Value and arbitrage

Traders talk about "edge" as if it were one thing. It is really two, and they ask almost opposite skills of you: arbitrage takes a margin already sitting in the prices in front of you, while value takes a price your own assessment says is too big. Here is what each demands, and what each can cost.

Arbitrage: a margin already in the prices

Arbitrage means covering every outcome of a market, across whichever prices you can reach, at odds that together imply less than 100% of the probability. Size the legs correctly and the same amount comes back whichever result lands; the margin is settled the moment both bets match.

The test takes seconds. Convert each price to its implied probability by dividing 1 by the odds, then add those figures up. Below 1, a margin exists. At or above 1, there is nothing to take.

A worked example

Take a two-outcome market where the best price on one side is 2.04 at one exchange and the best price on the other is 2.00 at another. 1 ÷ 2.04 = 0.4902 and 1 ÷ 2.00 = 0.5000, so the two add up to 0.9902 — 99.02%, below 1, so a margin of just under 1% is on offer.

To split a stake so the return is identical either way, each outcome takes total stake × (1 ÷ odds) ÷ the total, and the return on any outcome is total stake ÷ the total. With £202 committed: 202 ÷ 0.9902 = £204, and the stakes are £204 × 0.4902 = £100 and £204 × 0.5000 = £102.

Both rows return the same £204 against £202 laid out — a margin of £2, or 0.99% of the money committed. All the figures on this page are before commission. Nothing here depends on there being two outcomes: add up 1 ÷ odds across however many selections you cover and the same two formulas apply, which is the arithmetic behind Dutching as well.

Why this is not risk-free

Arbitrage is often described as risk-free. It is not, and the example shows why: the margin is £2, so almost anything that goes wrong costs more than the margin is worth.

• Commission. Exchanges charge it on net winnings, and a sub-1% raw margin can disappear into it entirely. Rates differ — the exchange guides carry the published figures.
• An unmatched leg. Each leg is a real bet that has to match. One matched and one resting leaves you holding a one-sided position on a sporting event.
• A price that moves first. The second bet goes on after the first, and in those seconds the price you were counting on can shorten or vanish.
• Voided or rescheduled events. One leg voided while the other stands leaves the remaining bet running uncovered.
• Differing rules. Two operators can settle the same event differently — non-runners, dead heats, abandonment — so legs meant to cancel do not.
• Account standing. Operators may restrict or close accounts, and some dislike this activity.

The event stops being your risk; execution, settlement and account standing become it instead.

Value: a price your own estimate says is too big

Value never produces a certain result on a single bet. The implied probability of a price is again 1 ÷ odds, and value exists when your own assessed probability is higher than that figure.

A selection at 3.50 implies 1 ÷ 3.50 = 28.6%. If your assessment puts it at 33%, the price is paying more than you believe the outcome is worth: 0.33 × 3.50 = 1.155, more coming back than going out, on average. The phrase "on average" is doing every bit of the work there.

Variance, and what a short run tells you

A 33% shot loses two times in three when your estimate is exactly right, so over a long series, runs of ten or more consecutive losers are ordinary rather than evidence of anything. Value is a claim about long-run averages; a handful of bets tells you close to nothing about whether the assessment behind them was sound.

The estimate is the entire edge, and it can simply be wrong. Nothing confirms your 33% — the price is the market's own estimate, made by many people with money at stake, and it is often better than yours. The only check is your record over a long period: slow, noisy and honest.

What BetTrading.net does, and what it does not

BetTrading.net is execution, not selection. It shows the prices at the exchanges you have connected, works out stakes, routes and places bets quickly and consistently, and keeps track of what you hold — which matters when a margin is thin and both legs must land at the prices you saw.

It doesn't find edges. It doesn't assess probabilities, rate selections, tell you what to bet on, or scan for arbitrage opportunities on your behalf. Where an edge comes from stays your work; the application removes the friction between deciding and having the bet on.