How we rate sportsbooks on value

The one number the ranking is built on

It is tempting to lead with the biggest welcome bonus a book will print. A bonus is claimed once, cleared against its wagering terms, and gone. The margin is different. It is the cut a bookmaker folds into its odds, known in the trade as the hold or, on an American line, the juice, and it is charged on every single bet for as long as an account stays open. Over a season it moves far more money than any sign-up offer, which is why it sits at the centre of the grade here and the bonus is treated as a tie-breaker.

That cut goes by several names, the overround and the vig among them, and whatever it is called it is straightforward to measure. Take a market, convert each price to its implied probability, add those probabilities up, and subtract 100 percent. A set of genuinely fair odds would imply exactly 100 percent between the two outcomes. Anything above that is the book's built-in edge. A market priced at 1.98 and 1.98 carries a lighter overround than one priced at 1.90 and 1.90, so it hands more back to the bettor on a winning stake.

The reason to build a ranking on this figure rather than on features is that it is the one cost a bettor cannot opt out of. A welcome offer can be taken or ignored, streaming can be watched or muted, an app can be tolerated or replaced, but the price baked into the odds is paid on every accepted bet whether the bettor notices it or not. It is the closest thing sports betting has to a price tag, it maps straight onto a bettor's expected loss over time, and a ranking that ignores it is comparing everything except the price.

How the margins are captured

The figures shown across this site are an illustrative, dated snapshot of public odds. They are captured manually, once, for a small and stable set of fixtures and market types: a handful of fixtures across a few common market kinds, read from each book's public prices at roughly the same moment. For every book, the implied probabilities across those sampled markets are summed and 100 percent is subtracted, and the results are averaged into the single margin figure you see beside each name.

The capture follows a deliberate order so the reading stays fair. The same fixtures and the same market types are read for every book, at close to the same time, before lines have had a chance to drift apart between one book and the next. Prices are taken from each book's public front end, the same odds any visitor would see, never from a private feed or a pre-match estimate. Reading the identical markets across the whole field is what makes the numbers comparable: a book only looks sharper or wider than another because it was measured on the same bet at the same moment, not because a kinder fixture was chosen for it.

Two things follow from that method, and both are stated plainly wherever a margin appears. First, the numbers are not live. They are representative editorial figures dated to when they were captured, not a real-time feed, and nothing on the site is betting advice. A book's true prices move constantly; the snapshot is a fair like-for-like reading taken at one point, useful for comparison, not a promise about the price on offer right now. Second, a book that does not publish capturable public odds for the sampled markets is left out of the ranking entirely rather than estimated or guessed. An empty guess would be worse than an honest gap.

A worked margin calculation

The method is easier to trust once you have run it yourself, so here it is on a single market. Take a football match priced at 2.10 for the home win, 3.40 for the draw and 3.80 for the away win. Convert each price to its implied probability by dividing one by the odds: 1 divided by 2.10 is 47.6 percent, 1 divided by 3.40 is 29.4 percent, and 1 divided by 3.80 is 26.3 percent. A perfectly fair market would total exactly 100 percent between those three outcomes.

Add them up and the total comes to 103.3 percent. Subtract the fair 100 percent and the margin on this market is 3.3 percent, which lands the book a B. Now compare a two-way market priced at 1.91 on each side. Each price implies 52.4 percent, the two add to 104.7 percent, and the overround is 4.7 percent, a wider B despite looking like a fairer, more even market at a glance. That is the whole point of measuring rather than eyeballing: a price of 1.91 and 1.91 feels balanced, but it hands back less than the three-way example above. The margin captured beside each book on the site is this same calculation, averaged across every sampled market, condensed into one number.

A season of betting: the bonus against the margin

The claim that the margin outweighs the bonus is easy to assert, so here it is with numbers over a full season. Take a bettor who stakes 100 a week, a little over 5,000 across a year. A book graded D on a 7.4 percent margin charges roughly that share of the turnover in expected cost, around 370 over the year, and hands over a one-off welcome offer worth, say, 50 in realistic terms once its wagering is cleared. The net cost of a season there is about 320. Now run the same turnover through a book graded A on a 2.6 percent margin. The expected cost is around 130, and there is no welcome offer at all.

The gap is more than 190 in a single season, in favour of the book that gave nothing away at the door. That is the whole case for reading the price first and the bonus last, drawn straight from the two ends of the grade scale. A welcome offer is a fixed amount collected once; the margin is a percentage of everything you stake, so the more you bet the wider the gap grows, and a heavier-turnover bettor sees it open faster still. A large bonus can close a little of the distance, but it cannot close a difference that compounds on every bet across a year. The figures here are illustrative, but the shape of the arithmetic holds at any realistic turnover, and the value calculator lets you run it on your own stakes.

From margin to grade

The margin maps to a letter grade through a fixed, published threshold. The mapping is the same for every book, applied automatically, so the scoreboard and any report drawn from the same data can never quietly disagree.

A book at or under three percent earns an A. Three to five percent is a B, five to seven percent a C, and anything wider than seven percent lands a D. The boundaries are deliberately blunt: a single measured margin decides the letter, with no discretionary bump for a book we happen to like or a promotion that looks generous. Features such as live streaming, cash out, app quality and market depth are described in each review, but they do not change the value grade. The grade answers one question only, which is the odds quality a book offers: how much it charges on the price.

The editorial rating, alongside the value grade

Each book carries two numbers, and they answer different questions. The value grade, A to D, measures one thing: the margin built into the odds, which is the price paid on every bet. The editorial rating out of five is broader, a considered read of the whole account, weighing the app and site, market depth, live streaming, withdrawal speed and reliability. The two move independently, so a book can price sharply and still be a plain product, or run a polished app on middling prices, which is why they are shown side by side rather than blended into one score. For a bettor focused on the return, the value grade leads; for one who weighs the day-to-day experience too, the rating fills in the rest. Neither replaces reading the strengths and watch-outs in each review.

How affiliate links do and do not affect the ranking

Some of the links on this site are affiliate links. When a reader follows one of those links and later signs up, this site may earn a commission from that book. That arrangement is disclosed openly, and it is kept structurally separate from the ranking. The grade is decided from the captured margin before any commercial relationship is considered, and no book can pay to rank higher, to soften a watch-out, or to have a weak price presented as anything other than what it is.

It is worth being precise about where the commercial layer touches the site and where it does not. It never touches the grade, the order of the table, or the wording of a review's strengths and watch-outs, all of which are fixed by the measured margin and the underlying record of facts. What it does affect is which links carry a tracked redirect: a book we have a commercial arrangement with is reached through a labelled, sponsored redirect, while a book we do not is reached through a plain external link. A reader can therefore see, from the link itself, whether the site stands to earn from a given click, and can check that seeing this changes nothing about where the book sits in the ranking.

The order of the table is set by measured price, best value first, and it would read the same whether a book paid a commission or not. If a commercial partner posts a wide margin, it sits low in the table with a C or a D beside its name, because that is what the number says. Keeping the method mechanical is the point: it removes the temptation to reward the books that happen to pay the most.

The unmonetized benchmark

To make that independence checkable rather than merely asserted, the ranking includes at least one book we earn nothing from. Pinnacle is carried as an unmonetized editorial benchmark. It consistently prices among the sharpest in the market and keeps some of the tightest closing lines going, so it anchors the standard the rest of the field is held to, and there is no affiliate arrangement behind its listing. Its link carries no commercial tag and is a plain external link. A ranking that only ever featured books it profits from would be hard to trust; keeping an honest, high-scoring book we make no money from is a deliberate part of the design.

How often the figures are updated

The snapshot is a dated reading, so it is refreshed on a regular cadence rather than left to age. When it is refreshed, the whole set moves together: the same fixtures and market types are read again across the field, the margins are recalculated, the grades are recomputed from those margins, and the capture date is updated in one place. Because every page draws its numbers from a single shared data source, a refresh propagates everywhere the figures appear at once, and there is no way for the scoreboard, a report and a review to drift out of step with one another.

Every page that carries a figure also shows a visible last-updated date, and that date is the honest marker of how current the reading is. If a book changes its typical pricing between captures, the site does not silently edit the old number; it waits for the next full capture so the whole field is re-measured on the same basis. Between refreshes, the standing advice is the same as it is at the top of every page: the figures are illustrative and dated, real prices move, and the tools on this site let a reader check any current price for themselves.

The limits of the method

This approach is honest about what it does not cover. A snapshot cannot capture how a book prices every market on every day, and margins genuinely vary between sports, between leagues, and between a headline event and an obscure one. The sampled set leans toward common markets, so a book that prices its mains tightly but widens sharply on niche competitions may look a little better here than a specialist would find it. Where a book has no capturable public price it carries no grade at all, shown as a dash rather than a filled-in guess. The method is meant to be a fair, transparent, reproducible read on price, published in full so anyone can see how each letter was reached, not the last word on a book.