top of page
Illustrative racing action: jockeys and horses on wet turf

8,785 Horse-Racing Favourite Bets: October Held Up. November Cost 152 Units.

25 minutes ago
6 min read

At a racecourse, the favourite is the horse the betting market regards as most likely to win. It is a reassuring word, especially when the field is large and the runners' form is hard to compare. But being the most likely winner is not the same as being a profitable selection. Trade Carefully's Favourite Horse Collector recorded 8,785 settled BACK bets with a clear win or loss between 3 October and 17 November 2026. Of those, 3,247 won and 5,538 lost. Recalculate each at one unit and deduct an assumed 2% commission from winning profits, and the collector lost 146.85 units, a return of −1.67%. What makes the story particularly revealing is the path: October finished slightly ahead, before November erased that gain many times over.

This is part of Trade Carefully's Research Lab, where small real-money collector bets are examined using their recorded selections, matched odds and outcomes. Our Live Experiments programme describes the wider approach. The normalised units here are not the original cash P/L: every losing BACK bet costs one unit, and every winner earns (decimal matched odds minus one) multiplied by 0.98. One additional workbook row carried a nonstandard SETTLED status without a win/loss marker and is excluded from these calculations.

Illustrative horse racing action, jockeys riding through muddy turf; not a photograph of a recorded collector race

The promise of the market's first choice

The first selections on 3 October illustrate the difficulty. Quddwah at ParisLongchamp, matched at 1.98 in a Group 2 race over a mile, lost. Captain Cairney at Southwell, priced at 2.68 over two miles, also lost. Arcadian Emperor at Gowran Park, priced at 2.08 in a beginners' chase, lost as well. Three different venues, three favourites, three lost stakes. The workbook confirms the races, runners, odds and settlement; it does not tell us how they travelled, whether they led, or what happened at the last obstacle. There is no need to invent a dramatic finish to understand the arithmetic.

The collector did not simply lose every early bet. Play It Cool, backed at 4.10 in a claiming race at Laurel Park on 3 October, won and earned 3.038 normalised units after commission. One result like that could offset three one-unit losses. But the same ledger that records a satisfying winner also records every favourite that never got home. That is why a collector must be assessed as a sequence, not as a selection of memorable races.

October looked sustainable. November did not.

Across 5,304 October selections, 1,991 won. The month finished 5.60 units ahead, approximately +0.11%. That was not a substantial edge, but it was a genuine positive result over thousands of bets. For a trader following the bank each day, such a near-flat month could feel reassuring: a favourite-based rule had apparently absorbed the normal losses without much damage.

November was a different experience. The 3,481 bets with clear win/loss outcomes, plus one nonstandard-status row excluded from the calculations, generated 1,256 winners and a normalised loss of 152.46 units. The November return on the eligible bets was approximately −4.38%. That single month turned a small cumulative gain into a loss of 146.85 units. It is not proof that something fundamental changed in horse racing. It shows how quickly a fragile historical result can be overwhelmed by a less favourable run.

There is a psychological trap in a result like this. After October, the collector had 5,304 opportunities to prove itself and had almost exactly broken even. A person wanting the method to work could call that resilience. After November, the same person might be tempted to say the system only needed a better filter. Both reactions risk turning the data into a story designed to protect an expectation. The bank does not care whether the runner was the favourite or whether last month looked encouraging.

The shortest odds did not provide protection

At matched odds below 1.50, the collector won 234 of 335 eligible selections, almost 70%, but lost 22.39 units (−6.68%). Those wins returned small profits, while every losing favourite cost the whole stake. The 1.50–2.00 range was also negative: 618 winners from 1,107 bets and a loss of 33.76 units (−3.05%). The market was often right about which horse would win, but that did not make the prices attractive enough.

The largest part of the sample lay further out. Among 3,206 bets at 2.00–3.00, the collector lost 7.94 units, approximately −0.25%. Among 3,673 at 3.00–5.00, it lost 43.71 units (−1.19%). Another 465 bets at 5.00 or higher lost 39.05 units (−8.40%). No major price band finished profitable in this dataset. The least damaging was 2.00–3.00, but even there the historical return was slightly negative.

These splits are useful diagnostics, not preapproved betting rules. We only know which band came closest to break-even because the results have already happened. Selecting the least bad band now and presenting it as an established advantage would confuse retrospective analysis with a forward test.

A winning favourite is not necessarily a short-priced favourite

The label also hides differences in field size and market confidence. On 4 October, Nouvel Espoir won a six-furlong handicap at the Curragh at matched odds of 9.40, returning 8.232 normalised units. Mayamar won a mile-and-a-half handicap at ParisLongchamp on 13 October at 8.80, returning 7.644 units. These were recorded by a collector named Favourite Horse, even though their matched prices were much longer than many readers would expect from a favourite. The label describes the collector's classification, not a universal price threshold or an independently verified market rank at the moment of the bet.

That distinction matters. A favourite in a small field priced at 1.30 and a horse classified by this collector as favourite at 9.40 are not economically comparable simply because they share a strategy label. To understand whether the selection logic contains an edge, we would need the precise ranking method, the market snapshot used to classify each runner and the timing of the matched price. The export supplies the bet and settlement, not the full reasoning behind the classification.

Australia was the exception — and it changed what we kept testing

One important finding is hidden by the overall −146.85-unit result. In Australia, the Favourite Horse collector recorded 3,344 settled win/loss bets, including 1,346 winners. On the same one-unit, 2%-commission basis used throughout this article, those Australian selections gained 72.53 units, approximately +2.17%. The remaining 5,441 eligible selections outside Australia lost about 219.39 units. This was not a uniformly losing collector across every region. Australia's positive contribution was substantial, even though it could not rescue the combined result.

The geographical contrast became even more consequential when we compared our other horse collectors. Australia's 3,273 second-favourite selections lost 261.31 normalised units, while its 2,674 third-favourite selections lost 248.69 units on the same calculation. The earlier research therefore led us to a deliberately asymmetric testing decision: retain Australia for the Favourite Horse collector, but exclude it from the second- and third-favourite approaches. This was a response to observed performance, not a claim that Australian favourites possess a permanent advantage. The Australian figures quoted here have been recalculated directly from the master workbook; they replace earlier provisional summaries.

There is a subtle but essential distinction between a research decision and proof. Keeping Australia in a forward test may be reasonable because the historical data raised a question worth investigating. It would not be reasonable to present the +72.53 units as guaranteed future profit, or to quietly remove the losing countries from a retrospective performance chart. The honest approach is to disclose the full original result, explain the geographic hypothesis, and judge the Australia-retained rule only on fresh selections collected after that decision.

What 8,785 bets can and cannot tell us

The scale of the sample makes the negative result worth taking seriously. It also makes the October-to-November deterioration difficult to dismiss as one unlucky afternoon. But 8,785 bets do not prove that every way of backing favourites loses money. They describe this collector's selections, timing and matched prices under a consistent commission assumption. A different rule, different prices or a different period could produce different outcomes. Any proposed refinement must be specified before it is tested.

Our Tennis Favourites investigation found a related problem: a high winning percentage was not enough to produce a profit. Our

For now, the most important conclusion has two parts. October's 5.60-unit gain was genuine, November's 152.46-unit loss was genuine, and the complete collector finished negative. But the Australian subset gained 72.53 units and gave us a specific hypothesis worth carrying into a fresh test. The task now is not to declare victory for Australian favourites. It is to see whether that geographical difference survives new data, with the selection rules and exclusions documented in advance. A favourite is the market's preferred runner, not a promise to the person holding the ticket.

Comments


Contact us

Disclaimer

Trade Carefully is platform that focusses on using data-driven approaches to build the right mindset to have any chance of success long term. Sports Trading is an extremely difficult path to follow. It requires strict discipline, patience, and can result in losses.  

 

If you struggle with gambling addition, please get the support you need from organisations such as GambleAware

2026 - TradeCarefully

Contact: support@tradecarefully.com

bottom of page