Thursday, 14 June 2012

The Euro Combined Systems......

This is ultimately the most important post to date on the European Systems. Although understanding the single European systems and their results is vitally important, at the end of the day, whether or not the European systems are a success is going to be judged on the performance of the combined systems.  I feel like this is the best way to judge the performance and very similar to the UK systems, I’m going to be asking the Secret Betting Club to proof the European systems based on the 6 combined systems.

Right, the best place to start is probably just the results of the 6 systems, side by side.  Here they are:

Based on what we have seen with the UK systems, I would expect to see an increasing ROI trend as we move down the table. Hence, the more selective the system is, the higher the ROI.  Clearly, this trend isn’t showing through here on all systems.

The trend is apparent on E1-E6, E2-E6 and E3-E6 as the ROI goes from 25.4% to 26.5% to 33.1%.  If think about the UK systems for a moment, then the equivalent systems are 6-21,7-21 and 8-21.  What did this trend look like?

Well, the trend for the backtested results was 27.1% to 27.0% to 29.1%.  It’s quite funny as I’m actually shocked seeing that trend on the UK systems! I thought it was much more clear that as you moved up the systems, the ROI increased.  It’s another example I think of me having some sort of misguided belief about the UK system results during backtesting.  My one defence is that it was in early 2010 when these systems were created and of course, we’ve had two seasons of live results since, so my memory is somewhat blurred between what I have happen live and what I saw during backtesting. A lot of my current thinking is based on what I have seen happen live rather than what I remember during backtesting, so go easy on me as we pick our way though these findings!  I’m seeing much of this for the first time as I’m drafting the post,

So, I would say that on these 3 Euro systems, there is actually a better trend than what I witnessed on the equivalent UK systems. 

How do the other 3 systems look?

Well, on E1-E7, E2-E7 and E3-E7, the ROI trend is 35.6%,33.6% and 36.9%.  Now, the first thing that stands out is just the sheer size of these ROI’s. Much higher than the other 3 Euro combined systems.  The second thing is the fact the trend doesn’t hold true and interestingly, as we go from E1-E7 to E2-E7, the ROI falls even though the bet number reduces by 470 bets. It is pleasing to see system E3-E7 with the highest ROI of all systems (would be the equivalent of system 8-22 remember!) and therefore, I’m intrigued to see how the UK systems compared.

Again, I’m as surprised as anyone now I see these results.

The ROI trend on 6-22,7-22 and 8-22 was 34.1%, 35.7% and 37.4%.  Right, first thing to point out is how close the ROIs are to the European ROIs. It’s actually very scary when you think about it! As we can see though, the UK systems did experience an increase in ROI as we moved up on the systems.

Does this raise any concerns for us then?  Well, I’ll be honest and say that I would have liked to see an increasing ROI as we filter on every system (it happens on the single systems) but we have to bear in mind the combined systems are pretty unique in the way they are constructed. It’s hard to get too anxious about systems not showing a certain trend we’re looking for when they have ROI’s in excess of 30%. Most people and to be honest, most wannabe footie analysts couldn’t backfit a system to get an ROI of 30% over a high number of bets if they tried to do this, so the fact I can do it with combining other single systems means I’m not going to beat myself up for a trend not looking quite perfect!

OK, so I’m comfortable with the basic returns on each combined system at a high level.

Let’s look at the split by season:

Firstly, there is a lot of information to take in here. It’s easy for me to look at this table and see things clearly as I analyse data and trends in my job. Others see tables like and no doubt get turned off quite quickly by the amount of information in here, so I’ll try to go slowly.

The first thing to notice is the reducing number of bets by season. I’d picked up on it for the single systems, so not surprising it’s apparent here.  Just in case I was imagining the same thing with the UK systems, I’ve looked quickly and the bet number on the 6 UK combined systems went from 1,274 in the 2006/7 season to 781 in the 2009/10 season and it declined every season.  Hence, I’m not surprised we’re seeing the same trend on the Euro systems.

One thing that I quite like with the Euro systems is the strike rate per season we’re seeing. 4 of the 6 seasons have a strike rate of 45.3%, 45.8%,45.6% and 45.7%.  Quite amazing really when you think there are on average over 1,700 bets in these 4 seasons! (NB. I’m never amazed at things when I look at data as I’ve seen a lot over the years but this is a stunning stat. You couldn’t aim for this and achieve it if you tried!)

I think the more I see the returns from season 2010/11, the less I want to put any weight on them. I think it’s a bad combination of my systems doing well in this season, along with the fact I’ve used some of the data in the backfitting and it is throwing up some amazing results. The fact I can cross refer the two algorithms and get an ROI of 48.2% over 1,309 bets in a season just makes me laugh.

Basically, I’d be very wary of placing any emphasis, never mind too much emphasis on this season if you are looking at the results. Just ignore this season!

The other issue is obviously the first two seasons as they have a high ROI over a greater number of bets than the other seasons. This is identical to the issue I had with the UK systems when they were first produced. I didn’t obviously have enough other seasons though to analyse data, so I was sort of forced to use these two seasons when I analysed the data.  Here’s a question that’s worth posing now….

Should I ignore seasons 2006/7, 2007/8 and 2010/11 when analysing the European systems then?

Well, I think the true answer is yes and no.  Yes, you don’t want to place too much emphasis on the data in these seasons (especially 2010/11) but then again, if I take out these 3 seasons, it basically reduces my data sample down by around 60%!  In this game, you can never have too much data and just ignoring big lumps of data is a last resort in my eyes.

I’m going to take comfort from the fact I used the 2006/7 and 2007/8 seasons data in my original UK systems analysis and it didn’t cause any great issues. Readers of the blog and followers of the service just have to ensure that they don’t get caught up by the returns in these ‘funny’ seasons. If you want a better indication of performance, then the other 3 seasons will provide it.

Anyway, that caveat out the way, let’s get back to the performance table.

So, in the 3 seasons with no backfitting to muddy the waters, we’re looking at ROI’s overall for the 6 combined systems of 18.5%, 32.9% and 23.8%. An average of 25.1% or thereabouts. For the record, if I look at UK combined systems and seasons 2008/9 and 2009/10, the ROI’s were 25.7% and 24.6%. An average of 25.2%.  (Freaky coincidence again I think!)

Here’s these results for the 3 seasons for the Euro systems:

The biggest difference between the UK combined systems and the Euro systems is actually the bet number. For the Euros, the combined systems have in excess of 1,000 bets a season easily, whereas we didn’t see so many bets in the UK systems.

In a way, I actually take greater comfort from the higher bet number. One issue I’m seeing with system 7-22 at the moment is the fact it has less than 100 bets a season now. If we look at the equivalent system here (E2-E7), it has 158 bets and 169 bets for the last two seasons. Of course, this number may continue reducing but during the backtesting, this system has about 50% more bets than system 7-22.

Assuming that the quality of the bets hasn’t reduced on the European combined systems (appears not looking at the ROI’s on the combined systems), then it will always be better to have more bets than less bets I think.

Here’s the split by Home and Away on each system:

The thing that jumps off the page here is the ROI on the Away bets. It’s not totally surprising given Aways did better on the single systems but ROI’s of 40%+ on every system is a bit crazy! Considering I know that the AH0.5 returns are probably easily beatable in practice given we can get better draw odds than I’ve used during backtesting, then an ROI of 10%+ on this method seems like a licence to print money to me. Of course, for all I know (I haven’t looked at anything like this by system at all), there could be massive drawdowns every season and they could make huge losses in some months due to volatility etc. Hence, before we get too carried away with Aways on these Euro systems, they need to be looked at in much greater depth which is the last part of this Euro system project I guess.

The Homes are also fairly decent considering we’ll be looking at much lower average odds and hence, they should provide a smoother journey than the away bets. I really like the fact the Strike Rates and ROI increase as we move up the systems and this is exactly what we’re looking for. The Aways are the ones causing the trends to not look perfect but of course, with such high average odds on these bets each season, all it takes is a few good or bad results each season and it can knock out the trends. Having seen the Aways ROI, I won’t knock them too much!

Last but not least, let’s have a look at the performance by League:

Similar to the other tables, there is a lot of information to glean from a table like this.  Firstly, you can see that around 29% of the bets on the combined systems are in France, 26% in Italy, 25% in Spain and the remainder in Germany. As mentioned on the earlier post, Germany has less games each season, so this isn’t totally surprising that there are less bets in this league.

There is a clear split in terms of profitability though. Germany and Italy are twice as good as France and Spain is about 75% as profitable as Germany and Italy and 50% more profitable than France.  That sounds like one them puzzles you get in newspapers! ;)

Very interestingly, (and this is why I love having the AH returns these days), if you look at the AH0.5 returns, you will see a different picture. France is best and is over 3 times better than Spain! Basically, what this means is that the draw clearly affects the returns in France more than the other leagues. However, if you cover the draw, you will do OK. 

To prove my point (I’m breaking my own rule as I said to myself I wouldn’t deviate from these 4 tables I’ve shown above for all other systems), here’s a very quick look at the Home/Away split for France.

An absolutely stunning statistic (and not one I’ve seen before for any league or system over a large number of bets), you can make as much in France by covering the draw as you can by backing outright. More interestingly, no matter how you cover the draw, it makes similar returns.

Now, I know (as you lot do as I’ve told you!), that the AH returns are understated on these systems as the draw odds are too low. Well, that impacts AH0.5 the most, then AH0.25 and then lastly, AH0. So, we know that AH0.5 is probably the best way to play all French Away games based on the historical results to date on these combined systems. Not only that, you can actually make a very good return hopefully from doing so, so as well as limiting the risk, reducing losing runs and so on, you can probably actually produce a very high return too.  Doesn’t that sound good? 

Of course, reading the earlier table, it was easy to see that France was probably a rubbish league but simply, by having these AH returns available, we can quickly see that Yes, backing outright Aways in France may not be the most profitable way to make money but simply, by using the draw, you can probably turn these Aways into something very good!

Phew, glad to reach the end of this post. It’s been a long one! So, it’s 11pm now and I’m shattered. Taken me nearly 3 hours to do this post and analyse the tables. Doesn’t time fly?

Next step is two-fold. I need to do a blog post to answer Rowan’s questions about whether we should consider following these Euro systems next season. I also need to get back into some spreadsheet work (had a break for a week!) and produce an analysis sheet which will be available for download which looks at these systems in depth.  This is probably going to take me into next week now.

Enjoy the weekend. Without sounding like an advert, the service is slowly filling up and I’ve only about 30 places remaining for next season. If people reading this are interested in following next season, then I’d suggest they join sooner rather than later. Once these places are filled, they are filled. I’ll probably operate a waiting list or something for next season if people contact me thereafter.

No comments:

Post a Comment