Wednesday, 2 October 2013

September 2013 Monthly Review

Over the years, I’ve tended to dread writing the monthly reviews after a losing month but I’m actually looking forward to this review. I’m not looking forward to discussing the results though and it has been a very painful month for me and everyone else following the systems I suspect but as I’ve seen over the years, we get these months now and again. The key is how the systems respond to a month like this and historically, the systems have always bounced back fairly quickly, so I’m not giving up hope yet but my confidence has taken a massive knock this month. If my confidence has taken a knock, I dread to think what new subscribers are feeling like but hopefully they keep a level head.

OK, in terms of these monthly reviews this season, I think as well as looking at the systems performance, I need to spend time looking at the performance of the value ratings. Until this season, I’ve never cared too much about the value ratings but I know a number of people are using the ratings rather than the systems this season, so I need to review both ways of following the service. Clearly, the systems rely on the ratings and it has been a terrible month for the ratings and systems, so I can’t imagine anyone has managed to do anything other than suffer a large loss this month but it will be interesting to see how the value ratings have fared compared to the systems.

Similar to previous seasons, I’ll review one set of systems at a time. Due to the fact this has been such a horrendous month, I’m going to look in depth at the results a little more than I would usually do but I think this is fair given the performance this month. Don’t expect this level of analysis every month but as I’ve said in emails to subscribers, my way of coping with the size of loss I’ve had this month is to overanalyse things a little. It doesn’t change the results, it won’t change what happens next but I really need to be able to rationalise the loss for my own mental state. I’ve lost over 12% of what I’ve won in the past 3 seasons and I’ve lost over 25% of my betting bank this season. Hence, this isn’t just a normal loss for me. This is a substantial loss.

I think before getting onto the systems, it’s worth discussing the results we have seen so far to date in all games I’ve rated so far this season. I deliberately try to avoid discussing trends in the results as at the end of the day, what we win or lose is all that matters to me and anyone following and when things are going well, I don’t tend to write about how easy it is to make money as the betting Gods are smiling on me.  However, I would be doing myself a disservice if I don’t comment on the season results to date.

I have rated just less than 190 games this season in terms of finding selections. I wait until 6 games are played in each league although this season, I dipped into League One just before this as a number teams hadn’t played 6 games but some had. In these 188 games, what do the results look like?

Backing all Homes outright has created a 10% profit, backing all Aways has created a 20% loss and backing  all Draws have created a 20% loss. Simply placing 1pt on every outcome of these games has created over a 25% loss this season. Long-term, I know that the average loss is 5% for each of the 3 outcomes. What does all this mean?

Well, given the vast majority of my bets are Away bets, trying to turn a profit when all the Aways have lost 20% is very difficult. The same comment goes for the draws. Trying to achieve a profit backing draws when they are heavily loss making is nearly impossible. On the other hand, making a profit should have been pretty easy if backing Home wins.

Keep these stats in mind when reading the performance of the rating algorithms and the systems so far this season.  :)

Here’s the review for each set of systems and rating algorithms.

Est Systems

I think it’s probably easier to discuss the ratings first and then we can look at the performance of the systems.

Here are the results for rating algorithms 1 and 2 for this month:

Well, to say it has been a shocker of a month for both rating algorithms is putting it mildly. Much of the loss has come from the lowest value bets as these have the highest turnover but the highest value bets haven’t covered themselves in any glory so far this season either. It’s difficult to draw conclusions from such small samples and then splitting it down into 7 more groups makes it impossible to draw any meaningful conclusions but basically, algorithms 1 and 2 have been rubbish this month.

One thing that does stand out is the impact that the draw has had. Both algorithms have made reduced losses if using draw coverage and the more draw coverage, the more the loss is reduced. However, even if using draw coverage on every bet, a loss would have still been made and therefore, we can’t pinpoint too many draws as the reason the algorithms lost this month.

Probably worth looking at the split between homes and aways for both algorithms.  Here’s the results:

I think this picture goes a long way to explaining what has happened this month. Algorithm 1 has had 6 winners from 46 away bets for a loss of 23.3pts. Algorithm 2 has had 5 winners from 39 away bets for a loss of 19.1pts. With respective average odds of 3.98 and 3.87, we’d have expected another 5-6 wins.

Homes paint a different picture though. Algorithm 1 has had 11 from 21 homes winning for a profit of 3.9pts and Algorithm 2 has had 8 winners from 15 home bets for a profit of 2.8pts. Therefore, Homes have performed in line with expectations.

When you break it down like this, you can see that ultimately, the problems this month stem from a small sample of Away bets. We’re looking at 5-6 games becoming wins rather than draws/defeats and the ratings would have made a profit on the month. I’ve discussed many times before how this game is full of thin lines and I think this again points to this.

Is it likely to be the case that Algorithm 1 and 2 have lost the edge that was so prevalent for the last 3 seasons over the first 39 and 46 away bets this season? I doubt it and if I look at the live results for both algorithms for Away bets (even after this month), algorithm 1 has a profit of 4.5% from 926 bets and algorithm 2 has a profit of 9% from 881 bets.

It can only be put down to variance and the fact that neither algorithm has had enough time to show the true underlying edge this season.   I don’t believe that we can dismiss the performance of 850+ live bets over the last 3 seasons for the exact same type of bet and think that the edge is no longer there.

OK, I think that’s my conclusion regarding algorithms 1 and 2. The only issue this month appears to be a small sample of Aways and the fact that Homes are in profit gives me confidence that the overall ratings haven’t lost the edge overnight.

I think for most of us, the performance of the systems is what we are really interested in for this month. Here are the results for this month:

The first thing to point out is that the draw has had a big impact on the overall performance of the systems.  I pointed it out with the base ratings but clearly, the teams who have appeared on more systems have been impacted by draws. A 92.4pt loss if backing outright is reduced to a loss of only 22.1pts if using AH0.5. That’s a significant shift in profitability.

The second thing to point out is the fact that as we move up systems 6 to 7 to 8 etc. the ROI’s are getting worse. This goes against everything we have seen over the past 3 seasons where the ROI improves as we move up each system. However, it all down to the draw.

Take system 7-21 as an example. 4 winners from 20 bets this month and a loss of 10.1pts. Shocking and I expect this system is in most people’s portfolios. However, if you look the AH returns, AH0 was a loss of 2.9pts, AH0.25 is a profit as is AH0.50. The same is true for systems 7,8,7-21 thru to 8-22. Therefore, the higher combined systems have really been impacted by draws.

So, I think my first conclusion is that if we look at the overall month for the Est systems, although I wouldn’t say all of the month’s losses is due to an above average number of draws, it definitely helps to explain some of it.

I think if we look at the split of homes and aways overall on the Est Systems, we’ll see where the issue lies this month with the performance of the systems. Here are the results:

A loss of 100pts on the Aways from 191 bets and a profit of 7.8pts from 71 home bets. I think we can see where the issue lies this month!

Again, if you look at the AH returns, any sort of draw coverage on the Aways improved the results substantially and it’s hard to ignore this.  I can pinpoint a couple of games in the last two weeks where I think we should have had winners (Hartelpool last week lost a 90th min equaliser, Middlesboro threw a 2 goal lead) but at the end of the day, that’s the name of the game. In the last two weekends, we’ve  had Plymouth (5.00), Partick (3.75), York (3.85) and all 3 played well, two of them took the lead and couldn’t hold on. When so many appear on multiple systems (Partick and York appeared 11 times), one goal for either team would have caused a 40pts swing. Goals for two teams would be an 80pt swing. 

I’ve discussed this lots of times on the blog and in emails over the last 3 years but at the average odds we play on Aways, it doesn’t take much to turn a profit into a loss. It takes a few things to go wrong to create a 100pt loss from 191 bets but we achieved it this month.  Does this mean the edge is gone and the systems have turned into a pile of shit overnight? No, I don’t think it does.

Lastly, I think most following the systems are keen to understand just how this loss compares to other losses the Est Systems have suffered in the last few seasons.

I kept referring to Sep-11 as being very comparable to this month but as long-term readers will be aware, in the first season of the systems going live, the systems suffered very bad losses in Feb-11 and Apr-11. Here’s the 4 months side by side and showing the P&L split between home and aways:

There are so many ways to look at this I suspect and others may disagree with me but I would say that based on the table, my ratings performed slightly worse in Sep-11 than Sep-13. Backing outright, Sep-13 is the worst month the systems have ever had but I think if you want to judge the ratings, then the draw has had a bigger impact in Sep-13 than Sep-11. Funnily enough, I think if you look back at Feb-11 and Apr-11, you can see that although the losses are comparable if backing outright, as I pointed out at the time, it was all down to draws in these months. Hence, I would say these were much better months than Sep-11 and Sep-13.

One thing that does stand out in the table is the Away performance this month though and clearly, it has been a freak set of results. If the ratings were as bad as they have appeared to be this month, would we expect the Homes to have made a profit?

I think that brings this monthly review of the Est Systems to a close. In summary, a terrible month but I think it’s nowhere near as bad as the results suggest. The Homes performed adequately but the Aways have had their worst month ever by a long, long way. A 13% strike rate at average odds of 3.73 is pretty tough to do if I was laying teams, so it’s a nightmare month but at the end of the day, it has happened and I can’t change it.

Having spent a fair bit of time playing with the data, I can’t see anything too wrong with the selections this month. I picked out some Aways that should have won above but I could also pick out some bets that we’ve had this month which actually lost but with hindsight, we were on an amazing value bet. The best example I can see is Macclesfield. The systems had them 11 times (a max bet) against Chester in a midweek game. Chester won 2-1. Macclesfield have played 3 games since and won all 3. Chester have only 1 win in 5 now (against Macclesfield) and have slipped to 2nd bottom in the league.  With hindsight, Macclesfield was the best bet I gave this month by far. They lost.

There were 5 max bets this month. Partick (took the lead and game finished 1-1), York (took the lead and game finished 1-1), Blackburn (0-0 against Huddersfield but couldn’t score and yet, scored 3 in the first half away from home on Saturday in their next game), Macclesfield (lost at Chester but won next 3 games) and last, but not least, Newport, who won at odds of 3.2.  Newport won again on Saturday and York managed to overcome the odds against Portsmouth and beat them 4-2.

With hindsight, they were 5 very, very good bets and we staked 55pts on these and only hit 1 winner from 5. If my ratings can find another 80 max bets like this over the rest of the season, I’ll be amazed if the Est Systems have a poor season.

New Systems

Here’s the performance of both these rating algorithms split by Home and Away bets.

Right, I think the first comment to make is the fact that algorithms 3 and 4 have outperformed the first 2 algorithms by a long way this month. Algorithm 2 is historically the strongest algorithm and therefore, it is interesting that algorithm 3 and 4 have wiped the floor with algorithm 2 this season.

Algorithm 3 managed to create a profit this month which in a disaster of a month, is somewhat surprising. When you look at the breakdown by Home/Away though, you’ll see it has been a special month for Home bets on algorithm 3. 27 bets and 20 winners at an average odds of 2.38 to create a profit of 22pts. Stunning results and for a small sample of bets, it shows what can happen when things go well. Algorithm 4 wasn’t quite as strong but still achieved a profit of 14.9pts from 35 home bets.  I thought the home bets on the first two algorithms did well but these results are much better.

When we look at the aways though, we see the same issues as the first two algorithms. Massive losses. 18.2pts lost on algorithm 3 and 23.8pts on algorithm 4. The draw definitely impacted algorithm 4 more than algorithm 3 this month but I don’t think either algorithm was impacted as much as the first two algorithms.

Overall, not a great month for algorithms 3 and 4 but compared to algorithms 1 and 2, different class!

Here are the results of the New Systems (seems a weird name now they aren’t new anymore!):

This is where the picture starts to get really confusing to those who understand the ratings, including me! I have just came to a clear conclusion that algorithms 3 and 4 have done much better than 1 and 2 but then you look at the system results and think, WTF has happened here?  31-41 has made a profit amazingly (to make a profit in a month where all outcomes achieved a 25% loss is absolutely stunning!) but working down the combined systems, the losses soon mount quickly.

Another amazing stat is that system 32-42 only hit 3 winners in 20 bets! As the systems have filtered the bets, they have done an amazing job at picking out losers! Take system 31. A profit in the month (really can’t exaggerate how impressive this is) but then, system 32 takes these bets, filters out all the winners and produces a loss of 14.6pts. Crazy figures and although I’ve seen something like this before, not sure I’ve seen it to this extent with the systems.

If you look at the AH returns, the draw clearly hasn’t had a big impact at all on these systems and quite simply, the filtering process just hasn’t worked at all. I’ve seen this happen once or twice before (if you read the monthly reviews on the blog, you can see it yourself in some months) but it’s annoying when it happens. The draw isn’t to blame on these systems and therefore, you have to say it has been a worse month for the New Systems than the Est Systems. Strange saying that considering algorithms 3&4 were better than 1&2 but that’s the way it has turned out.

I think the next picture paints the picture well for what has happened on these systems.  Here’s the P&L split by Home and Aways:

Let’s deal with the positive results. One of the best results for Homes that these systems have had to date. A 39.5% ROI on Homes is amazing and although I’ve said it was an easy month to make profits on Homes, it wasn’t that easy! Crazy return in a crazy month and for those people asking themselves whether or not my ratings are working well, then I think this shows there isn’t too much wrong.

On the other hand, the Away return is crazy for the opposite reason. A 146pt loss from 298 bets is difficult to do I think in a month! A 14.8% strike rate from 298 bets at an average odds of 3.78. A brilliant performance for a laying system and if I was laying these, I would doubt I think I could have a better month! Given I’m backing these, hard to imagine a worse month than this.

In terms of how bad this is historically, all the New Systems have only been around since last season although the algorithm has been around for 2 previous seasons.  Saying all of that, even if I look at all of the backtested results, I can’t see anything as bad as this or even close to this. Here are the 3 worst months historically:

Both of the previous large losing months were in some way down to the draw as you can see by the AH returns but this month, the draw had very little impact. Therefore, I have to admit that this is the worst month these systems have had. What does this mean? Well, it means we are into unprecedented territory and for those of us who set betting banks based on historical drawdowns, we are likely to see new historical drawdowns if we haven’t already and therefore, our bank may come under pressure sooner rather than later.

Last season, these systems made over 100pts in a month twice in the space of 6 months and therefore, if any set of systems can bounce back quickly, these are maybe the systems. Then again, if any set of systems can win and lose 100pts in a month, it’s never going to be a smooth journey and until we see more data, I’m slightly worried we may have got sucked into these systems too early. Last season, I didn’t follow these systems at all and watched in envy as huge profits were made across the board and I’m not alone in jumping in with both feet on these systems this season. Therefore, I am slightly anxious having seen this size of loss in the first month but I’ll give it another month and then decide what I want to do. If another big losing month was to hit, bank preservation would possibly kick in and I may look to exit these systems for a loss in my own betting portfolio but we’ll see what happens.

I think that completes the review of the New Systems. A really disappointing month for the systems but the ratings weren’t anywhere near as bad as the systems are making out. It’s difficult though as the ratings are only the lynchpin which the systems are based on but at the end of the day, we need to the ratings to do well for us to make money long-term and algorithm 3 managed it this month. Hence, I would give algorithm 3 pass marks but algorithm 4 was rubbish and the systems overall were a disaster. Roll on next month. 

Misc Systems

No ratings to judge here, so the system results are all that matters (and they were terrible!). Here’s the results:

Given the losses on the previous systems, these systems didn’t stand much of a chance! I would go as far to say that I’m really pleased with the performance of TOX, STOY and STOZ as they lost less than most other systems with a similar level of turnover, so they weren’t quite as bad as some systems. I like the fact that the 3 systems I follow that aren’t rating based still lost badly as it shows that no matter what I did this month, I don’t think I could have avoided a loss.

6-32 has done the impossible and hit 4 winners from 25 bets at odds of 3.17! Horrible month and it will be a long way back for this system this season. Break-even would be looking good I suspect.

21-31 and 6-21-31 have had nearly the same bets (apart from 1 which was a loser as well) and overall, it’s a horrible month. Not much more I can add. I’ve already discussed all 3 algorithms in depth.

One thing I have touched on before on the blog and also in various emails to subscribers is the performance of Home bets on systems TOX, STOY and STOZ. Traditionally, these have been the strongest home bets my systems have found and in a month where Homes have done well, I would hope these bets have done well too! Here are the results of all 6 systems split between Home and Away bets:

I’ll be honest and say that I’m disappointed with the returns from the Homes on these systems this month. Yes, they achieved a profit but compared to some of the algorithms and the returns on Homes,  I would say these under achieved a little this month.  On the other hand, they haven’t had as many Away bets as some systems this month and ultimately, that has meant these systems haven’t had massive losses (it’s all relative when you’re losing like my systems are this month).

System 6-32 has hit 1 winner from 17 away bets at average odds of 3.67. All I can say is WOW! Again, if I was laying these, I wouldn’t think anyone could achieve these sort of results, so to hit this on a backing system is a disaster!  The draw was a little bit of a nuisance on these systems but nothing too bad compared to what I’ve seen in the past.

In terms of historical results, these systems have had a number of poor calendar months but overall, I would say this month has been the worst. Here’s the results from the worst months:

We need to be careful here as the bet number over the last 2 seasons is much higher than previously as the backtested results contained some filters which I removed a few months after the systems went live and I never went back to update the backtested results. However, if we ignore the overall pts lost, we still have the ROI for each month and each betting method.

Although a couple of months have had worse ROI’s, these months were clearly impacted by draws. The AH0.5 loss this month is the highest ever seen on these systems by a long, long way, so that’s as good an indication as any that this month has been a terrible month IMO.

I think I get a little more comfort on these systems given the fact I feel like they have been there and done it before and large losses are not totally unknown in some months.  I guess I have more confidence in the Misc Systems than the New Systems given the fact they have an extra season of live results but my own portfolio is staked that way. I stake more on the Misc Systems than the New Systems, so although my confidence may be better, my P&L is just as bad this month on both sets of systems!

That will do as a review of these systems, don’t want to write the same over and over again. Aways horrible, Homes good. End of.

Under/Over Systems

I didn’t think I’d be saying this 4 weeks ago but these were the best performing systems by far this month. Of course, this is probably linked to the fact that it’s the only set of systems that I don’t have any systems from within my own portfolio, so when things aren’t going your way, it’s like another kick in the balls that these have made a profit.

Here’s the results:

A profit of 4.5pts from 27 bets may not make anyone rich but compared to the other systems this month, it’s just great to see a profit instead of a loss! Said it before a lot on the blog but I personally don’t like these systems, I don’t believe there is that much of a long-term edge and I think it’s probably a slow way to win or lose 1%-2% over the long-term. So far, they have made an ROI of 3.9% over 863 live bets but compared to all my other systems, this is poor and they were no better than break-even last season.

I found out last week someone is following these systems this season and it has saved their month (gulp!) but I’m not sure I’d want to be relying on these systems long-term to make a money from following my ratings. Don’t think they’ll do much harm but not sure they’ll make much money either but we’ll see how they do over the next few months.

Draw Systems

Before I get on with this review, I feel like I should share my own thoughts on these systems. I’ve mentioned this on the blog before and I had a decent exchange on Twitter with a subscriber about the systems but for me personally, I haven’t enjoyed betting on the draw systems. I’m only following a couple of systems and I’m only a couple of points down, so this isn’t me speaking through my pocket but I’m not sure these systems are for me.

It’s a strange one as people may say it’s because of the losing runs and the high average odds but that’s rubbish. The average odds for my footie bets are around a similar price to the draw prices as I follow more Aways than Homes in my portfolio and the Aways are usually decent odds, so it’s not linked to average odds or expected losing runs.

I know this sounds a little bit irrational but I can’t help but feel there is an element of luck involved in this on a game to game basis. I actually feel the same way about the Under/Over bets and I sometimes sit and watch a game where it has 0-0 and Under 2.5 goals written all over it and then in the space of 5 mins, it’s 2-1 and the Unders bet is gone. I think I feel the same way about the draw systems.

Now, I know it’s irrational as the backtested results must have all of this factored in which means the ratings in the long-run don’t really depend on luck in any game to make a profit but I think I’m maybe scarred by the fact that my first decent bet on a draw was ruined by a 97th minute goal to make it 1-0!

Anyway, that’s just my own thoughts but it’s based on nothing but my own thoughts on placing the bets. These systems may end up being the best systems we’ve ever seen and they make me a great profit long-term but in the first month,  I’ve been left a little underwhelmed to say the least.

Bearing in mind what I said in the introduction around the fact that backing draws in all the rated games so far created a 20% loss, I don’t think these results are too bad at all! Here’s the results:

I don’t want to overanalyse such a small sample of bets and therefore, not going to make too many comments. I would probably come across as biased anyway given what I just wrote above and therefore, let’s give these systems a little more time to see how things pan out.

If the late goal that occurred in the month had not happened, would have basically been a break-even month to kick off their first live month and that would have been a fair result. Not sure the current results are a true representation of the month for these systems but we’ll learn more as the season goes on.

Ok, I think that’s enough of a review for a single month. Some people analyse a season in less words than I’ve analysed a month! I did start reducing the length of the reviews as I worked through this as there is only so many times I can tell you that Homes had a good month and Aways had their worst month ever. Could have written the review in 5 mins. ;)

I’m not sure about any of you reading but I personally feel much better about the month we’ve just encountered. Yes, it’s the largest monthly loss I’ve ever suffered personally from betting, yes, it’s the worst month the service has ever had, yes, anyone following the systems this month lost a massive amount of money but at the end of the day, when you are backing mostly Aways where all Aways are heavily loss making, it is always going to make things appear worse than they are in reality.

We’ll see what happens next month and hopefully we see a bounce back from the Away bets next month to correct the P&Ls of many of these systems!

No comments:

Post a Comment