Wednesday, 9 March 2011

Just track the draw.....

One thing I've really struggled with recently has been the performance of the systems since the turn of the year. Up to the end of November, all my systems had been behaving exactly like the historical results suggested they would. Steady monthly returns, short losing runs and fairly consistent returns. ROI's of 20%+ were not uncommon on most systems and the very best systems had ROI's of 30%+ this season which again, was not too unlike the past.

December then came along and over a shorter number of games than we'd usually see in a month (due to postponements), the systems produced exceptional profits, the like some of the systems had never witnessed before. As mentioned in my summary for December, the overall ROI was 40% on the single bets and this was a phenomenal month.

Clearly, I didn't look at December in too much depth as when you are winning money hand over fist and the systems look unbeatable, it doesn't really seem like the best time to over analyse what is happening!

January then appeared and after an OK start, some of the systems fell away at the end of the month and a few of the systems had a losing month which doesn't happen very often. However, nothing unusual seemed to happen I think although I was conscious of the fact that there had been a lot of draws on the systems at the end of the month.

It was then onto 'Armageddon February' (as it is now known by me!) where everything appeared to go wrong and the strike rate for the systems dropped to an historical low. As I said in my monthly review I think, I was conscious that there had been a lot of draws in the month but due to the fact I don't track the draw at all, I couldn't really say for sure whether it had been any less normal than usual.

Well, I've spent the last couple of days collecting all the data I need to look at the draws and I find the results interesting and quite informative. I personally feel like I can put February out of my mind now as these sort of months are going to happen now and again and it's important to always think long-term and not get too drawn into the short-term, no matter how bad it gets at times.

So, where do I start........

Season Comparisons

This table below shows the split of homes/aways/draws by season on the systems overall. I've also graphed the % at the end of each season (taking into account all data beforehand) to the start of this season.

A quick observation of the first 4 seasons would be that the numbers of games finishing a draw has been increasing year on year but it has been stable at around 26% for the last 2 seasons.

As you can see, at this point in the current season, we're sitting with a draw % of 29%. This is up a good bit from the last two years but not out with the boundary of possibility. Hence, this probably explains why my systems haven't been as profitable this season as they have been for the last couple of seasons.

Monthly Comparisons

This table shows the % of homes/aways/draws across each month across all systems overall.

As you can see, the overall draw % is 25% across all the games which isn't too surprising and in line with what I thought it would look like. Doing a couple of quick statistical checks, you would expect 9% of the draw % to lie within 2 standard deviations of the mean. Therefore, you would expect 95% of the observations to lie between 12.5% and 37.5% in this table.

As it turns out, only 2 of these observations lie outwith this boundary. Which months are they? December 2010 and February 2011!

An amazing statistic that 2 of the outlying months come within 2 months of each other. I'm not sure if there is some sort of correction factor that happened at the start of 2011 to correct for the fact the end of 2010 was so profitable but I do find it interesting. The draw % was only 11.7% in December whereas it was 38.5% in February. They're both lying about the same distance away from the mean, so they were as likely as each other to happen.

What does this mean? Well, I think we can ignore the fact that February was the biggest losing month the systems have ever had as it is unlikely to happen again in the near future. Likewise, we can ignore the 40% returns that were achieved in December as this is also unlikely to happen again in the near future.

What is does show though is that when the draw % is behaving and is in line with the long-term norm, the TFA ratings and more importantly, the TFA systems have an edge here that can be exploited.

Systems Comparison

You need to be really careful about looking at how systems compare to each other with things like this. The obvious reason why is simply the data is very thing on all of the systems. When you are looking at trends over a few hundred observations, it isn't easy to jump to the wrong conclusion and therefore, I don't intend on commenting on the system by system comparison of how the draw is behaving. Clearly, with a fewer number of games, you are much more prone to big swings in the percentage of draws you can hit at any point in time.

What does it all mean?

Well, I think we can safely say that the exceptional month of December 2010 isn't likely to be a regular occurrence and likewise, the nightmare month of February 2011 is also something that shouldn't be appearing too often.

We can definitely say that the draw is the enemy when it comes to my systems and my ratings. When the overall draw % gets very high, then it is inevitable that my systems will struggle to make money. Likewise, when the draw % gets very low, my systems will outperform in the short-term.

Overall though, the draw % has been fairly steady year on year in these leagues I am looking at and therefore, I expect my ratings to remain profitable as long as there isn't a fundamental shift in the number of draws occurring in each of the leagues.

For those that like to see this sort of thing on a graph, here is the correlation between the ROI by month of my systems and the % of games that finish a draw by month. As you can see, the systems have an exceptionally high ROI in the months when the draw % is very low and when the draw % is very high, the systems will lose money in the month or struggle to make much money. Importantly, when the draw stays within the 'normal' deviation in most months, my systems are profitable which is all that matters to me.

As the title of the post indicates, I don't need to track how the systems do on a daily/weekly/monthly basis, I just need to track the % of games that finish a draw to work out if I will make a profit or not using my systems. ;)

No comments:

Post a Comment