Backing Horse Racing Favourites: 16,325 Bets Reveal the Cost of Betting Blindly

There is something deceptively appealing about backing the second or third favourite in a horse race. The favourite might be too short to offer an attractive return, while the outsiders appear too unpredictable. Somewhere in the middle, surely, there must be value. That assumption sounds reasonable until it is tested against thousands of actual bets.

During October 2026, the Trade Carefully Research Lab did precisely that, and the results offer a revealing lesson about the difference between finding winners and finding an advantage.

Our automated horse racing experiments recorded 16,325 settled selections across three market positions: the favourite, second favourite and third favourite. These observations covered racing in Australia, the United States, Britain, France, Ireland, New Zealand and South Africa. Rather than selecting individual horses based on form or opinion, we systematically recorded what happened when horses occupying these market positions were backed at their matched exchange prices.

The purpose was not to make money from every experiment. We deliberately use very small stakes because the primary objective is to collect real-world data, test hypotheses and learn how markets behave. A losing experiment is not automatically a failed research exercise. It can be extremely valuable if it helps eliminate an unproductive approach or reveals where further investigation is justified.

October gave us plenty to investigate.

The favourite collector recorded 5,304 settled selections and 1,991 winners, producing a strike rate of 37.54%. After allowing for 2% commission on winning profits, it finished 4.60 units ahead, a return on turnover of just 0.09%. In practical terms, backing favourites indiscriminately was approximately break-even.

The second-favourite experiment told a less encouraging story. Across 5,768 selections, it recorded 1,207 winners, a strike rate of 20.93%, but lost 181.24 units. That represented a negative return of 3.14%. The third-favourite collector performed worse still, recording 744 winners from 5,253 selections and losing 185.52 units, equivalent to a negative return of 3.53%.

Combined, backing the second and third favourites produced 11,021 selections and a loss of 366.77 units. Those figures are normalised to one-unit stakes, not the actual cash lost during our small-stakes research. The distinction matters: we are measuring the performance of the ideas consistently, regardless of the experimental stake sizes.

But the headline losses are only the beginning of the story. When we examined the results by country, some fascinating differences appeared.

Australia: The Most Striking Contrast

Australia provided the largest national sample in all three experiments, making its results particularly interesting.

Backing Australian favourites produced 808 winners from 1,986 selections and finished 78.51 units ahead, a positive return of 3.95%. Yet backing Australian second favourites lost 107.95 units across 2,212 selections, while third favourites lost a remarkable 191.28 units across 1,999 selections.

That third-favourite result represents a negative return of 9.57%. Only 250 of those 1,999 selections won, a strike rate of 12.51%.

The contrast is difficult to ignore. In the same country's racing markets, backing favourites produced a positive October result while moving just two positions down the market produced substantial losses.

Does this establish that Australian favourites are consistently underpriced or that third favourites are systematically poor value? Absolutely not. One month cannot establish either proposition. But it does give us a specific research question to pursue. Rather than treating every horse racing market as interchangeable, we can now investigate whether the relative performance of market positions differs meaningfully between countries.

United States: A Different Pattern Emerges

The American results were noticeably different.

The favourite collector recorded 1,671 selections and finished almost exactly break-even, losing just 2.31 units, a return of −0.14%.

But the second-favourite collector produced 407 winners from 1,753 selections and finished 29.01 units ahead, a positive return of 1.65%. Third favourites, meanwhile, lost 42.81 units across 1,638 selections, a negative return of 2.61%.

That gives us an interesting contrast with Australia. In October, Australian second favourites performed poorly, while American second favourites delivered a modest positive return.

Our recorded winners included Summerstateofmind at Charles Town, matched at odds of 4.00, and Lebombo at Remington Park, matched at 3.45. At Charles Town, Quality Of Essence was also recorded as a winning second-favourite selection at 7.20. These individual examples are not evidence of an advantage, but they illustrate the range of winning prices contributing to the American result.

The important question is whether the difference between Australia and the United States reflects something persistent about their racing markets or simply the natural variation of a relatively short observation period. We do not yet know. What we do know is that combining the two countries into a single global figure would conceal this distinction.

Ireland: A Surprising Result That Needs Scrutiny

Ireland produced one of the most eye-catching findings in the entire dataset.

The third-favourite collector recorded 34 winners from 223 Irish selections and finished 70.56 units ahead, representing a remarkable 31.64% return on turnover.

That is an extraordinary contrast with Australia, where third favourites lost 9.57%, and France, where they lost 18.51%.

However, the Irish sample is much smaller. With only 223 selections and a 15.25% strike rate, a relatively small number of winners at substantial prices could have a disproportionate influence on the overall result. Before drawing conclusions, we need to examine the distribution of winning prices, the contribution of individual races and whether the result persists as the sample grows.

Ireland also produced a negative result for second favourites, which lost 36.03 units across 247 selections. Its favourite collector, meanwhile, finished 9.06 units ahead from 227 selections.

This is precisely why we collect data rather than rush to conclusions. A 31.64% return is exciting to observe, but it would be irresponsible to present it as a discovered edge without substantially more evidence.

France and New Zealand: Further Reasons to Question Blanket Approaches

France delivered losses across all three market positions. Favourites lost 46.39 units from 316 selections, second favourites lost 18.12 units from 345, and third favourites lost 58.11 units from 314.

That translates into negative returns of 14.68%, 5.25% and 18.51% respectively. In other words, none of the three broad approaches worked in the French sample during October.

New Zealand showed another uneven pattern. Favourites lost 21.79 units across 270 selections, and second favourites lost 51.07 units across 295. Yet third favourites managed a small positive result of 8.35 units from 250 selections.

Again, these are not findings that justify immediate changes to a betting strategy. They are signals that the broad global averages may be concealing important differences. Country, race type, field size, odds range and liquidity could all influence the outcomes, and those possibilities need to be examined carefully.

What the Data Is Really Telling Us

One conclusion is becoming increasingly difficult to avoid: blindly backing horses across countries, race types and market conditions is not a convincing route to a sustainable betting advantage.

Our October evidence does not prove that every indiscriminate betting approach must lose money. The favourite collector was approximately break-even, and some individual countries and market positions recorded positive results. But the losses from the second- and third-favourite experiments demonstrate the danger of assuming that a seemingly sensible market position automatically offers value.

At Southwell, our third-favourite collector recorded Tinkety Tonk as a winner at odds of 8.00. At Fontwell, Glance Of Glory won at 4.10. At Wolverhampton, the second-favourite collector recorded Decem Carolinae winning at 5.10. These are the sorts of selections that can make an approach appear attractive when viewed individually.

Yet the complete record tells a different story. Thousands of unsuccessful selections consumed more units than the winners generated. It is easy to celebrate an eight-to-one winner. It is much harder, and far more useful, to account for every losing bet that preceded or followed it.

That is why we are not building the Research Lab around attractive screenshots, selected winning races or hypothetical profits. We are building it around recorded outcomes, actual matched prices and consistent performance measurement.

The small stakes are essential to that approach. They allow us to run numerous experiments, including those we suspect may ultimately prove unprofitable, without confusing the research objective with the pursuit of immediate income. The value we hope to create is in the accumulated evidence and the understanding that emerges from it.

There is also an important warning about analysing countries retrospectively. Once we have examined the results, it is tempting to discard Australia from the third-favourite experiment, keep Ireland and declare the problem solved. That would be a serious methodological mistake. We would simply be selecting the best-performing historical subset after already knowing the outcome.

A genuine advantage needs to survive further observation. If Ireland continues to perform unusually well, or Australian third favourites continue to underperform, we can begin investigating why. Until then, those results are hypotheses to test, not strategies to sell.

The Research Continues

October's 16,325 horse racing selections provide a useful starting point, but they are not the complete history. We began collecting wider Research Lab observations around 23 September 2026, and those earlier records will be incorporated into our cumulative dataset.

From November onwards, our analysis will increasingly focus on results since the experiments began, rather than treating each month as an isolated exercise. We will examine changes in performance, the distribution of winning prices, differences between countries and the impact of individual high-priced winners.

We will also investigate whether apparent patterns persist as thousands of additional observations accumulate.

The most valuable outcome may ultimately be identifying which broad approaches should be abandoned, which deserve more focused testing and which apparent opportunities disappear under closer scrutiny.

For October, the evidence is clear. The favourite collector was almost break-even. The second and third favourites collectively lost 366.77 units. Australia showed a particularly severe deterioration further down the betting market, while American second favourites and Irish third favourites offered contrasting positive results.

Those differences make the research more interesting, but they do not yet establish reliable advantages.

The data does not care how persuasive a betting idea sounds. It records what happened. Our responsibility is to follow that evidence, including when it contradicts our assumptions.

That is the point of the Trade Carefully Research Lab.

Research note: This analysis covers October 2026 live research observations across seven countries. All returns are standardised to level one-unit stakes using actual matched exchange prices, with 2% commission deducted from winning profits. Actual experimental stakes are deliberately small. Country-level results are exploratory and should not be interpreted as proven profitable strategies. Past performance does not guarantee future results. Betting involves financial risk.