Wednesday 29 May 2013

2012/13 Season Review Part Four - Systems 31-41,31-42,32-41,32-42,33-41,33-42

I think I’ve had some fairly decent feedback regarding the reviews I’ve written so far. Had a few tweets and emails from subscribers and I think the general feeling I’m picking up is that these reviews are starting to open up people’s minds to the fact the service isn’t only about the 6 Est Combined systems. Being honest, it’s difficult during the season to stand back and take stock of what is happening at times and that’s partly why I spend so long reviewing all the systems after each season. I actually learn a lot about the systems myself from reviewing them in depth and I try my best to drop the odd hint or idea about how we use the knowledge going forward.  Part of the reason my own betting portfolio did so well last season was because I learnt a lot from the reviews last Summer!

One of my long-term subscribers who happens to be a mathematical statistician by trade has correctly pointed out that my statistical significance section is useless and makes no sense at all. Appears a Chi-Square test isn’t the correct thing to use for what I’m doing given the summarised results and therefore, the results for the P-value are inaccurate.

I knew it was a bad idea to dip my toes into statistical analysis given it’s not my background at all (people often think because of my job and footie modelling, I know lots about stats and maths but I actually dropped maths at university as it was my worst subject ;) ). Anyway, given this, I think I should drop this statistical measure from the subsequent reviews. Happy to hear from anyone via email who can give me something to measure my summarised results by that actually makes sense. Only suggestion so far is multiplying ROI*Bets but not a fan of this myself although it allows me to rank systems which is what I’m looking for really.

On to more important things. The two parts of the eason review I’ve most looked forward to writing for a while are the next two parts. The New Combined systems have just completed their first season and the results have been stunning. Likewise, the Misc systems (particularly the Similar Games Systems) have completed their second season and the results have been amazing too. Hence, I expect most subscribers are going to be looking seriously at whether they take the plunge with a New Combined system next season or a Similar Games system as part of their portfolio.

Similar to what I said on the last review, for the detailed analysis section, I’ve used the last 3 seasons worth of data. Hence, two seasons are backtested and one season is live. The last season of backtesting was a losing season, so I think there is some weight to using the last 3 seasons worth of results to try to draw any conclusions once we drill down into the results.

For any new readers, these systems are obviously combined systems from the other 5 systems 31,32,33,41 and 42. Hence, 31-41 means a team that appears on system 31 and 41.

Season 2012/13 Performance


These systems started the season well and at one stage, looked like they were going to achieve a crazy return this season but the last few months were more of a struggle. Even so, looking at the results for each of the 6 systems, they are very good and look like they have great potential going forward.

In terms of ROI, it’s pleasing to see the two highest systems lead the way. System 33-41 achieved an ROI of 35.5% from 103 bets and system 33-42 achieved an ROI of 30.5% from 91 bets.  Not bad for their first live season and although the bet number isn’t great, it’s slightly better than the Est Systems 8-21 and 8-22 which makes these systems slightly more powerful I think although with only one season behind them, it’s too early to be jumping to that conclusion.

Given the fact system 32 didn’t have the best of seasons itself, 32-41 and 32-42 struggled a little bit but even so, an ROI of 9.3% and 11.6% is much better than the 5.8% that system 32 achieved, so cross referring the bets has worked well.

I think system 31-41 did well in its first season and an ROI of 7.7% across 612 bets is respectable. I think the system that did best from this whole set of systems is system 31-42 and to achieve an ROI of 14.8% across 366 bets in its first season is outstanding. 42 is clearly a better system than 41 and this shines through in this result.

Overall, a profit of 214pts from 1,641 bets (13% ROI) is a great return for the first season for these Combined systems. As a comparison, the Est Combined systems achieved just over 16% ROI in their first season, so although the New Combined systems look like they’ve started well, they have started as well as the Est Combined systems.

All Live Results


Since this is the first season, the same comments apply from above in terms of the returns.

Looking at the ROC figures, I think the figures are very healthy indeed. System 31-42 leads the way with a ROC of 108.3% last season. Amazing performance I think and it will be interesting to see if it can follow it up next season with a similar performance.

System 33-41 isn’t far behind with a ROC of 91.3% and then you have 31-41 with a ROC of 86%. The other systems all achieved very good ROC’s of between 51.5% and 69.4%.

I think it’s pleasing that these systems not only obtained a high level of ROI but also a very high ROC too. The advantage these systems have over the Est Systems is the bet turnover and therefore, the ROC potential is greater with the New Systems over the course of a season.

Targets for 2013/14 Season


Similar to most of the systems I’m reviewing, the targets for next season don’t look overly stretching if we’re only looking at last season but I’m always wary of the fact that you don’t want to be too bullish after one good season. The targets are all lower than what was achieved last season and therefore, even if the systems don’t repeat this season next season, there is still hope the targets could be met.

31-42 leads the way with a target ROC of 77%, closely followed by 31-41 with a target of 65%. The other 4 systems have targets of between 40% and 50%. Considering the lowest ROC for last season was 51.5% on any system in this group, the targets aren’t impossible.

One thing I should mention is the bet number. When you only have one season’s worth of live results, you can’t be sure that the bet number we saw last season will be repeated again next season. Some systems seem to pick up bet volume and other systems reduce bet volume and trying to second guess the bet volume at times is difficult, so again, if we find these systems have less bets next season, the targets may be difficult to achieve.

Detailed Analysis (Not going to discuss this too much given the fact it is a mixture of backtested and live results)

Performance by Season


A couple of things to comment on here.

Firstly, the systems made a loss during backtesting two seasons ago which is the first ever systems I’d built that had a losing season during backtesting! This is basically what put me off these systems this season and ultimately, they have shown the loss to be a blip I think.

The bet number is very interesting though as since they went live, the bet number is much higher than during backtesting.  Discussed this before on the blog but backtesting will never be a proxy for live results since the odds are captured at different times and given the odds are important in deciding where the bets go on the systems, if I run the systems at the odds at different times, I’d get very different results.

Interestingly, the higher bet number is all on the lower combined systems. Systems 33-41 and 33-42 have a similar number of results as backtesting. The results from this season are much like the results from backtesting in 2010/11 on these systems. The higher bet number is coming from the lower combined systems and I suspect there is a trade-off here between the strength of the bets and the number of bets looking at the ROI.

You can see the impact the draw had two seasons ago where AH0.5 made a profit and yet, backing outright was loss making. Last season, backing outright was profitable but AH0.5 was loss making. Maybe the systems got no luck two seasons ago but got plenty of luck last season? Hard to know for sure but I would have been happier if the systems made a better return when covering the draw. Implies that when the teams weren’t winning, they were losing! I much prefer to know teams are drawing when not winning. :)

Performance by League


All the leagues are highly profitable although the SPL and League One have the lowest ROI. Some interesting trends by League where I see that the highest combined bets on League Two are poor but the lower combined bets in that league are fine! Small samples as always though, so more than likely to be variance.

Performance by League and Season


Some leagues look great last season, others look rubbish! Can’t base any decisions on one season though and the past data is hard to fathom I think.

Performance by Home and Away


Right, I do find this very interesting. Home bets do not look anywhere near as good as Away bets on these systems. That’s the complete opposite to the Est Combined systems if you remember. A 19.1% ROI on Aways and yet, a lowly 4.3% on Homes. 

First thing that pops into my head would be to pick the Aways from these New Combined systems and Homes from the Est Combined systems and see how the results look. Obviously, a bit of backfitting going on in my thinking here as I’m picking the best bets from both sets of systems but it’s definitely something to consider.

I know that on the Est Combined systems, I showed in the past it was better to stake more heavily on Homes but this wouldn’t be the same here. It’s important that people don’t take a broad brush approach to these sort of things and what may seem like a good idea for some systems won’t work on them all.

Interestingly, you can see that the Homes are badly affected by draws as the returns are quite similar across each of the betting methods. Can’t say I’ve ever seen this before too often and I suspect the Homes must be affected a fair bit by draws given the average outright odds are 2.22, so the AH0.5 odds would be heavily odds on, so making a 4.3% ROI by backing at heavily odds on is very rare.

Performance by Home and Away by League


Lots of small data sets that are losing money here but with one season’s live results (and one season included which is loss making), I think we need to see another season of live results before we jump to conclusions. BSq Prem Homes, League One Homes, Premiership  Aways and SPL Homes are all loss making! A little strange considering the overall returns but all it means is the Aways must be much better for each league which fits in with what I said above with Homes v Aways.

I think quite a few people are going to be interested in these combined systems next season but I think the key thing from this review is the fact that Homes look a little dodgy and Aways seem to have huge profits for outright betting but I’d be slightly wary of using AH betting on the Aways after last season’s performance where the teams tended to win or lose!

No comments:

Post a Comment