The story
Here is a number worth carrying to the next big handicap you bet.
Take the same horses, at the same prices, in the same races. Back them each-way where the bookmaker pays four places, and you lose. Back them where the bookmaker pays six, and you are in front. Nothing else changes. Not the horse, not the price, not the race. Only how far down the field the place half of the bet gets paid.
The swing between those two is more than twenty percentage points of return.
Most punters shop for price. In a twenty-eight-runner handicap, our data says they should be shopping for places, and it is not close. That is the finding on this page, it is useful on any Saturday of the year, and it is the part the old version of this experiment got right.
The part it got wrong was the conclusion it drew. It read "places transform the bet" as "so this bet wins", and told readers to hold out for six places and get stuck in. Recomputed against our own race record, that does not hold. At the place terms bookmakers are actually obliged to offer, the bet loses, and the positive that appears once the extra places are switched on rests on a sample too small to bank on.
So the honest version is more interesting than either of those. The lever is real and bigger than anyone expects. What sits at the end of it is a bet that gets cheaper rather than one that starts paying.
Why everyone's heard it doesn't work
The cynical case is worth stating properly, because the recompute ends up agreeing with most of it.
You pay the margin twice. An each-way bet is two bets, a win bet and a place bet, and the bookmaker's margin sits on both halves. The place half is the one doing the damage, because the place market is usually the worse-priced of the two.
The place fraction is thin. A quarter of the odds sounds generous until you work out what it buys. On a 10/1 shot the place half returns 5/2. For that to be worth having, the horse needs to finish in the frame considerably more often than one time in three and a half, and in a big field it does not.
The measured returns are negative. Flat-stake each-way betting across all races is a well-known loser, and our own numbers on these three races agree with that starting point rather than contradicting it.
The rule was written for the median race. Eight to twelve runners, three places, a short favourite. In that race the cynics are simply correct.
So the question was never whether each-way is a good bet in general. It is whether the big-field handicap is a real exception or a comfortable story, and until now this page had not tested it.
The catch the cynics miss
Two things really do change in a twenty-eight-runner handicap, and both move the punter's way.
The winners come at long prices. Across the nine renewals in our record, of the Wokingham, the Royal Hunt Cup and the Ascot Stakes, the favourite won twice. Seven times the race went to something else, often at a price. The market is not confused about this, it is that a big handicap is an open race, and the folk wisdom is right that the winning prices skew long.
The place terms widen, and this is the big one. In a small field you are looking at a fifth of the odds for three places. In a big handicap the standard is a quarter of the odds for four, and in race week the firms compete on top of that with five, six and seven. Going from three places to six does not improve the bet slightly. It roughly doubles the number of finishing positions that pay you, on a bet where the place half is most of what you are buying.
That second point is where the value on this page lives, and it is worth pausing on, because it inverts the habit most bettors have. Shopping around for a better price on a 16/1 shot might move you to 18/1. Shopping around for an extra place moves the entire place half of your bet. In a field this size the second is worth far more than the first, and almost nobody bets that way.
The distinction the old page missed. The terms a bookmaker must offer come from a published schedule: a quarter of the odds for four places in a handicap of sixteen runners or more. Five, six and seven places are extra-place specials. They go up on race morning, vary by firm, and come down when the firm chooses. They are a promotion, not a property of the race.
That matters, because the whole argument rests on it. If the exception only exists at six places, then it is something a bookmaker hands out and can take back, rather than something structural about big fields. Testing it properly means testing the guaranteed terms and the promotional ones separately, which is exactly what the old page did not do.
How we tested it
The three races are the Wokingham Stakes, the Royal Hunt Cup and the Ascot Stakes, all heritage handicaps with fields of twenty or more. Our record holds nine renewals, three of each, from the Royal Ascots of 2024, 2025 and 2026. The old page claimed the last ten renewals of each. That was never a sample this site possessed.
The bet is the one the page has always described: back every runner priced at 10/1 or bigger, each-way. One unit on the win, one on the place, two units staked per horse. That comes to 189 bets across the nine races. It stakes far more than any real punter would, which is the point, because it measures the terms themselves rather than anyone's horse-picking.
Every bet is settled at industry starting price with no best odds guaranteed and no early prices. The place half returns a quarter of the odds when the horse finishes inside the places.
The test then runs three times over the same 189 bets, changing only the number of places paid:
- Four places, the terms the published schedule guarantees for a handicap of this size.
- Five places, and six places, the extra-place specials that get advertised in race week.
Running one population through three sets of terms isolates the thing in dispute. If the big-field exception is real, it should survive at the guaranteed terms. If it only appears once the promotions are switched on, then what the old page called a structural edge was a bookmaker's marketing budget.
As with the companion favourite study, every return carries a 95 per cent interval, and the intervals are the reason this page no longer recommends anything.
What the data showed
The same 189 bets, settled three times, changing nothing but the number of places paid.
| Place terms | Bets | Return | 95 per cent interval |
|---|---|---|---|
| Four places, guaranteed | 189 | -14.15% | -55.0% to 26.7% |
| Five places, promotional | 189 | +3.17% | -38.8% to 45.1% |
| Six places, promotional | 189 | +7.41% | -34.6% to 49.5% |
The lever is real, and it is enormous
Start with what that table proves, because it is the useful part and it does not depend on the sample size at all.
Every row is the same 189 bets. The same horses, the same starting prices, the same nine races, the same finishing order. The only thing that moves between the top row and the bottom is how deep the bookmaker pays, and the return moves from -14.15% to +7.41%.
Two extra places are worth more than twenty percentage points on identical bets.
That is not a claim about whether the bet wins. It is arithmetic on a fixed set of results, and it is the single most actionable thing in either of these two studies. If you are betting each-way in a big field, the place count is the term to shop for. A better price is worth a fraction of what a deeper place is worth, and the industry advertises the prices far louder than the places.
What sits at the end of the lever
Now the part the old page overreached on.
At four places, the terms you are actually entitled to in a handicap this size, the bet returned -14.15%. It loses, at roughly the rate the old "each-way is a mug's bet" line would predict. The exception this page was built on does not exist at the guaranteed terms.
It appears once the extra places go on, and there the honest answer is that we cannot yet tell. The six-place return of +7.41% carries an interval of -34.6% to 49.5%. On 189 bets with six winners between them, that is what an interval always looks like when a handful of long-priced results carry the column. The number is positive and worth watching. It is not yet evidence.
Where the positive actually comes from
Splitting by race, at the most generous terms tested, is instructive.
| Race, six places | Bets | Return |
|---|---|---|
| Wokingham Stakes | 68 | -8.64% |
| Royal Hunt Cup | 74 | -23.65% |
| Ascot Stakes | 47 | +79.52% |
The two sprints the old page led with both lose even at six places. Everything positive in the headline comes from the Ascot Stakes, which is a two-and-a-half-mile stayers' handicap and the least like the race the argument was built around. Three runnings, 47 bets, and an interval from -28.5% to 187.6%.
Three runnings of one race is not a finding about that race either. It is where the variance landed, and it is a good illustration of why the by-race split is the first place to look when a headline number seems too good.
What we cannot check
Our record starts in October 2023 and holds three renewals of each race. The decade of Wokingham winning prices the old page carried, a named favourite from 2019, and a count of thirty renewals between 2015 and 2024 are not in any file we hold, so they have been retired rather than repeated. Of the nine renewals we do have, the favourite won two, and neither was a Wokingham.
The verdict
The thing to actually use
In a big-field handicap, shop for places before you shop for price.
That is the finding, it comes straight out of the table, and it holds regardless of whether the bet ends up winning. The same 189 bets returned -14.15% at four places and +7.41% at six. Nothing changed but how deep the bookmaker paid. No amount of price-shopping on a 16/1 chance gets you anywhere near a swing that size.
Most betting advice has this the wrong way round, because prices are what firms advertise and places are buried in the terms. In a twenty-eight-runner handicap the places are the headline and the price is the footnote.
The honest answer on whether it pays
At the terms you are entitled to, no. A quarter of the odds for four places is what the published schedule guarantees in a handicap of sixteen runners or more, and at those terms our nine renewals returned -14.15%. The old cynicism about each-way survives this test rather than failing it.
At extra-place terms, we cannot yet tell. Six places lifts the same bets to +7.41% on an interval of -34.6% to 49.5%, and almost all of it comes from one race in three runnings. That is a number to keep watching, not one to bet on. It is also worth being clear about what it would mean even if it held up: it would be a promotion making a losing bet cheaper, handed out at the bookmaker's discretion, rather than a flaw in how they price a race.
So the verdict is mixed rather than flat. The mechanism the old page identified is real and larger than it claimed. The conclusion it drew from it went further than the evidence, and further than our own data now supports.
If you are betting these races anyway
Check the place count before you check anything else, and know which part of it you are entitled to. Four places is the floor. Anything beyond that is a promotion worth hunting, because on our numbers it is the difference between a bet that loses clearly and one that loses very little.
That is a way to bet these races less expensively. It is not a way to beat them, and this page no longer suggests otherwise.
Why the page changed
This used to be the only experiment in the Lab with a verdict of "works", and it was the one built on the least of our own data. It cited ten renewals of each race when our record holds three, and it treated a race-morning promotion as though it were the standing rule. Recomputed from our own database, the mechanism survived and the recommendation did not. The full recompute and every retired figure are published as a dataset beside the page.
Nothing here is a tip and nothing here is an edge. The model's own track record stands at −13.9% across 10,609 races and regenerates nightly. This experiment does not use model output at all. It measures place terms.
The companion study, Does backing the favourite work at Royal Ascot?, came out of the same recompute with a finding of its own: Royal Ascot favourites win at exactly the rate favourites win at everywhere else.
Responsible betting reminder. Every figure here measures a small sample of the past, not the future. Set a limit before you bet. Free, confidential help is at BeGambleAware.org.