipsters admit.

step 4 — size the stake by Kelly, then halve it

The Kelly criterion is the classic answer to "how much?". It sizes a stake in proportion to the edge divided by the price: f* = (b·p − q) / b, where b is the decimal odds minus one, p is your probability and q is 1 − p. In the example, b = 2, p = 0.40, q = 0.60, so f* = (0.80 − 0.60) / 2 = 0.10 — full Kelly says 10% of the bankroll, which is £100 on a £1,000 bankroll.

Full Kelly maximises long-run growth but is brutal in practice: it assumes your probability estimates are exactly right, and it swings violently. Most serious staking plans run half or quarter Kelly — most of the growth, a fraction of the variance. The staking calculator does this without any algebra: bankroll, decimal odds and probability in, and it returns the Kelly fraction plus the full, half and quarter stakes. It also has a points-staking mode — declare the bankroll as 100 points and it tells you the cash value of 1, 0.5 and 2 points — which is the practical way to run a tips service or a personal betting log.

Two sizing rules that survive contact with reality. First, a bet with no edge gets no stake: if the calculator shows f* at 0.00%, Kelly is telling you the price does not beat the chance — pass. Second, stakes are set by the arithmetic, not by how confident you feel. Confidence is not a probability; it is a mood.

step 5 — when one bet is not enough: dutch the value

Sometimes the value is not on one horse but spread across two or three. You might think the favourite is overrated but cannot separate the next two, or you want exposure to several live chances in the same race. That is what dutching is for: splitting a total stake across selections so the return is the same whichever one lands. The dutching calculator does the split, with optional weights, and shows the combined implied probability and the book % — what the dutch costs you.

The same value logic applies to the dutch as a whole. Say you back two horses at 4.00 and 5.00 with £100: the dutch puts £55.56 on the first and £44.44 on the second, and returns £222.22 — a £122.22 profit — whichever wins. Combined implied probability: 1/4 + 1/5 = 45%. If your combined true probability for those two runners is above 45%, the dutch is a value position, and you can size the whole thing through the staking calculator as if it were a single bet at those combined odds. If your combined estimate is below 45%, the dutch is a losing bet in fancy clothing — the calculator shows exactly where you stand.

A note of caution: dutching inside a single bookmaker's market usually means paying the overround on every leg, so the combined implied probability typically comes out well above 100% — the book % line shows the gap. Dutch when the prices justify it, not because "covering more horses feels safer". Covering more horses with no edge just spreads a losing bet thinner.

the loop

Value betting is a loop, not a trick: form the estimate, convert the price, check the edge, size by fractional Kelly, repeat hundreds of times. What separates the people who make it work from the people who do not is rarely the picking — it is the discipline around the picking: prices checked against at least two books, stakes set before the bet and never raised after, and a written reason that names the value rather than the hope. That is the checklist mro.tips runs on.

And the honest endnote: an edge is an expectation over many bets, and even a genuine edge produces losing streaks — long ones. That is precisely why the stake sizes above are so deliberately small. Bet within your means, treat the bankroll as money you can afford to lose, and if betting stops being enjoyable, stop. 18+ | BeGambleAware.