Draw No Bet/Double Chance Analysis

Firstly, I should add that I mentioned last week about doing this analysis on the blog and I’ve been discussing doing this in the TFA forum for a wee while. Some may just look at this weekend’s results and think it’s a knee-jerk reaction to even consider doing something with the draw and these systems (especially given how much profit they have made this season!) but I never consider changing anything based on a short-term run of results. Simply, I’ve been disappointed about the results from some of the systems this season and after a piece of analysis I did on the draw a few weeks back, I decided the next step was to look at Draw No Bet and Double Chance betting for each of the systems.

As you can imagine, it isn’t an easy job to gather this data but thankfully, I’m a bit of a whiz with Excel as you have probably noticed, so I can usually do things with data that others can only dream about! A lot of work has gone into collecting this historical data but thankfully, it will be easy to keep it updated going forward, so this has been a worthwhile piece of analysis.

When carrying out a big piece of analysis such as this, there are some golden rules you need to follow. In my opinion, you should state any assumptions around the data you are using, you should state any priori beliefs/hypotheses about what you expect to find, you should carry out the detailed analysis and then lastly, you should draw any conclusions including referring back to whether your hypothesis was correct or not. If you stick to these steps, you won’t go too far wrong with analysis I tend to find.

Assumptions

As discussed in various places on the blog, I don’t think that the results in season 2006 and 2007 can be used for any meaningful analysis. These two years were involved in the backfitting of my systems although as I have said before, the backfitting is probably only 50%-75% in these two seasons. I obviously had other data from earlier seasons too when I was developing the ratings and therefore, I didn’t use all of the data from 2006 and 2007. However, I used enough of the data for me to say that these years suffer from backfitting and therefore, I don’t want any analysis or trends to be used involving these two years if I can help it.

Apart from excluding these two seasons, I’m happy to use all the data from 2008 and 2009 as well as all results to the 23^rd March 2010 to draw any conclusions. You need to a draw a line in the sand with this sort of thing as your data would be moving all the time. In total, I will be looking at 7,488 bets and these are split 2,483 homes and 5,005 aways for the data from 2008 to now. For parts of analysis which include 2006+ data, there are 14,361 bets. Overall, I think there’s enough data to be able to draw meaningful conclusions even though I have missed out the first two years of data when drawing any conclusions.

Priori beliefs/Hypotheses

I think my main priori beliefs before starting out on this analysis were:

1). DNB betting will be less profitable than traditional H/A betting although it may provide a smoother P&L

2). DC betting will be much less profitable than H/A and DNB betting and I don’t think this will be worth following in future. I wouldn’t be surprised if it was even loss making

3). DNB betting will be best utilised on away bets rather than home bets as away bets have a much higher average odds and a lower win strike rate

4). DNB betting will work well on teams priced 2/1+ as I think this is the area where the systems lose out on most profits as many of these big priced away teams draw regularly

5). Lastly, DNB will not work on short priced aways (less than 2/1) as covering the draw when you are backing at short prices away from home can’t really be a profitable method

Overall, my main hypothesis is that DC betting will not work on these systems and DNB betting may provide a smoother P&L but it is likely to be much less profitable than traditional H/A betting.

Detailed Analysis

I think the first place that is worth looking at is simply the performance of H/A v DNB v DC by year and by system. I have also graphed the 3 P&L’s on the same graph for each system. The drawdowns are only for the H/A betting. For information purposes only, I have also showed the key metrics by each system and league although I don’t think we can use this for any analysis.

Below is a picture of the results:

OK, I could really go to town on what trends I see but I think there are better ways of analysing this data than I’ve shown here. A few things worth mentioning are:

• On every system, the overall returns from H/A beats DNB and DC
• However, this is partly due to the results in 2006 and 2007. If we look at the graphs of each system, the DNB P&L looks parallel to the H/A P&L over the last few seasons
• Looking at 2010 in particular, DNB beats H/A on many systems and overall, it looks fairly close between DNB and H/A betting

I think the above is interesting but it doesn’t really help us to understand how HvA betting compares to DNB and DC overall as it includes 2006 and 2007 data. Let’s exclude this data and just look at the returns by season:

OK, this is starting to look more interesting. Based on the last 3 seasons, the ROI for H/A is 18.3% and the ROI for DNB is 16.1%. DC ROI is lagging behind but is still a very respectable 7.5%.

Interestingly, looking at 2010, DNB easily beats H/A betting for my away bets. The same was also true in 2009! Hence, this isn’t just a trend this season, it was already there at the start of the season.

I think the next stage is to look at the performance of home bets and away bets separately. I could look at this by season but I don’t think it is going to add much by doing this, so I’ll look at it by system and overall. Here it is:

Now, this is getting very interesting. As you can see, Away bets have an ROI of 16.4% for H/A betting whereas Home Bets have an ROI of 22.3% for this type of betting.

Amazingly though, Aways have produced an ROI of 18% for DNB betting and Homes have only produced a profit of 12.3% for DNB betting.

DC is fairly consistent with homes producing a profit of 6.9% and aways producing a profit of 7.7%.

What does all of this mean then? Well, I don’t want to draw too many conclusions without further analysis but clearly, DNB is going to work better on Aways rather than Homes and secondly, the ROI is higher on Homes than Aways for traditional H/A betting, so I don’t think homes are the issue here. Clearly, it is the away bets that are suffering from too many draws and pulling down the ROI of my systems. This backs up what I said above looking at the last two seasons.

So, now we know it is the Away bets that are suffering from too many draws. Let’s drill down a bit further into these bets. This shows the performance of the away bets by season:

As you can see, this clearly shows that DNB betting is better than traditional H/A betting on my away bets. When you consider the average odds of DNB bets will be much lower and will have a much higher strike rate, if given the choice of two systems who produce the same returns, you will always choose the system with the highest strike rate, lower losing runs and smaller betting bank. Hence, DNB is better than traditional H/A betting on my away bets.

What does it look like by odds range though? Well, if you think back to my hypothesis when I started out, I believed that DNB would work better on longer prices but on shorter prices, DNB could never work……..

Here are the results of teams at 2/1+ playing away from home:

As you can see, traditional H/A betting has produced an ROI of 18.5% on these bets. However, DNB betting has only produced a return of 16.1% and DC betting has only produced a return of 8.1%. I find this very surprising as it shows that for teams at 2/1+ away from home, you are better with H/A betting rather than DNB!

So, what does it look like for teams priced less than 2/1 playing away from home:

Well, you can knock me down with a feather duster! Covering the draw when you are backing teams at less than 2/1 away from home has produced much bigger profits than traditional H/A betting.

The ROI from traditional H/A betting is only 14.9% whereas the ROI from DNB betting is a remarkable 19.4%! Over the 3 seasons, it comes to 130pts profit across the 2,896 games.

I rest my case your honour……..

Conclusion

After doing all the analysis, spending a few hours analysing and then writing up this report, I’m just left to try to decipher what the hell it all means!

Let’s go back to each of my priori beliefs and check which ones were right and wrong as a starter for ten.

1). DNB betting will be less profitable than traditional H/A betting although it may provide a smoother P&L

Over the last 3 seasons combined, the statement is true that DNB betting is less profitable. However, in 2010, DNB has an ROI of 13.7% and H/A betting has an ROI of 13.1%. Interestingly, if we split this into Home and Away bets, DNB on away bets have beaten H/A betting in 2009 and 2010. DNB on Home bets has never beaten traditional H/A betting.

As for producing a smoother P&L, I would have to undertake further analysis but based on the fact the strike rate for DNB needs to be much higher than H/A betting to make a profit, it’s safe to say that DNB must provide a smoother P&L overall.

2). DC betting will be much less profitable than H/A and DNB betting and I don’t think this will be worth following in future. I wouldn’t be surprised if it was even loss making

I think it’s safe to say that DC betting is much less profitable than the other types of betting. However, it still produces a steady ROI of 6%-8% a season and therefore, it isn’t exactly loss making like I thought it might have been!

The overall ROI for backing DC Aways is actually just short of 8% and when you consider the strike rate you need for DC betting to be successful, I don’t think it’s the worst system I’ve come across!

3). DNB betting will be best utilised on away bets rather than home bets as away bets have a much higher average odds and a lower win strike rate

This statement is definitely true although DNB betting on homes still isn’t bad and produces an ROI of 12.3%. However, this is much less than the 18% on away bets. In addition, H/A betting beats this ROI on Home bets, so there is no reason to consider DNB for Homes I think.

4). DNB betting will work well on teams priced 2/1+ as I think this is the area where the systems lose out on most profits as many of these big priced away teams draw regularly

This is the most surprising thing from this whole analysis but it is actually more profitable backing these big priced away teams to win outright, rather than covering the draw. It’s hard to understand why but clearly, there must not be as much draws in this category as I thought.

I think finding this out is a massive step forward. Without doing this analysis on odds, I would have been in danger of making a massive mistake and considering backing these bigger priced away teams DNB I suspect.

5). Lastly, DNB will not work on short priced aways (less than 2/1) as covering the draw when you are backing at short prices away from home can’t really be a profitable method

Well, I was pretty certain that I would find out that this was true but it is most definitely not true!

It’s quite hard to believe that covering the draw when you are backing short priced teams away from home (less than 2/1) is actually a better method than backing them outright but this is definitely the case. The difference in profit is actually quite substantial with traditional H/A betting producing an ROI of 14.9% but DNB producing a profit of 19.4%. As I said above, it is 130pts profit across 3 seasons.

So, the piste de resistance of the analysis, what will I do going forward? Well, at the moment, nothing! For a start, we’ve got 9 weeks left this season and therefore, I’m not going to be suggesting that we start covering the draw on away bets at the moment. I’d much rather wait for the season end and take another look at this analysis in the summer before making any wholesale changes. We are playing with profits at the moment, so we don't need to make rash decisions.

Importantly, if I did decide to go down the route of covering the draw with away bets, I would do it with all bets and not just shorter priced bets. Overall, the ROI has been better for the last two seasons on all the away bets and therefore, I don’t want to reduce the sample size by only doing it on a proportion of the away bets.

The other thing to consider for my own betting and anyone else following this is how practical is it going to be to cover the away bets each week with the draw? It is pretty easy to knock up a DNB calculator in Excel where you can enter the stake and best odds for the away team and draw and that will tell you what to stake on each team. However, is this practical for people to follow? I honestly don’t know……..

I’ll leave it there for the moment and hopefully people will have some questions which I can answer. For those that question how powerful this sort of analysis can be, look at the results from this weekend so far…..17 bets, 14 draws, 3 wins. 7 individual games, 3 wins, 4 draws. May look like a freak weekend but read back for the past few months on the blog and in particular, read this post from a few weeks back. The draw is a pain for these systems this season and if backing the draw can increase your profits, surely you have to consider doing this?

http://the-football-analyst.blogspot.com/2011/03/just-track-draw.html