Do chase favourites make money?

Pascal, the Lab's eternally hopeful punter, has a soft spot for the winter game, and he has a theory to go with it. Fences find out the pretenders. A hurdle can be brushed through and a Flat race can be stolen from the front, but a steeplechase asks a horse to meet a run of big, fixed obstacles at racing speed, one after another, with tiring legs in the closing stages. A horse can't bluff its way through that. So by the time a chaser has built a proper record over fences, Pascal reasons, the market has seen it jump, seen it stay and seen it cope when the pressure came on. The favourite in a chase is the horse that has already sat the exam and passed. Back that one in every race over fences, he says, and you are backing a horse that has proved it can do the job.

There's a real kernel here, and we'd rather start with it than sneer at it. Jumping form is information. A horse that has gone round cleanly over fences has shown something a recruit fresh from hurdling hasn't, and the betting market pays close attention to it. Favourites over fences win a good share of their races. Nothing about Pascal's starting point is silly.

What we wanted to know is whether that information is still worth anything once it has been turned into a price. So we ran the blunt version. We took every steeplechase in our database of real British racing, 3,876 races, and backed the favourite in all of them: one flat stake, settled at the official Starting Price. Hurdle races and bumpers stayed out, because this page is about fences alone. And every horse that fell, unseated its rider, got brought down or pulled up was counted as exactly what it is to the punter holding the ticket: a losing bet.

The chase favourite has a better story behind it than most favourites, and the story is why people trust it. Over fences you can see the evidence with your own eyes. You watch a horse meet its fences on a good stride, stand off at the open ditch, land running and keep finding more up the straight, and you feel you've learned something solid about it. Flat form arrives as times and distances. Chase form arrives as a picture, and pictures stick.

Then there's the experience Pascal leans on. Chasers are usually older, more exposed horses, many of them with a hurdling career behind them before they ever saw a fence. By the time one heads the market, punters have had whole seasons to watch it run. All that public form makes backing the favourite feel like reading out a verdict that's already been reached. The jumps favourite page tested the wider version of this feeling across every kind of jump race, and the chase favourite is its purest form.

The win column helps too. The favourite wins 34% of these races, roughly one in three, so the reward turns up often enough to keep the faith. Each winner confirms the theory in the punter's mind.

The losers get filed somewhere else. When a well-backed chaser comes down four out while travelling well, nobody blames the method. They blame the fence, the ground, the jockey or the loose horse that cut across. A faller feels like an accident that happened to a sound selection, and accidents don't seem to count against a theory. Pascal remembers every chase favourite that jumped like a stag and came home clear. The one that turned over at the first is, in his memory, a separate event that tells him nothing. That habit is what this test exists to correct, because the bookmaker settles both bets on the same sheet.

Pascal is right that fences find things out. Where he goes wrong is in assuming they only find out the pretenders.

Start with the price. Everything that makes the chase favourite look trustworthy is public: the clean rounds, the completed races, the way it travelled last time, the trip it stays. The market has had the same seasons of evidence the punter has, and it had them first. A horse with a proven record over fences is priced as a horse with a proven record over fences. The jumping form Pascal relies on is real information, and the price has already spent it.

Then comes the bookmaker's margin. Add up the chances implied by every price in a chase and the total comes to more than a whole. That surplus is the overround, the house cut, and it is taken from every bet you place, winner or loser. So the favourite is priced roughly right for its chance and then shortened a little further to pay the book. Roughly right minus the margin is a loss that arrives slowly and arrives every season.

The third part belongs to fences. A chase favourite can be the best horse in the race and still not finish it. It can misjudge a fence and fall, make a mistake that sends its jockey out of the saddle, be brought down by a faller in front, or be pulled up when the race goes wrong. Every one of those is a losing bet for the person holding the ticket, with the full stake gone. The market knows chasers don't always get round and prices that risk in too, but the punter still carries every faller in full, and a short price leaves very little room to absorb them. Jumping is what makes a chaser look reliable, and jumping is also what can end its race at any fence.

The filter is simple, and we'll state it exactly. A race qualifies if it is a steeplechase, a race run over fences, in our database of real British racing. Hurdle races are left out, and so are National Hunt Flat races, the bumpers run without any obstacles at all. Our jumps favourite experiment puts all three types together; this one isolates fences. That leaves 3,876 chases out of the 30,221 real British races we hold. Nothing within that is picked. Handicaps and level-weights races, small fields and big ones all stay in, because the theory is about every race over fences and so is the test.

In every qualifying race we backed the favourite with one flat stake. Each bet is settled at the official industry Starting Price. Non-runners are voided, so a horse that never set off costs nothing. Where two or more horses share favouritism, the stake is split between them, so a tie can't quietly slip a winner into the count. A race needs a recorded winner to be included at all.

Then the rule that matters most over fences. Fallers, unseated riders, horses brought down and horses pulled up are all counted as losing bets, with the full stake gone. That is how a bookmaker settles them, so that is how we settle them. Leaving them out would make the figure look kinder, and it would also make it wrong. A chase is a jumping test, and the races where the favourite fails the test belong in the ledger as much as the races where it passes.

Starting Price is the generous measure, too. It assumes no commission, no early-price shading and no account restrictions. A real punter taking real prices does a little worse than the number you're about to read, so if our test errs at all, it errs in the bet's favour. Nothing was tuned to make the answer look better or worse.

Here is the number. Backing the favourite in every steeplechase returns -9.24% across 3,876 races, flat stakes to Starting Price. For every £100 staked, about £90.76 comes back over the long run. The favourite wins 34% of these races, roughly one in three, which is a healthy strike rate and exactly why the bet feels sound. The 95% range on the result is [-13.4,-5.7]. Every figure inside that range is a loss, so this isn't an unlucky patch you could sit and wait out.

Now the comparison everyone wants, and we'll handle it with care. Over hurdles the favourite returns -7.3%. Across all jump races together it returns -8.1%, and across every race on the board, Flat and jumps alike, -9.0%. It's tempting to read a story into the gaps, the bigger obstacles costing the favourite more. The data can't carry that story. The chase sample is a slice of the jumps sample, so its range is wider, and the ranges on these figures overlap. The honest reading is that the favourite over fences sits in the same neighbourhood as favourites elsewhere, and any difference you can see is within the noise. The hurdle favourite page runs the same test over the smaller obstacles if you want to set the two side by side.

What the number does show is the fate of Pascal's best piece of evidence. Chase favourites carry seasons of public jumping form, about as much as any horse in a betting market, and the result lands roughly where favourites usually land. The record over fences is in the price, the margin sits on top, and the fallers are counted. Put together, that gives a steady loss that looks very like every other favourite's steady loss, and that resemblance is the finding.

These figures live-pull from the current data. Every jumps season adds chases, the range tightens, and this page moves with it.

Chase favourites don't make money on any evidence we hold. Across 3,876 steeplechases the favourite returns -9.2%, and the whole 95% range, [-13.4,-5.7], sits below zero. That is a firm answer to the yes-or-no question. How the chase favourite compares with the favourite over hurdles is a softer answer, and we've kept it soft on purpose, because the ranges overlap and the gap could be chance.

Pascal deserves his credit. Fences do find things out, and a horse with a clean record over them has shown something real. The favourite wins its share, and the theory starts from a true observation about jumping. Where it goes wrong is in thinking the punter gets that information for free. The market saw the same rounds of jumping, priced them and added its margin. Then the fences kept their old habit of putting any horse on the floor, favourite or outsider, and the stake goes with it when they do.

The useful lesson travels well beyond chasing. Evidence that everyone can see is evidence the price has already used. A proven jumper at a fair price minus the margin is the same losing bet as any other fair price minus the margin. And a faller goes into the ledger as a loss, whatever it gets filed under in the memory of the punter who backed it.

None of this is a reason to stop enjoying winter jumping. It is some of the best racing there is to watch, and a bet on it can be worth its cost as entertainment. If you do bet, keep to a flat stake you can afford to lose. Past performance is not a guide to future returns, our Starting Price settlement with no commission flatters the figure a little, and no staking plan can turn a negative edge into a positive one, however carefully it is laid on top.