Tuesday, 17 July 2012

What does system 6-21 bring to the party?

I think the next post is an important post. One thing that a few people have said to me over the Summer and even before the last season ended was that they were going to be dropping system 6-21 and simply following the other 5 combined systems.  I think the general consensus was that system 6-21 has a lot of bets and makes the combined systems quite tough to follow over the season due to the additional workload but at the end of the day, it doesn’t actually increase your profitability too much.  Is this true or is it the case that system 6-21 actually helps smooth the P&L in the long-run and importantly, provides an additional number of bets that allow you to increase your ROC than just following systems 6-22 to 8-22?

I’ll be honest and say when I started looking at this, I sort of hoped the second thing was true.  I hoped the additional bets provided by system 6-21 did increase your ROC even though I appreciate following system 6-21 reduces your overall ROI.  If this wasn’t true, then I feel a little bit of a fool for suggesting people follow the 6 combined systems when they could have followed 50% fewer bets and made a similar level of return from doing so.

To be fair, the first person who asked me to look at this was Les. I think I paraphrased his question but here’s what I think he asked in essence.

What does 6-21 add to the 6 combined systems? Does dropping 6-21 improve performance without increasing any variance?

I should also add that Alvin, Gary and Grant have had similar discussions with me about this or have asked me to look at this when I had time.

The issue with answering something like this is that there is never going to be a perfect answer. I know many of the subscribers last season used variable staking plans on Homes and Aways, some followed Homes outright and covered the Aways with draws etc. and therefore, I can never answer a question like this unless I take it at a very high level and assume we’re all playing 1pt on every bet and if we choose to follow an Asian Handicap route, we do so for all bets, even though we know AH isn’t an optimal solution for Home bets.  Hence, any conclusion isn’t perfect but as always, this blog is only my take on things and I’m sure if you asked 100 analysts to do this piece of work, we’d all end up with different ways of looking at the problem and difference answers.

I’ll paste the workings first and then discuss how I have approached answering this.  As I say, no answer is perfect but I think my conclusion is probably correct but I’m sure someone will pick me up on it if they think I’ve looked at it wrongly.

Here’s the summary I’ve come up with:
The first thing to point out is that the summary is very basic.  I make no apologies for this and it’s linked to my point above. If I try to make this more complicated, it would muddy the conclusion and if people want, they can do their own in depth analysis of the data and come up with their own conclusions.

So, what we have here is a table showing the profitability by season for the 6 combined systems and the 5 combined systems.  You’ll be familiar with these P&L tables by now on the blog!

Underneath each table, I have stated what the max drawdown is for each betting method (outright and the 3 Asian handicaps).  I have again suggested that a doubling of the max drawdown would be a proxy for the betting bank required. People pick me up on this all the time but in all my time betting, assuming you have a decent sample of data to work from, I usually find this is as good as any sophisticated mathematical formulae based on strike rate and expected returns. We all know we haven’t had the worst historical drawdown yet and to understand this, you would have to run a Monte-Carol simulation but as I say, if you double the historical drawdown of thousands of results, you never end up too far away from the right answer.

Now that we know the betting bank and returns for each betting method for each season, it’s simply a case of putting these into a table and calculating the ROC under each scenario. It is shown in the picture above but here are the two tables next to one another.

The first thing that jumps off the page when I looked at this was how similar the returns are for each method and each season.  I’d ask everyone to put more weight on the most recent two seasons since these are live results and beyond that, the previous two seasons as these are fully backtested results. The first two seasons are the least reliable and being honest, the ROCs are overstated due to an element of backfitting involved in both seasons.

If you look at it, it’s safe to say there isn’t much in it. The figures at the bottom in red let you know which portfolio is better. Anything below 100% means the 6 systems are better and anything above 100% means the 5 systems are better.

If we take it one season at a time.

Last season has the biggest discrepancy.  This is because system 6-21 had an obscene number of bets and I would have expected the ROC to be better if you include system 6-21 in the portfolio.  The previous season wasn’t as clear cut and you would have to say that the 5 systems were slightly better overall under every betting method.

If you look at the previous two seasons, you would again favour the 6 systems on every method. The first two seasons tend to favour the 5 systems for AH betting and the 6 systems for outright betting but I’d not pay too much attention to this.

At the overall level, you’d say that you would expect the ROC from following the 6 systems to be around 5%-10% better than the returns from following 5 systems.  As you use AH more, the gap narrows. If anyone is using a combination of outright betting on Homes and AH betting on Aways, I think you’ll find there isn’t much in it at all at the overall level.

Of course, next season is more likely to be like the 2011/12 season than any other season, so this does lead me to believe that following the 6 systems would be slightly better than following the 5 systems. However, I wouldn’t bet that system 6-21 would have quite as many bets next season and therefore, I would expect the gap to narrow between both methods anyway.

One very important piece of analysis that is missing from all of this is simply the number of bets.  Yes, it may be the case that following all 6 systems gives you a 5%-10% higher ROC but is it really worth the time and effort?

Well, if we look at last season, dropping 6-21 halved the number of bets you would be having during the season. Hence, although you would have ended up with a lower ROC, you would have halved the time it takes to follow the systems.  How much is your time worth compared to 10% of an increase in ROC?

In summary then, I would say there isn’t much in it between following all 5 systems and all 6 systems based on past performance.  I would expect following all 6 systems would lead to a slightly higher ROC than following 5 systems but then again, you would save a lot of time and effort only following 5 systems, rather than 6, given 6-21 has a lot of bets that appear on that system and that system only in the combined systems.

To answer the question that was posed to me, I believe system 6-21 does add a little to the ROC in the long-run but it should be weighed up against the time you’d save by not following it.  I don’t believe dropping 6-21 would improve the performance of the portfolio (improves the ROI but not the ROC) but importantly, dropping system 6-21 doesn’t appear to bring in any additional variance, so there is no increase in risk from doing so.  Hence, whether you follow 6-21 or not as part of the combined systems portfolio is solely down to the individual following. 

Hopefully people are enjoying these brief discussion topics and it’s whetting their appetite for building their portfolio of systems to follow next season!


  1. Graeme,

    Very interesting. The point about workload, is quite important, particularly at times like March 2012 when I think one day 6-21 produced 24 bets.

    What do the numbers look like if you also exlude 6-22?

  2. Hi Andy.

    I think if you go back and look at the system reviews from earlier this Summer for 6-22 to 8-22 mate, you’ll be able to draw some sort of conclusion.

    6-22 has less bets than 7-21, so in a way, dropping 6-22 doesn’t actually save you too much time at all I would think.

    If you ever want to look at a portfolio of systems, just sum up the returns for each system. You miss out on the leverage factor a little but for the combined systems, there isn’t much leverage anyway as the bets are so correlated. In addition, if you are simply comparing portfolios, if you miss the leverage factor out in both cases, you will still get a meaningful conclusion. :)