Working With Market Odds

Converting greyhound odds into implied probability

Decimal odds are a probability statement dressed up as a number. Implied probability is one divided by the decimal odds, so 4.00 means 25%, 2.50 means 40% and 10.00 means 10%. Once you can do that in your head, you stop seeing prices and start seeing forecasts. Convert every runner in a race and the total almost always comes to more than 100%, usually somewhere between 105% and 130%.

That excess is the overround, the bookmaker's built-in margin taken before any winners are paid. A total of around 125% is roughly normal for a six-runner race in the UK and Ireland. It is not profit on every race, but a long-run cut. A useful habit is to total the implied probabilities for each card: a figure close to 100% usually means you are looking at an exchange price rather than a bookmaker one.

Bookmaker prices versus exchange prices

A bookmaker price is set by a trader balancing liability across the book, with margin baked into every price. An exchange price is the live meeting point between people backing a dog and people laying it, with commission of typically 2 to 5% charged on winning bets instead. Bookmaker prices are stable and easy to take; exchange prices are often closer to true probability on busy races but thin on small fields and quiet meetings.

The chapter recommends using the exchange price close to the off, when liquidity is highest, as your probability benchmark, while treating bookmaker prices as what you actually stake at. Allow for commission when converting exchange odds. With UK greyhound turnover falling in real terms, prices on lower grades and midweek cards may be less efficient, so a favourite on a busy Saturday final is a more reliable number than one on a quiet Tuesday.

What market moves and steam really signal

A dog drifting from 5.00 to 7.00 before the off is being abandoned; one shortening to 3.50 is being backed. Both belong in your model, but neither should be obeyed blindly. Timing matters, since late moves say more than slow drifts over hours. Volume matters, because the same percentage move means little on a thinly traded race. Liability matters too: a bookmaker shortening a price may simply be protecting margin, so treat those moves with more scepticism.

Log the opening price, the price an hour out, ten minutes out and the starting price, because the shape of that trail tells you more than any single number. Steam moves, where several dogs on one card shorten together, often reflect someone else's plan rather than which dogs are fastest. Use them as a check on your own overconfidence, not as a direct reason to bet.

The favourite fallacy in greyhound betting

Favourites in graded UK races win about 33% of the time, which means the market's most likely winner loses roughly two races in three. Backing favourites because they feel safe is paying for a feeling rather than accuracy. The more useful point is the spread behind that average: at strong, predictable tracks favourites win nearer 40%, while at more chaotic tracks or in lower grades they win nearer 31%.

If you build your own ratings, the aim is to judge, race by race, whether the favourite's true chance sits near the top or bottom of that range. The chapter also notes that punters love long shots, which tends to leave rank outsiders over-bet. Spotting races where the favourite is weaker than its price suggests, or where an outsider has a hidden chance, is where your own estimates can add something.

When it makes sense to disagree with the market

A model that just agrees with the market and lags it adds nothing. Value comes only when your estimate and the implied probability differ by enough to overcome margin and variance. With a 125% overround, a model rating a dog at 25% against a 20% implied price may be close to break-even once margin, commission and variance are considered. The threshold is higher than most beginners assume, so write down your estimate, minimum edge and the race overround before checking prices.

Reasonable grounds for disagreement include the market pricing in crowd narratives your data contradicts, such as a kennel on a cold run; the market reacting slowly to new information like changed conditions; and situations it may weigh poorly, such as a return from a break or a first run at a venue. Disagreeing on a strong feeling is not one of them. If a hunch cannot become a variable, log it and move on.

Building market odds into your betting system

The market can serve as a baseline, where you assume it is right and bet only when your model is far enough away, or as a feature, feeding implied probability into a model alongside form and conditions. Either way it becomes a timestamped column in your database. Three habits help: log your estimate before seeing the price, compare against the exchange rather than other bookmakers, and weight late prices far more than early ones.

The market is rarely wrong by a lot, and it tends to be wrong in similar ways across thousands of races, which is what makes it a useful benchmark. The goal is not to beat it on every race but to find a small, repeatable slice of races where your structured data says something the market has not yet fully priced. Betting always carries risk, and a gap on paper is not a guaranteed return.