I’m a tennis bettor. Specifically, I’m a underdogs men's tennis bettor. In my more than 4,000 selections since 2016, my average price (always using Pinnacle’s odds) has been 2.600.

Over the years, I have discovered that my betting results in grand slam matches were much worse than my results in other tournaments.

I decided to investigate this imbalance to see if there were any significant differences between how I bet on grand slams (featuring best-of-five set matches) and how I bet on ATP 250, ATP 500, and Masters 1000 tournaments (featuring best-of-three set matches).

Is there some kind of bias, making it more difficult to be successful when betting on the outliers in the grand slams? Is there any reason why my underdog selections perform worse in the grand slams than in the other tournaments?

To make this analysis, I gathered the Pinnacle closing prices of all the main-draw matches played since 2013.

After cleaning the data due to incorrect and missing odds and taking out withdrawals and retirements when the first set was not completed, I came up with a total of 4,885 fixtures in grand slams and 19,356 fixtures in non-grand slam events. Challengers, ITF matches, and Davis Cup games were excluded.

Grand slams

Should you have bet blindly with a stake of one unit on all the underdogs and all the favourites in grand slams, these would be the results you would have achieved:

If you had backed all of the favourites over the last 10 years, you would have finished virtually flat (-2.5 units), with a -0.05% yield. On the contrary, should you have bet on all the underdogs, the loss would have been remarkable, -680 units and a -13.92% yield.

Non-grand slam events

I also conducted the same exercise for all the ATP 250, ATP 500, and Masters 1000 matches, with the following outcome:

You would have lost 571.6 units (-2.95% yield) if you had bet on all of the favourites and -533.3 units ( -2.76% yield) if you had backed all of the underdogs.

Analysis of the results

At first sight, we can extract the clear conclusion that betting on the underdogs in grand slams is a disaster.

The first reason that comes to my mind to explain such a difference is the favourite-longshot bias. The favourite-longshot bias is the tendency to undervalue favourites and overvalue “longshots” in sports betting.

On average, the odds offered by the bookmakers at the higher prices have a higher margin and vice versa.

Moreover, from certain levels, the higher the odds the higher the margin. This means that, for a bettor, the likelihood of winning when betting on longshots is lower, as a bigger margin for the bookmaker translates into a lower expected yield.

However, why can’t we see such a gap when analysing the results of underdogs and favourites in the non-grand slam matches? In fact, in non-grand slam matches, the difference between backing the favourites and backing the underdogs was tiny, a -2.95% vs. a -2.76% yield.

In order to have more information, I have specifically analysed the favourite-longshot bias in both sets of data.

I have divided the dataset of grand slam fixtures into five groups of 977 bets, ordered by odds. Furthermore, I have taken the same odds ranges to form five groups of bets for the non-grand slam fixtures.

This way, we can compare both groups. Although the sample size is nearly four times higher for non-grand slams (NGS), we can see clearly that the favourite-longshot bias is more pronounced in grand slams (GS).

In the first range of the highest-priced players (8.710 to 81.000), we’ve got a -33.6% yield in GS vs. a -2.0% in NGS. Within this group, the number of bets is relatively low (874 in NGS) and only a few longshots can change the overall picture. This -2.0% yield in non-grand slam matches will likely be more negative as the sample size increases.

However, the difference with the grand slams group, where the yield is -33.6%, is too wide.

A low sample size is clearly not the only factor that can explain the difference. Apart from that, we can see that, for every range, the negative yield is higher in the grand slam matches.

Also, the magnitude of the differences is so high in some ranges that there is no need to make any statistically significant tests to address the issue of the different sample sizes.

Up to now, we already knew that betting on the underdogs during grand slams (-13.92%) was a worse idea than betting on the underdogs in the non-grand slams (-2.76%).

Favourite-longshot bias is higher during grand slams

But now we also know that the favourite-longshot bias is much higher during grand slams than in the non-grand slams.

For all of the ranges, if you had bet on the underdogs you would have obtained worse results. And this difference is the largest at the highest prices.

This is the same as saying that bookmakers apply, for the same level of odds, higher margins in the longshots of grand slams than in the longshots of non-grand slams. Why is there a difference?

If they take their decision based on risk aversion and price the underdogs disproportionately shorter relative to their fair prices than the favourites, why are they more risk averse in grand slams than in non-grand slams?

Could it be that, when grand slams are played, there is a higher number and proportion of recreational bettors who don’t bet on other tournaments and who are more prone to betting on the longshots?

Could it be that, to cope with this increased risk, some bookmakers might decide to apply higher margins on the longshots, knowing that these bettors are not very sensible when it comes to high prices?

That is, they don’t care much if the price is 16.000 or 13.000, as their demand elasticity with respect to underdogs is very low. Could it also be that punters are not fully assessing the influence of the “best-of-five sets” factor, where the underdogs are far less likely to win?

Conclusion

In summary, if you are betting on the grand slams, I suggest you choose your underdogs wisely. The odds are totally against you.

Choose your underdogs wisely...

Moreover, the higher the price, the greater the bookmaker margin will often be and, therefore, you’ll be less likely to find bets with positive Expected Value. This is the favourite-longshot bias and is, therefore, applicable to all sports and events.

However, we’ve seen that in tennis, it is much more prevalent in grand slam matches than in non-grand slam matches. It is more difficult to make money betting on the outliers in the grand slams than doing so in in the any other tournament.

We have speculated what might be behind this higher favourite-longshot bias in grand slams vs. non-grand slams.

Some of the reasons might be a lower demand elasticity to underdogs in grand slams than in non-grand slams by bettors, bettors not fully evaluating the impact of the best-of-five sets factor, and/or bookmakers being more risk averse on the longshots due to an increased risk.

Check out more tennis betting advice from Nishikori at @nishikoripicks. Be sure to explore our Betting Resources hub for previews and predictions of the biggest tennis tournaments. Sign up at Pinnacle and experience great tennis odds with low margins.