Wednesday 6 November 2013

October 2013 Monthly Review

I’m going to try to keep this monthly review a little shorter this month for the simple fact that I’m fed up writing about the poor performance of the systems this season and no matter how much I write, it won’t change anything. I’ll pass comment on the systems and the ratings but you will see the issues for yourself. 

Clearly, the filtering of the systems hasn’t worked at all this season and ultimately, it is starting to look like filtering the ratings is maybe not the best idea at all. Of course, we have two months of poor results against a couple of seasons of really good results, so we need to keep the losses in perspective but for whatever reason, the bets that the systems perceive to be the better bets are letting me down badly. So bad that in some circumstances, it’s hard to believe just how bad the results are. 

Overall, as you will see, it is another losing month but the month was a strange month. The systems managed to get themselves into a really good position midway through the month but then lost all of the gains in the second half of the month. I guess the results in the first half of the month were the first indication that the systems could create profits this season but clearly, law of averages dictate that you’ll win some of the time, so maybe it’s not that surprising they had a couple of decent sets of fixtures before going back to losing again.

I suspect that for many reading this, the month probably feels worse than it actually was but as always, winning money and then losing money always tends to feel worse than losing money and then winning a proportion of it back.

The weekend past was a really good weekend which means November has started well but I’ve been here in October, so we’ll need to see what happens next. Based on the results so far this season, probably means the next downturn is just around the corner again but hoping the systems can buck the trend sooner, rather than later!

Here’s the review of each set of systems and ratings.

Est Systems


Algorithm one made a profit of 7.9pts from 91 bets at an ROI of 8.7%. This is better than the long-term average for this rating algorithm and therefore, I’d have to say it has had a good month. Breaking it down by value band doesn’t really make for a very clear picture with the results being quite random between each band due to the low volume sizes. I guess the good news is the highest value band hit 2 from 4 correct this month for a really nice profit but the lowest band also made a profit. I suspect depending on where you cut the value off, it could have been a better or worse month than just following all the bets thrown up.


If we break it down by Homes and Aways, we see a different picture to last month with Aways doing better than Homes. Aways made 12.2pts at an ROI of 21.7% whereas Homes lost 4.3pts. Again, looking at the value bands, nothing really stands out due to the small sample size. I suspect we’ll struggle to gain much insight from the value bands until we start looking at the season results to date. 



Algorithm two made a profit of 7.4pts from 74 bets (10% ROI obviously!). Even more biased towards Aways here with a profit of 16.3pts at an ROI of 29.7% whereas Homes lost 8.9pts from 19 bets. The value bands are more interesting on algorithm two though with the two lowest bands creating a substantial loss and the higher bands doing better. That’s the sort of pattern we’re looking for here as ultimately, we want the day to come when we can use the systems and value ratings in synch and hopefully increase our profits in the long-term.

Overall, a much better month for both algorithms and they looked much more like the rating algorithms I remember from last season. However, the issue this month lies with the filtering of the systems and I very much doubt the filtering has ever been as bad as this in a single month. Basically, the systems took two profitable rating algorithms and turned the month into an absolute mess.

Here’s the results by system.


It’s impossible to not be taken aback here by the poor performance of systems 7,8 and 22. It’s hard to explain and ultimately, the filtering of the systems this month has really worked against the ratings.

To see system 7 hit 4 winners from 21 bets, system 8 hit 1 winner from 9 and system 22 hit 3 winners from 16 is unbelievable. At the very least, I would expect these systems to be able to filter the bets and at least achieve the same returns (even if they have no edge, if you are picking from a winning sample, you have a chance!) but they’ve done a great job of filtering out the best bets and keeping the worst bets.

The combined systems therefore didn’t have a chance and quite simply, when you cross refer losing systems, I’ve noticed it tends to make things worse at times than the systems you are using and this happened this month. System 6-21 at least gets pass marks as it didn’t make things too much worse than 6 or 21 but the other systems 6-22 to 8-22 were shockingly bad.

No winners from 9 on system 7-22 and no winners from 5 on 8-22 sums up the month as historically, these have the highest ROI’s of all my systems since they went live. When the highest performing systems lose badly like this, anyone following has to take a hit.
Overall then, systems 6-22 to 8-22 lost 33.9pts from 60 bets. Given the fact system 6-21 had a 7.1% profit, it is a scary month for these systems. If they lose like this when the ratings have an OK month, what will happen if the ratings have a poor month?

I’m trying to not panic too much but quite simply, this is about as bad as you can mathematically do I suspect. It is pushing the boundaries to take a set of bets with a 7.1% profit and turn them into a loss of over 50%!

The only explanation is that the systems have been a little unlucky and if you look at the historical results, the filtering does appear to work OK and I’ve seen nothing to suggest that we should see these results as anything but a shock. The issue with filtering of bets is always variance and we have had some massive variance this month and it does beg the question why we bother filtering the bets in the first place. I know people don’t like placing lots of bets and they look for an easier way to make money but maybe I’m starting to think that leads to some massive swings in profitability and therefore, is it really worth looking for shortcuts?

Overall, the systems in total lost 33.5pts from 340 bets. That makes it two monthly losses in a row to start the season and going into the season, these systems have suffered 6 losing months in 27 months (3 seasons). Hence, an average of 2 losing months a season for the last 3 seasons since the systems went live. On one hand, it leads me to think they’ll go on a run like we’ve seen before and hit many months of profit in a row or we now face up to the fact that whatever edge these systems had over the last 3 seasons has vanished over the Summer.

I think it’s difficult to draw conclusions based on 2 months of data where it hasn’t been the easiest of trading conditions but if I had a RAG status for these systems of mines, the Est Systems are now flashing RED which is an unfortunate place to be given the fact they have been so good to me and many of us over the last 3 seasons but it’s an honest assessment of how I feel at the moment. The next month or so is key for these systems.

New Systems


Algorithm 3 was the only algorithm to make a profit last month and as I said at the time, to make a profit when the trading conditions were so difficult and when every other algorithm playing in the same leagues made a loss shows the strength of the algorithm. When trading conditions are favourable, any algorithm can make a profit but very few could last month. Well, with slightly better trading conditions this month, algorithm 3 made another decent profit again. 9.6pts profit from 100 bets.

Similar to the first two algorithms, Aways did much better with a profit of 16.5pts from 70 bets. Homes made a loss of 6.9pts from 30 bets. The value bands don’t appear to show up anything due to the low volume of bets again.


Algorithm 4 was the only algorithm to make a loss this month. A loss of 2.8pts from 114 bets. I personally rank algorithm 4 as my lowest algorithm and I’m never too surprised when it had a losing month. Homes were the issue again with a loss of 5.5pts from 40 bets. Aways made a profit of 2.7pts from 74 bets. The higher value Aways bets outperformed the lower value Away bets and that’s something that’s interesting I guess. Again, a very small sample but I think long-term, it gives me a little more hope that we can use the systems and the value ratings together to increase profits.



In terms of the systems then, a profit of 13pts from 558 bets. Not the best month by any means but considering the loss on the Est Systems, a profit is a profit!

It’s a very similar story to the Est Systems though when it comes to the filtering of the bets. System 31 to system 32 was OK and it actually increased the ROI.  The issue lies in moving from system 32 to system 33. System 32 made a profit of 13.6% and yet, system 33 selected 13 bets and only hit 1 winner. A loss of 9.4pts.

When you start to add this together with what we learnt from the Est Systems, you can see the issues we are having this season. As we move down the systems to the ‘so called’ higher ROI systems, the performance is dropping off a cliff. System 33 has had a nightmare month as have systems 7,8 and 22. Traditionally, there will be many bets that appear on all these systems together and anyone following combined systems including these 4 systems have to be making fairly substantial losses.

If we look at the combined systems here, everything is fine down to 33-41 and 33-42. It goes wrong badly here though with 1 winner from 13 on 33-41 and 1 winner from 10 on 33-42. 

On one hand, I think it’s just variance and 10 bets or 13 bets is nothing really but on the other hand, I think wow, that’s really, really bad! I saw a comment on another blog recently that said my results are so bad, it has to be bad luck or variance as no system could be that bad at picking winners and I sort of think that’s maybe true! If I was selecting lays, to hit 1 winner from 13 at football bets would be the stuff dreams are made of, so for it to happen on a system which has an ROI of 15%+ is difficult to do I suspect. 

Overall, it’s another tough month for the New Systems but the base ratings aren’t doing badly at all. System 31 (algorithm 3) has now had two winning months in a row in the toughest run my ratings have ever had. If that’s not promising, I’m not sure what is! Wait till we hit a good spell and see what system 31 can do. :)  I’ll take the small profit and move on. 

Misc Systems


At a high level, an OK month for this set of systems and a profit of 12.9pts from 318 bets at an ROI of 4%. However, when you look at the results in a little more detail, things were not quite as good as the trusty home bets had a nightmare this month whereas the aways did really well. A complete reversal of last month. 

Dealing with the systems first though, an OK month for 6-32 with an ROI of 7.2% from 39 bets. System 32 had a decent month as did system 6, so no real surprise this system did OK. 

The same comment applies to 21-31 and 6-21-31 in the sense the rating algorithms all had profitable months and therefore, betting on the teams that appeared on each algorithm wasn’t a bad strategy. 21-31 did best and achieved a profit of 14.2% from 72bets. 6-21-31 missed out 3 bets and only achieved a profit of 12.6% from 69 bets. I can see from the strike rate that one of the 3 bets missed out was a winner! 

Overall, a decent month for these 3 systems and no complaints. My compliant probably lies with systems TOX, STOY and STOZ. Losses on all 3 systems and after a poor first month of the season, they have followed it up with another poor month.

STOZ was the best performing of the three with a loss of only 0.9pts but TOX and STOY lost 3pts and 5pts respectively. Not large losses and nothing to be overly concerned about but it’s disappointing that these systems performed worse than the other 3 Misc Systems. Historically, they have performed much better than these systems, so it’s disappointing that didn’t happen this month.

I expect STOZ is a very popular system after the results last season and therefore, to see is struggling this season after two months is disappointing but hopefully it’s a short-term blip.

The system does have the ability to win large amounts in short spaces of time which makes it so powerful but with the increased return comes increased risk and it can be quite volatile compared to other systems. Hopefully it settles down and gets back to something like the long-term norm.

As I mentioned in the introduction, the split by Homes and Aways is interesting as you can see here:


Traditionally, the Home bets on these systems have been the strongest Home bets my systems can find (especially STOY and STOZ) but for whatever reason, they didn’t win last month. However, the great thing about looking at data at a granular level is you can try to understand why and in this case, we can see the clear issue the Homes had last month. When you can make a profit backing AH0.5 on home bets, there must be a very high draw %. 

Looking quickly, the 9 bets on TOX and STOY yielded 6 draws. When 66% of the games are draws, it’s difficult to make a profit!

Of course, with the low average odds, it’s not an excuse but so often, there is a thin line between a draw and a win. If I look quickly, I can see 4 of the 6 lost leads. Hence, in a more normal month, I wouldn’t expect 4 of the 6 best Home bets to lose a lead and draw and therefore, it is just bad luck I suspect what has happened this month with the Homes.

Interestingly, if you look at the AH returns for the Aways, you see the complete opposite trend. AH0.5 created a substantial loss for Away bets and therefore, if the bets weren’t winning, they were losing!

I’ve talked about this before on the blog but if you only look at the outright results of systems, it can be misleading. I would go as far to say that the Home bets were probably very unlucky this month and the Away bets were very lucky and therefore, the systems did better than they deserved to this month. However, given many people stake more on Homes than Aways, we would have preferred the Homes to carry more luck I suspect. 

Overall, a profitable month but hoping to see these systems improve as the season goes on and in particular, hoping that STOZ can reach the heights of last season!

Draw Systems

A month can be a long time betting and I’ve gone from having very little confidence in these systems to really embracing them after the last few weeks. I think my issue in month one was that I tend to look at how much value I think I’m getting from looking at closing odds lines compared to what I’m backing at and in the main, I couldn’t see much value at the odds I was playing at.

However, the other way to look at value is simply at the results and as I said in the first monthly review for these systems, they were a little unlucky and if a couple of late goals had not happened, things would have looked much better. Well, after a profit this month, I’ve changed my tune a little but not simply due to the fact a profit was made. The games thrown up seemed to be the sort of games where backing the draw made sense and in most cases, many of the games were very close even though they may not have finished a draw. Many of the games were low scoring and we also hit a number of 0-0 draws this month (the perfect draw?) and adding all that together, I’m feeling much better about these systems. 

The last thing that encourages me is simply the fact that when I’m placing the bets, there is basically no issue with liquidity or odds movement in the main. All of my bets are with Pinnacle and if I look at the odds at kick-off, they’re pretty close to the odds I’m backing at, so considering I’m looking at this game long-term and ways to make money long-term, being able to find a profitable edge at Pinnacle is a big advantage.

Saying all of the above, we are two months in and the systems have barely had any live bets, so I’m not getting carried away but I can at least see me continuing with a couple of these systems in my portfolio. Last month, I was close to writing these systems off after month one.  Here’s the results from last month: 


In total, a profit of 64.1pts from 210 bets this month. The strike rate was 37.6% and the average odds are 3.52, so you can see why the systems had such a good month.

As I said last month, I think we have an issue with these systems in terms of the combined systems not having enough bets once you go past D2-D7 and the same is true this month. 3 or 4 bets a month is no good for any system and therefore, you have to think D3-D6 and D3-D7 are a waste of time. However, the turnover is decent enough on the other combined systems and the issue is actually D3 which only had 4 bets this month. I expect D3 will have no future going forward and we’ll only have D1,D2,D6 and D7 and the 4 combined systems.
D2-D7 is a system I follow (equivalent to 7-22 which many of you will be aware of) and to see the system hit 7 winners in 11 bets this month is fantastic. Small sample and it had a losing month in month one but it’s up and running now.

Long-term, I still hope that the Draw systems have a part to play in the TFA portfolio of systems but we’ll see what the next 7 months hold. 

Euro Systems 

The first month for the Euro systems this season and a small profit was achieved. I’ve deliberately played down these systems after seeing how difficult it was in these Euro leagues last season to find an edge and therefore, my exposure to these systems this season is very low. I probably put too much faith in the systems last season in my portfolio and I got lucky in a way with the systems I followed doing OK but only playing a small number of bets this season to see how the systems do.

It’s actually really interesting that the results for month one mirror the results for last season. The ratings algorithms both made a loss (E1 and E6) but the filtering worked well and the combined systems all made a profit this month.  Thinking back to last season, the results from the base ratings were very poor but the combined systems did OK and the higher combined systems had an OK first season. It would be interesting if the higher combined systems could again do well this season as it would mean that we have two seasons of OK results and that would give me hope for the future.


Overall, a profit of 8.9pts from 134 bets. An ROI of 6.6%. The combined systems made a profit of 13.8pts from 52 bets which isn’t a bad start to the season.

I think my thoughts on these Euro systems are similar to the Draw systems. I would love to create myself an edge in these leagues as quite simply, there are no issues with liquidity and that’s a massive advantage in a game where I know my time is limited with many of my current bookmaker accounts and more and more of my betting is going through Pinnacle. 

Early days and we’ll see what the next 7 months bring for these systems.

Under-Over Systems


A poor month for these systems with only 17 correct from 44 games. A loss of 8.7pts.

Not really sure why I spend my time doing these systems to be honest. If they don’t make a return this season, they won’t be here next season! Too much time spent producing bets and recording results on these systems that I don’t think anyone should follow.

Overall

A tough month again although if you look at all systems combined, it’s not anywhere near the disaster of a month we had in September. In totality, a profit of 56.7pts was achieved across 1,604 bets. The frustrating thing this month has been the performance of the Homes and whereas the Aways caused the issue in September, in many ways, the Homes have been the issue this month although the Aways have still not been great.

I said last month that October was a big month for the service but November is massive. Never had a losing November in the last 7 seasons of results (both live and backtested) and most systems have never had 3 losing months in a row. Put these two facts together and if November doesn’t improve, we have to seriously start getting worried about some of these systems.

I’ll do a post next which summarises the results to date this season to the end of October and the results since the systems went live. Will help provide a little bit of context about how bad these results are this season! 

2 comments:

  1. Hi Graeme

    It might be interesting to track the returns from betting on U2.5 goals in the games the draw systems select, could be another angle to look at if the draw systems prove to be successful.

    ReplyDelete
  2. Hi Brian.

    Don’t you think I have enough systems to track? :)

    Yes, it’s fair point. Looking quickly (and I mean quickly!), many of the draws are being thrown up on the Under 2.5 system but they are being filtered out by the fact the system uses a min odds cut-off of 2.00. When the system was first built, it really struggled with odds on bets as the strike rate wasn’t high enough. I guess what we will see then is many of the draws appearing on the Under system but at odds of less than 2 as the goal expectancy of the draw games is usually always low.

    I expect the Under-Over systems will be in their final season before they retire, so I doubt I’ll have much appetite to do another system like this but if it was simply tracking the draws for Under 2.5 goals, that seems doable and less work for me!

    I’ll take a look at it next Summer as it shouldn’t be too hard to see if there is an edge backing Under 2.5 on all draws.

    Cheers for the comment.

    Graeme

    ReplyDelete