Do favourites win novice and maiden races?

Pascal, the Lab's eternally hopeful punter, has a soft spot for the races most people skip in the form guide. Novice and maiden races are a staple of British racing, on the Flat and over jumps, and on paper they look like guesswork: a field of horses that have never won, some running for the first time, a few with one quiet outing behind them. The form book has very little to say about any of them. That, Pascal argues, is exactly the point.

When the public knows nothing, the only people who know anything are the ones who watch these horses work every morning. So watch the money. If one horse in a field of beginners is attracting all the support, the yard must have seen something at home, and the market is simply passing the message on. His rule fits on a beer mat: follow the confidence.

There is more to this than wishful thinking. Novice and maiden races are often uneven contests. A well-bred horse from a powerful stable can line up against rivals who are still learning the job, and the gap between them can be wide. Stables do know their own horses better than anyone, and the betting does tend to find its way to the ones they rate. If any favourite was going to carry inside knowledge onto the track, you would expect it to be this one.

So we tested it without arguing about it. We took every race in our database with novice or maiden in the title, Flat and jumps alike, which sweeps in maiden hurdles and novice handicaps as well as the plain novice and maiden events. That gave us 7,799 races. In every one we backed the favourite, one flat stake, settled at Starting Price, and counted the result honestly, beaten favourites and jumping fallers included.

Much of the appeal is how often it lands. The favourite wins 46% of these races, which is a high figure for a favourite: across every race in our database the everyday favourite wins 35% of them. A punter following the novice and maiden rule spends a lot of afternoons collecting, and a steady run of collects feels like evidence. It feels as though the method is reading something real.

The rest of the appeal is the story about the yard. Racing loves the idea of the stable whisper, the horse that was "expected" after a good piece of work at home. In a race full of unraced or lightly raced horses that story is at its most believable, because there really is less public information to go on. Liam, the stable loyalist on Pascal's Panel, follows his yard through thick and thin on much the same instinct: the people closest to the horse must know best. In a maiden, that instinct seems to have the field to itself.

Memory does its share of the work too. A well-backed newcomer that wins easily is the kind of race people talk about for weeks, because it looks like the moment a secret came out. The heavily backed debutant that ran green and finished fifth is forgotten by the next race on the card. So the evidence a punter carries home is tilted towards the wins, and the wins in these races come often enough to keep the tilt steady.

Pascal's version is that the confidence is the edge. Everyone else is guessing, he says, and the money is the one clue that isn't a guess. He is right that the money is a clue. Where he goes wrong is in assuming he is the only one reading it, and the next section is about what that assumption costs.

The trouble with following the confidence is that the confidence is the price. When one horse in a maiden is all the money, that money does not sit quietly waiting for Pascal to notice it. It shortens the horse. By the time the race goes off, whatever the yard knew, or whatever the market believed the yard knew, has already been turned into odds. A punter backing the favourite at Starting Price is buying the stable's opinion after it has been fully priced.

Pascal is right that these races are uneven, and the market can see that as clearly as he can. A strong favourite against a field of beginners goes off short, often very short, and a short price pays little when it wins. That is how a higher strike rate buys smaller returns. Winning close to half the time sounds wonderful until you look at what each win pays.

Then comes the margin. Add up the chances implied by every price in a race and they come to more than a hundred per cent. The surplus is the bookmaker's overround, and it is charged on every bet, winners and losers alike. A favourite priced almost exactly right, minus that margin, loses slowly and steadily, and nothing about a novice race switches the margin off.

The unknowns also cut both ways. The same lack of form that makes the favourite look like the only informed bet means the rest of the field is a mystery as well, and a newcomer with no public form is the hardest horse for anyone to price. Some of them turn out better than the market thought, and every one that does beats a favourite. Over jumps there is an extra cost. Novice races are, by definition, full of horses still learning to jump at racing pace, and a favourite that falls or unseats its rider is a losing bet, the full stake gone, exactly as if it had been beaten.

The rule is simple to state and simple to check. A race qualifies if its title contains the word novice or the word maiden. That is how the code picks the races out, by reading the race title rather than any hidden category, so the honest description is races with novice or maiden in the title. Flat and jumps both count. Novice hurdles, novice chases, Flat maidens and maiden hurdles are all in, and so are novice handicaps, because the word is in the title even though the race is run off the handicapper's weights. A race for inexperienced horses that carries neither word in its name is left out.

That filter leaves 7,799 races out of the 30,221 real British races in our database. In every one we backed the favourite: one flat stake, official industry Starting Price, non-runners voided, joint-favourites split, and a recorded winner required for the race to count. Over jumps, fallers, unseated riders and pulled-up horses are counted as the losing bets they are, because your stake is gone whether the horse is beaten a length or hits the deck at the last. It is the same settlement every experiment on this board uses, so the figures line up from one page to the next.

Starting Price is the kind figure. It assumes you got the official return with no commission, no early-price shading and no account restrictions, so a real punter taking real prices does a little worse than the number in the next section.

On sample size, this one stands on firm ground. Novice and maiden races turn up on cards all year round, Flat and jumps, so the test runs to thousands of races and the range around the result is reasonably tight. There is no horse selection being tested here and no staking trick. Back the shortest price in every qualifying race, stake the same each time, and count what comes back.

Here is the number. Backing the favourite in every race with novice or maiden in the title returns -6.72% across 7,799 races, so for every hundred pounds staked about £93.28 comes back over the long run. The favourite wins 46% of these races. The confidence range on the result is [-9.3,-4.5], and the whole of it sits below zero. Even the kindest reading of the data is a loss.

Now set it beside the everyday favourite. Backing the favourite in every race returns -9.0%, and that favourite wins 35% of its races. So the novice and maiden favourite wins more often and loses a little less. That is the part of Pascal's theory that survives: in these races the favourite is more reliable than favourites in general, and the leak is gentler. We'd keep the comparison modest, though. The two ranges overlap at their edges, so "a little less" is as far as the evidence goes, and we would not stake a reputation on the exact size of the gap.

For a sense of scale, odds-on favourites, the shortest-priced runners of all, return -5.2%. Short prices carry a thinner slice of the margin, and a novice or maiden favourite is often a short price, which goes some way to explaining why this is one of the gentler lines on the board. Gentler still means a loss.

The figure worth staring at is the gap between how often this rule wins and what it returns. A system that collects close to half the time and still hands money back shows how little a strike rate tells you on its own. The stable's confidence is real, and the market pays close attention to it. It pays so much attention that the confidence has been absorbed into the price before you arrive, with the margin taken on top.

Favourites do win novice and maiden races, often. Backing them in every one does not make money. Across 7,799 races the rule returns -6.7%, about £93.28 back for every hundred pounds, and the confidence range of [-9.3,-4.5] never reaches zero. It is one of the gentler leaks on the board, and it is still a leak.

Pascal deserves credit for the half he got right. These races are often uneven, and the favourite wins them more often than favourites usually do. Where the theory breaks is the step from "the yard knows" to "so I can profit from it". Once a stable's confidence shows up as money, it shows up as a shorter price, and the market has read the same clue Pascal has. What he ends up holding is the stable's opinion at a price that already reflects it, with the bookmaker's margin on top.

That lesson travels well beyond maiden races. Information only helps a punter while other people don't have it, and money in the market is public by definition. Following it means joining the crowd at the crowd's price. A high strike rate in those conditions makes for more winning afternoons, which is pleasant, but the long-run total still finishes on the wrong side of zero. Liam would recognise the feeling: loyalty to a confident yard is rewarded often enough to feel like faith repaid, and the price quietly collects on every other day.

If you enjoy these races, and there is plenty to enjoy in watching a well-bred youngster find its feet, treat any bet on them as entertainment and size it that way, with a flat stake you can afford to lose. Past performance is no promise of future returns, our figures are settled at Starting Price so the real world is a touch harsher, and no staking plan turns a negative edge into a positive one.