Sportpunter

With only a small profit made over the past 5 years on grass for ATP tennis, we decided to take a closer look at the analysis. The table below shows the analysis of all our grass bets over the last five years for ATP, with a description to come

Clearly this shows something that has long been suggested, in that grass is a favourites game. No wonder Sampras, Federer and a stack of pre-modern era players continually win on the green top.

Our analysis suggests that only bets of 70%+ probability have been profitable. This accounted for 13.2% ROI, with 146 of the 499 bets in total (29%). Similarly, only betting on favourites (where the odds are less than 2.0) resulted in 215 bets with 8.0% ROI. Betting on underdogs at odds of 3.0 or more, or even 4.0 or more provided very bad returns. 54 bets at odds of 4.0 or more resulted in a 50.7% ROI loss, with only a win happening only one in 8 bets.

Clearly on grass, these bets should be avoided. Simply by not betting on large outsiders (4.0 or more) increases the return from 0.1% ROI up to 5.1% ROI which is very acceptable.

I will leave it up to the individual what bets one might want to take here, but I am contemplating writing it into the program that only bets of 70% or greater probability be suggested. Keep in mind that it is a very small sample size for grass, and this could be why part of the problem occurs.

Interestingly, if one were to bet $100 on all the favourites on grass over the past 5 years, then you would have turned over $141,000 for a profit of $4,797 which equates to 3.4% ROI. Should one have bet $100 on all the underdogs, then your $141,000 would have yielded a loss of $29,291 for a -20.8% ROI.

A big difference indeed. Similarly, if one was to bet to win $100, then betting on the favourites would have put you back $718,693 for a nice profit of $20,133 or a 2.8% ROI. Betting on the underdogs to win $100 would have seen you fork out $20,133 for a loss of $9,780 or -14.7% ROI.

Seems to me to be a very big bias there. Perhaps our probabilities are generally pretty right., however it is the bias in the results that is causing such skew. It is important of course, when looking at biases in results not to think that this bias will stay for the long term. Once people start realising that grass court tennis strongly favourites the better player, then the odds will start to accommodate for this.

It may take a while, but at this stage, only betting on strong favourites seems to be most profitable.

Let us know what you think in the comments below.

Sign up to the Sportpunter Tenns Model by 31/12/09 to receive a free Sportpunter Polo shirt and free entry, along with myself, into day 1 of the Australian Tennis Open.

Last week we looked at the successful WTA 2009 season, however today we will look at how the tennis model went in the last five years since 2005 to give a better overall idea of how the model works. The data is shown below in the tables, and following this will be a good write up for discussion.

Unlike the mens analysis, all raw data without SP rules, showed good profits amongst almost all the probabilities (probabilities of 20% and below showed a loss), and all odds and overlays. Higher overlays resulted in higher returns which is a key for a good model to show.

When looking at the sportpunter rules, when the probability was between 30% and 40%, we basically broke even from 358 bets, we will look at this are more closely soon. All other probabilities showed good profits. Similarly all odds showed good profits, but also interestingly a loss was found when betting overlays between 5% and 10%. This loss was quite small (only -0.9% ROI), but with over 1000 bets it is sure to be significant. Perhaps a minimum 10% overlay for when probabilities are 50%+ should be undertaken.

All surfaces showed good profits.

When looking at the overlays when the probability of winning is between 20 and 30%, we can see that a minimum of 30% or 40% overlay shows a good profit. A minimum 40% overlay (to give some leeway to the big negative returns when the overlay is 20-30%) perhaps should be implemented.

For the 30% to 40% range, it would seem that the same 40%-50% minimum overlay range should be executed.

The 40-50% range interestingly shows a profit from even the 0% minium overlay. A 13.3%ROI when the overlay is less than 5% shows exactly what a low sample size can do. On average, the expected return could not be more than 5%, so the fact that the return in this period is so high is no doubt due to luck. There seems no reason why a minimum 10% overlay, to keep in line when the probability is greater than 50%, should not be used.

Hence, given the above, it is suggested that the new rules for WTA tennis betting in 2010 will be as follows:

Probability 40%+: min 10% overlay

Probability 20-40%, min 40% overlay.

Let us know what you think in the comments below.

Sign up to the Sportpunter Tennis Model by 31/12/09 to receive a free Sportpunter Polo shirt and free entry, along with myself, into day 1 of the Australian Tennis Open.

We previously looked at the year 2009 for ATP tennis, but now we will look at how well the model has gone over the past 5 years (since 2005). Note that this is based on head to head betting only and not set betting. Results are found below with commentary under that.

A lot of things can be discussed here. Note that because of the “Federer Factor”, betting on tennis players with probabilities under 30% was not profitable. Likewise, bets on players with odds of 6.00 or more was not profitable, and most likely for the same reason. This same reason once again means that it is hard to distinguish between profitable bets for the overlays, as low probability players with large odds are also most likely to have a large overlay.

Better to analyse the results under SP Rules, which looks at a 5% min overlay for all probabilities greater or equal to 50%, and a minimum 25% overlay for all probabilities between 30 and 50%.

Here we get some good results. A total of 7.7% ROI is very very good. Profits are found virtually amongst all probability and odds ranges. Similarly, all overlays had a good profit. It is good to see that the %ROI increased as the overlay increased which is what we would want.

As expected grass court only barely managed to make a profit, whilst all other surfaces did very well. Previously in the forum we talked about grass court and how favourites are more likely to win. Bets perhaps should be adjusted for this, we will talk about this more soon.

When looking at the near bet probabilities of 20% to 30%, we can see that no minimum overlays are profitable. Best to leave all these alone. Even the very high 100%+ overlays have not been profitable.

When the probability is between 30% and 40%, we note that higher range overlays are indeed very profitable. One can suggest a minimum 25% of perhaps 40% overlay for probabilities in this range, which is basically the same as what has previously been suggested.

Interesting to look at is the results for when the probability is between 40% and 50%. Higher overlays generally have a greater return despite the low sample size which is good, but results show that any bet under 10% overlay is not profitable. Previously we suggested bets only when the overlay is 25% or above for this range, but it seems that a minimum of 10% or even 15% to be conservative is more accurate.

I am thinking of changing the suggested bets to be as follows based on the above information:

Prob 50%: min 5% overlay

Prob 40%-50%: min 15% overlay

Prob 30%-40%: min 25% overlay

Let us know what you think in the comments below.

Sign up to the Sportpunter Tenns Model by 31/12/09 to recieve a free Sportpunter Polo shirt and free entry, along with myself, into day 1 of the Australian Tennis Open.

Here we have a brief look at the stats for the WTA betting for the year 2009. Sportpunter did very well betting on the women which is shown below with a return of 5.8%. It was higher mid year, but had a small down trend later in the year, which was unfortunate as this was the same time that ATP fell as well. Still, one can’t argue with a 5.8% ROI profit, and hopefully more profits were made betting sets as well.

Analysis shows good profits over all probability ranges, odds and overlays, which is always a great sign. When overlays were above 40%, returns increased to 34% ROI which is quite massive.

It was good to see that nice profits were made betting on the favourites, with odds of less than 1.70. This was particularly good, because in 2008 this failed to eventuate with so many big name players losing.

All the results point to hopefully another good 2010 season.

Sign up here for Sportpunter’s tennis model in 2010

Following the 2009 Tennis season, I’im going to have a look at how we did during the season.

First up is the ATP, which started well, had a terrible mid to late season and then finished with a good couple of weeks at the end.

The breakdown of all the bets that were made are shown below. This shows a couple of interesting things. Profits were not nearly as big as previous years where in 2008 11.2% ROI was made. Higher probabilities didn’t make a profit, however the amount of bets means we can’t read too much into this. What is interesting is that betting at odds of 3.00 or more failed to make a profit. This is shown in the overlays as well, where our big 100% overlays made a loss with only one win from 23 bets. This is normally our bread and butter. This area, 100%+ overlays was really where the model fell down. I will be analysing the data in more detail to determine why this was the case. With so few bets it could well be just a case of random variation, just not in our favour.

We will have the WTA analysis in the next post, which is a lot more encouraging. Full results to come.

Sign up here for Sportpunter’s tennis model in 2010