I mentioned to Justin in a comment earlier this week that I’d try to take a quick look at the weekend games to see if any bets stand out. Of course, my ratings can’t work until at least 4 games take place and I don’t start using them until the first 6 games take place, so I can’t actually model anything using data from this season.
However, if I assume that the teams just continued on from from last season, then I can get something from my ratings at the end of last season, adjusted for the season's performance to data. I wouldn’t place much emphasis on it though as clearly, teams just don’t start the next season the way they finished last season. There could be massive changes to any team and of course, fitness, pre-season form, managerial changes etc. etc.
Anyway, for a bit of fun, I’ve had a look at the SPL games. Surprisingly, there is actually a very decent value bet thrown up.
I have ICT at around 4/6 against Hibernian this weekend whereas they are available at 11/8.
Looking at why this has happened, I have ICT as one of the form teams in the SPL at home. There are too many stats to quote but the key stat looking at the ratings is that in their last 4 games against bottom half opposition at home, they’ve won 2-0,1-0,3-0 and 2-0. Against top half opposition away from home, Hibs have lost 8 of their last 9 games, including losing their last 6. These two factors combined appear to be throwing up a high % chance of an ICT win.
All the other rating factors also point to Hibs struggling to muster too many shots on goal away from home and combined with ICT’s strong home defence, it is all pointing to a strong home win.
To put it in perspective though, ICT played their first game this season away to Motherwell and lost 3-0. They mustered 2 shots on target in the game. Hardly encouraging.
However, Hibs lost 2-0 at home to Celtic and couldn’t muster a shot on goal there.
Based on the scores from last weekend, I’m not sure you could seriously contemplate backing ICT but if we assume the teams perform like they did last season, ICT will play much better at home this weekend and Hibs will struggle badly away from home.
I would have been all over ICT last season at the end of the season but happy to bide my time until the first 6 games are played this season. Incidentally, this is one of the games where DNB is also showing as value but this is down to the high home % rather than the draw being value.
I'm glad I don't need to give a write up for every game I give out next season. I'd never get the bets out each weekend!
Thursday, 28 July 2011
Wednesday, 27 July 2011
Review of The Football Analyst
I've noticed that Mat Hare has posted an independant review of my systems from last season on his excellent blog.
http://drownintheriver.wordpress.com/2011/07/27/the-football-analyst-analysed/
It's great to have another person look at the systems as it gives me some more confidence around what I'm doing although admittedly, many of Mat's conclusions are the same as mines on here.
How much weight do you put on the run experienced by the systems in February and April last season?
I'm not sure I can answer that but all I know is that before February came along, I thought I was onto something pretty special here, something that you maybe wait a lifetime to come across. Unfortunately, 3 months later, I was doubting whether my systems even had an edge to tell the truth.
Depending on whether you are an optimist or not, you can look at that as a blip or a start of a trend.
Before Feb-11, the systems had experienced 18 winning months during backtesting and live results, with an average profit of 48pts and an overall ROI of 20%. It was the best run ever experienced by the systems and thankfully, I caught the first 5 months last season which averaged 60pts profit and in excess of 20% ROI.
So, we had the best run ever followed by the worst run ever.
What will happen next season?
http://drownintheriver.wordpress.com/2011/07/27/the-football-analyst-analysed/
It's great to have another person look at the systems as it gives me some more confidence around what I'm doing although admittedly, many of Mat's conclusions are the same as mines on here.
How much weight do you put on the run experienced by the systems in February and April last season?
I'm not sure I can answer that but all I know is that before February came along, I thought I was onto something pretty special here, something that you maybe wait a lifetime to come across. Unfortunately, 3 months later, I was doubting whether my systems even had an edge to tell the truth.
Depending on whether you are an optimist or not, you can look at that as a blip or a start of a trend.
Before Feb-11, the systems had experienced 18 winning months during backtesting and live results, with an average profit of 48pts and an overall ROI of 20%. It was the best run ever experienced by the systems and thankfully, I caught the first 5 months last season which averaged 60pts profit and in excess of 20% ROI.
So, we had the best run ever followed by the worst run ever.
What will happen next season?
Tuesday, 26 July 2011
How does this sound?
One thing that I have been doing a lot of in the last few weeks or so is running different scenarios for a few people for next season. People are trying to build portfolios using my systems for next season and due to the fact I'm a pretty good analyst (self praise is no praise!), then I've been trying to help out where I can.
Clearly, this is a bit of an issue with the way the current systems are set up to work. Only I have access to all my historical results data and therefore, people need to rely on me to run scenarios for them and to try to understand how certain systems may work in a portfolio or not.
If I wind the clock back a little, I saw this football project as a 3 year project at the outset. The first year was about building ratings and systems that work. The second year was about consolidating what I'd done and building some more ratings/systems that work and the 3rd year was going to be about looking at ways to try to maximise profits from the systems for everyone.
I think I'm well into the 2nd year now (I started out in Jan-10) and therefore, I think I'm pretty much done with tweaking ratings and building new systems. Assuming next season went well and all my systems performed as I'd like them too, I'd have 20 unique systems to go forward with from the end of the season. 11 systems would have two season's of proofed results and 9 systems would have a season's proofed results.
The next question then would be, how do we use these systems and how do people build their own portfolios to suit their own risk aversion using these systems?
I'm the sort of person that always has ideas floating around their mind and you won't be surprised to hear I'm already thinking about how this would work. Of course, I'm not going to even contemplate looking at this until the end of next season as I don't want to get too far ahead of the game. At the moment, I only have 11 systems proven over one season. It's a big leap to assume that these will all perform as I expect next season and that I won't hit a problem or two along the way.
If we can just imagine that next season goes really well though, then what is the next logical step for this project?
Well, it has to be to find a way to allow people to build their own portfolios using these systems. I'm sort of seeing it already but people could have a drop down box of all the systems and they can pick what systems they want to follow. They can then pick whether they want to follow H/A or DNB for the system and of course, they could decide to back homes outright and cover the draw for away teams. They can then select their stakes for Homes and Aways and ss if by magic, the results would appear which would show the individual the portfolio performance over time, the returns achieved by month,season,league etc. It would also show KPIs such as longest losing run, largest drawdown and then calculate a betting bank based on how safe the person wants to be.
This would allow people to sit and compare lots of different portfolio's until they get a portfolio that suits their needs.
Even when I think about this now, it sounds great and it's the sort of thing that I can really see helping people make informed decisions around which systems to follow and how to best maximise their profits.
Anyway, I thought I'd just throw this out there as an idea of where this venture could be going in a season's time. Allowing everyone to choose a portfolio that they'd like to follow. I'm sure I'll continue to keep tweaking existing systems and probably developing more systems as time progresses but I think the next big leap has to involve helping others use these systems to maximise their own returns.
Sound like a good idea?
Clearly, this is a bit of an issue with the way the current systems are set up to work. Only I have access to all my historical results data and therefore, people need to rely on me to run scenarios for them and to try to understand how certain systems may work in a portfolio or not.
If I wind the clock back a little, I saw this football project as a 3 year project at the outset. The first year was about building ratings and systems that work. The second year was about consolidating what I'd done and building some more ratings/systems that work and the 3rd year was going to be about looking at ways to try to maximise profits from the systems for everyone.
I think I'm well into the 2nd year now (I started out in Jan-10) and therefore, I think I'm pretty much done with tweaking ratings and building new systems. Assuming next season went well and all my systems performed as I'd like them too, I'd have 20 unique systems to go forward with from the end of the season. 11 systems would have two season's of proofed results and 9 systems would have a season's proofed results.
The next question then would be, how do we use these systems and how do people build their own portfolios to suit their own risk aversion using these systems?
I'm the sort of person that always has ideas floating around their mind and you won't be surprised to hear I'm already thinking about how this would work. Of course, I'm not going to even contemplate looking at this until the end of next season as I don't want to get too far ahead of the game. At the moment, I only have 11 systems proven over one season. It's a big leap to assume that these will all perform as I expect next season and that I won't hit a problem or two along the way.
If we can just imagine that next season goes really well though, then what is the next logical step for this project?
Well, it has to be to find a way to allow people to build their own portfolios using these systems. I'm sort of seeing it already but people could have a drop down box of all the systems and they can pick what systems they want to follow. They can then pick whether they want to follow H/A or DNB for the system and of course, they could decide to back homes outright and cover the draw for away teams. They can then select their stakes for Homes and Aways and ss if by magic, the results would appear which would show the individual the portfolio performance over time, the returns achieved by month,season,league etc. It would also show KPIs such as longest losing run, largest drawdown and then calculate a betting bank based on how safe the person wants to be.
This would allow people to sit and compare lots of different portfolio's until they get a portfolio that suits their needs.
Even when I think about this now, it sounds great and it's the sort of thing that I can really see helping people make informed decisions around which systems to follow and how to best maximise their profits.
Anyway, I thought I'd just throw this out there as an idea of where this venture could be going in a season's time. Allowing everyone to choose a portfolio that they'd like to follow. I'm sure I'll continue to keep tweaking existing systems and probably developing more systems as time progresses but I think the next big leap has to involve helping others use these systems to maximise their own returns.
Sound like a good idea?
Monday, 25 July 2011
Terrible, terrible, terrible
I haven't tended to answer anonymous comments or even post anonymous comments lately on the blog but there was an intriguing comment posted yesterday which I think deserves a response.
The comment was:
On paper,this looks great but do you seriously think you can do as well next season?no, thought not.i'll be following with interest.
I always hate it when people ask me questions and answer them on my behalf in the same breath. Putting this annoyance to the side, I think this is a fair question. On paper, the systems and suggested portfolios all look very good but do I really think these returns are achievable?
Before I answer this, it's probably worth sharing a little more info about the systems and why they are built the way they're built. When I started out looking at building systems, I had one idea in my head. It was to achieve a return of 10% per season. Nothing more, nothing less.
That was my aim when I was building systems and it was also my aim all of last season when I was following bets produced by the systems.
Clearly, people can look at all of my systems and see that the returns are much higher than 10% for all systems during backtesting and backfitting. Even during the live season last season, many of the systems had an ROI in excess of 10% and for the majority of the season, most systems had ROI's above 20% although the ROI dropped significantly from February onwards.
This begs the question, why create systems with returns of 20%+ when you are only after a target return of 10%?
Well, it's hardly rocket science but everyone knows how difficult it can be to get systems to perform in line with past results. I've read about lots of systems that looked amazing when they were backtested and yet, when they went live, they've made substantial losses. Therefore, to combat this, I've DELIBERATELY created systems with very high returns.
Everyone knows that systems 6,7,8,21 and 22 had a poor season last season. They ONLY achieved a return between 27% and 37% of the returns achieved during the previous season (which were purely backtested results). Hence, these systems only achieved around a 1/3 of the return I thought they'd achieve last season.
Therefore, I was very disappointed with these systems as I've said numerous times on here.
However, when you look at the returns these systems last season, they ONLY created a return of 5.5%. They had a terrible season and produced a very low return of 5.5%.
Terrible, terrible, terrible. Only words to describe that low level of return they created. Waste of time I guess.
Anyway, you can see where I'm going with this......
Is 5.5% return so bad for a set of systems that performed terribly last season? I'll let all of you reading the blog decide that for yourselves.
I was aiming for a return of 10% last season with these systems, they had a terrible season (second half of season really) and yet, they produced a profit of 5.5%.
Simply, by aiming for something much greater than what you'd be happy with, then you can still achieve a decent profit even when things go wrong. If I had been following systems with an expected profit of 10% last season, I might have managed to achieve a return of between 2% and 3% last season.
The other reason for following systems with such a high return is that it is a long way from an ROI of 20% to making a loss on any system. Last season, some of my systems had a poor season but they still managed to make a profit.
So, to answer the question I set out answering......do I think these portfolio returns can be as good next season as they were last season?
Well, last season wasn't a great year as you can see by looking at the returns and last season, the returns of the portfolio were only 60% of the returns of the previous season. Hence, it was a disappointing season last season for the portfolios I'll be following next season.
What happens if they have an even worse season next season. Instead of only being 60% returns of the previous season, what if the returns are only half as good as last season (that's 50% anonymous!).
Last season, the portfolios made a return of 22.1% overall.
Next season, they might have a disaster of a season and only make a return of 11%. Terrible, terrible, terrible.
I think I'll probably pack in this game if my portfolio could only make an 11% return next season from nearly 1,500 bets I guess.
So, to answer your question anonymous, do I think I can achieve the same returns next season for my portfolio? In all honesty, I don't know but if the returns are half as good as last season, that will do me just fine.
The comment was:
On paper,this looks great but do you seriously think you can do as well next season?no, thought not.i'll be following with interest.
I always hate it when people ask me questions and answer them on my behalf in the same breath. Putting this annoyance to the side, I think this is a fair question. On paper, the systems and suggested portfolios all look very good but do I really think these returns are achievable?
Before I answer this, it's probably worth sharing a little more info about the systems and why they are built the way they're built. When I started out looking at building systems, I had one idea in my head. It was to achieve a return of 10% per season. Nothing more, nothing less.
That was my aim when I was building systems and it was also my aim all of last season when I was following bets produced by the systems.
Clearly, people can look at all of my systems and see that the returns are much higher than 10% for all systems during backtesting and backfitting. Even during the live season last season, many of the systems had an ROI in excess of 10% and for the majority of the season, most systems had ROI's above 20% although the ROI dropped significantly from February onwards.
This begs the question, why create systems with returns of 20%+ when you are only after a target return of 10%?
Well, it's hardly rocket science but everyone knows how difficult it can be to get systems to perform in line with past results. I've read about lots of systems that looked amazing when they were backtested and yet, when they went live, they've made substantial losses. Therefore, to combat this, I've DELIBERATELY created systems with very high returns.
Everyone knows that systems 6,7,8,21 and 22 had a poor season last season. They ONLY achieved a return between 27% and 37% of the returns achieved during the previous season (which were purely backtested results). Hence, these systems only achieved around a 1/3 of the return I thought they'd achieve last season.
Therefore, I was very disappointed with these systems as I've said numerous times on here.
However, when you look at the returns these systems last season, they ONLY created a return of 5.5%. They had a terrible season and produced a very low return of 5.5%.
Terrible, terrible, terrible. Only words to describe that low level of return they created. Waste of time I guess.
Anyway, you can see where I'm going with this......
Is 5.5% return so bad for a set of systems that performed terribly last season? I'll let all of you reading the blog decide that for yourselves.
I was aiming for a return of 10% last season with these systems, they had a terrible season (second half of season really) and yet, they produced a profit of 5.5%.
Simply, by aiming for something much greater than what you'd be happy with, then you can still achieve a decent profit even when things go wrong. If I had been following systems with an expected profit of 10% last season, I might have managed to achieve a return of between 2% and 3% last season.
The other reason for following systems with such a high return is that it is a long way from an ROI of 20% to making a loss on any system. Last season, some of my systems had a poor season but they still managed to make a profit.
So, to answer the question I set out answering......do I think these portfolio returns can be as good next season as they were last season?
Well, last season wasn't a great year as you can see by looking at the returns and last season, the returns of the portfolio were only 60% of the returns of the previous season. Hence, it was a disappointing season last season for the portfolios I'll be following next season.
What happens if they have an even worse season next season. Instead of only being 60% returns of the previous season, what if the returns are only half as good as last season (that's 50% anonymous!).
Last season, the portfolios made a return of 22.1% overall.
Next season, they might have a disaster of a season and only make a return of 11%. Terrible, terrible, terrible.
I think I'll probably pack in this game if my portfolio could only make an 11% return next season from nearly 1,500 bets I guess.
So, to answer your question anonymous, do I think I can achieve the same returns next season for my portfolio? In all honesty, I don't know but if the returns are half as good as last season, that will do me just fine.
Friday, 22 July 2011
Portfolio Performance
As I said in my last post, I was going to take a look at how my own personal portfolio would have performed in season's gone past if I had been following the systems I'm following this season and using the same staking plan.
Now, before I get onto this, a few caveats.
This is only one idea of a portfolio. With 20 individual systems this season and 2 ways to play each systems (H/A and DNB), that means there are 40 possible systems. Depending on how many systems you want to play, there are clearly millions of different combinations of portfolios. Throw in the fact that I'm playing higher stakes on homes than aways in one of my portfolios and you can see that the number of portfolios is infinite.
Clearly, there is absolutely nothing to say that this portfolio is any better than any other portfolio. I haven't tried to find a portfolio that works historically and decided I'm going to use this. I've decided which systems I want to follow based on historical results, decided how to follow the systems to suit my risk aversion and simply, combined these systems to form a portfolio. I'm not suggesting for one minute that this is an optimal way to create a portfolio. It's how I've created my portfolio and nothing more.
Last but not least, past performance can only ever be an indicative guide of the future performance.
OK, caveats out the way, how do the results look?
Well, I'm quite pleased with how the portfolio shapes up historically. This table shows the results by season for both portfolios. Portfolio one is based on the systems from last season and portfolio two is based on the new systems from next season.
As you can see, I've dropped the first two season's from these results. I personally don't place a lot of weight on the performance of my systems in the 06/07 and 07/08 seasons. I know there is some backfitted results included and therefore, to help set my expectations, I always ignore any season's which contain backfitted results.
The first thing that stands out is the fact that I'm playing far fewer bets in portfolio two. This is how I wanted it to look as these bets are much more unproven and therefore, I don't want to get to the stage whereby my established systems make me money and my new systems lose me all of this money.
I mentioned in my last post that I would play lower stakes on the portfolio two bets but I'm not sure I need to now I've looked at the number of bets. I basically wanted portfolio one to carry around 2/3 of the risk next season and by playing 1pt on each bet in both portfolios, this can be achieved.
Interestingly, portfolio two did better than portfolio one last season but I can put that down to the fact portfolio two used DNB much more last season. Portfolio one only uses DNB on aways priced 5/2 or greater.
The next interesting thing to look at is the results on a monthly basis.
As I also said in my last post, I was keen to see how bad portfolio one did during February and April as the systems had a nightmare in both months. As you can see, portfolio one had a nightmare in both months too. After winning an average of 50pts for the first 4 months of the season, the portfolio lost 27pts in February and another 51pts in April. OUCH!
On the bright side though, they won all of February's losses in March and they did rebound a little in May too. However, I can't get away from the fact the portfolio made no profit in the 4 months from Jan-11.
Interestingly, portfolio two didn't suffer anywhere near as badly. This portfolio suffered a loss in December and a small loss in February but somehow, the portfolio managed to dodge the substantial loss created by the established systems in April. Why is this? Simply, the second portfolio uses a lot more DNB betting and therefore, when we suffer from an obscene number of draws as we did in April, the portfolio doesn't suffer nearly as much. Or in this case, it can even make a profit when there are a lot of draws!
However, in the first 4 months of last season, portfolio two did nowhere near as well as portfolio one, so it has to be kept in perspective.
For those that prefer pictures.....
I think these pictures paint an interesting picture. People can rightly point to portfolio one and the fact it has made very little profit over the last 5 months of trading. However, I find it quite reassuring that portfolio two suffered a near identical 5 month spell during the 07/08 season and yet, the portfolio has bounced back over the last couple of seasons.
Clearly, looking at the graphs, you could argue that portfolio two should maybe carry more confidence but then again, I know portfolio one has been there and done it last season whereas portfolio two is all about backtested results and it has no real life experience yet!
I hope the above gives a little insight into how my own plans are shaping up for next season. Not long to go until we get to the first bets of the season now........(just another 6 weeks!)
Now, before I get onto this, a few caveats.
This is only one idea of a portfolio. With 20 individual systems this season and 2 ways to play each systems (H/A and DNB), that means there are 40 possible systems. Depending on how many systems you want to play, there are clearly millions of different combinations of portfolios. Throw in the fact that I'm playing higher stakes on homes than aways in one of my portfolios and you can see that the number of portfolios is infinite.
Clearly, there is absolutely nothing to say that this portfolio is any better than any other portfolio. I haven't tried to find a portfolio that works historically and decided I'm going to use this. I've decided which systems I want to follow based on historical results, decided how to follow the systems to suit my risk aversion and simply, combined these systems to form a portfolio. I'm not suggesting for one minute that this is an optimal way to create a portfolio. It's how I've created my portfolio and nothing more.
Last but not least, past performance can only ever be an indicative guide of the future performance.
OK, caveats out the way, how do the results look?
Well, I'm quite pleased with how the portfolio shapes up historically. This table shows the results by season for both portfolios. Portfolio one is based on the systems from last season and portfolio two is based on the new systems from next season.
As you can see, I've dropped the first two season's from these results. I personally don't place a lot of weight on the performance of my systems in the 06/07 and 07/08 seasons. I know there is some backfitted results included and therefore, to help set my expectations, I always ignore any season's which contain backfitted results.
The first thing that stands out is the fact that I'm playing far fewer bets in portfolio two. This is how I wanted it to look as these bets are much more unproven and therefore, I don't want to get to the stage whereby my established systems make me money and my new systems lose me all of this money.
I mentioned in my last post that I would play lower stakes on the portfolio two bets but I'm not sure I need to now I've looked at the number of bets. I basically wanted portfolio one to carry around 2/3 of the risk next season and by playing 1pt on each bet in both portfolios, this can be achieved.
Interestingly, portfolio two did better than portfolio one last season but I can put that down to the fact portfolio two used DNB much more last season. Portfolio one only uses DNB on aways priced 5/2 or greater.
The next interesting thing to look at is the results on a monthly basis.
As I also said in my last post, I was keen to see how bad portfolio one did during February and April as the systems had a nightmare in both months. As you can see, portfolio one had a nightmare in both months too. After winning an average of 50pts for the first 4 months of the season, the portfolio lost 27pts in February and another 51pts in April. OUCH!
On the bright side though, they won all of February's losses in March and they did rebound a little in May too. However, I can't get away from the fact the portfolio made no profit in the 4 months from Jan-11.
Interestingly, portfolio two didn't suffer anywhere near as badly. This portfolio suffered a loss in December and a small loss in February but somehow, the portfolio managed to dodge the substantial loss created by the established systems in April. Why is this? Simply, the second portfolio uses a lot more DNB betting and therefore, when we suffer from an obscene number of draws as we did in April, the portfolio doesn't suffer nearly as much. Or in this case, it can even make a profit when there are a lot of draws!
However, in the first 4 months of last season, portfolio two did nowhere near as well as portfolio one, so it has to be kept in perspective.
For those that prefer pictures.....
I think these pictures paint an interesting picture. People can rightly point to portfolio one and the fact it has made very little profit over the last 5 months of trading. However, I find it quite reassuring that portfolio two suffered a near identical 5 month spell during the 07/08 season and yet, the portfolio has bounced back over the last couple of seasons.
Clearly, looking at the graphs, you could argue that portfolio two should maybe carry more confidence but then again, I know portfolio one has been there and done it last season whereas portfolio two is all about backtested results and it has no real life experience yet!
I hope the above gives a little insight into how my own plans are shaping up for next season. Not long to go until we get to the first bets of the season now........(just another 6 weeks!)
Thursday, 21 July 2011
My Strategy......
One thing I have been putting off until now is sharing any information my own betting strategy for next season. One thing I'm always wary of is people thinking that I always know the answers of how to use the systems to maximise the potential returns but I showed myself last season that I'm no better than anyone else when it comes to the gambling side of things. I made some truly awful decisions last season about when to increase stakes and ultimately, my returns bared no resemblance to the returns from the systems.
I wrote a good post on the subject here which was for my benefit as much as anyone else's benefit. I have to learn from my mistakes last season if I want to do better this season.
http://the-football-analyst.blogspot.com/2011/06/lessons-learnt-from-last-season.html
However, I'm in the tricky situation here of being the person who has to do all the analysis and all the results updating and ultimately, I'm really the sort of person who struggles to take an interest in things unless I have some money riding on the outcome. This sounds very unprofessional really but it's an honest assessment of me as a person. I've never paper traded a system in my life and even when I was learning to trade on Betfair a few years ago, I would be doing it with £2 stakes rather than sitting there, pretending I was backing and laying a horse.
Therefore, similar to last season, I need to probably have more bets than I should really have. If I decided to just watch my new systems this season, I know for a fact that if they started badly, I'd have thoughts about dropping them from the portfolio or trying to change them mid season as at the end of the day, I'd have no money riding on them.
Taking into account all of the above, what's my strategy for this season?
I know some people followed me very closely in the TFA forum last season and when I was increasing/reducing stakes, they were following me closely and being honest, as a result of me making a mess of things, they made a mess of things too. Therefore, I'm not going to be writing a big post on this blog stating what systems I'm playing, what % of my bank I'm placing on each system, what bet size I'm having, what betting banks I'm using and so on as I don't really want people to start this season using the same methodology as me. For example, I might change strategy after a month depending on how things go and I'd then have to tell others who started with the same banks as me to change strategy too.
I'll keep my strategy very high level and leave it to others to come up with their own strategy.
Firstly, I won't be placing any money on systems 6,7,8,21,22 this season. I firmly see these systems as feeder systems for my combined systems now (6-21 etc) and therefore, I know that the bets that only appear on one of the two algorithms are actually weaker bets than bets that appear on both algorithms.
This may sound like a fairly obvious change but system 8 is my favourite system of all the systems I've built. When I first developed system 8, I loved looking at the backtested and backfitted results as they were better than anything I'd ever seen before for a gambling system I'd created. Of course, when I started combining systems, system 8 suddenly looked like an average system but I'll never forget my excitement in the TFA forum when system 8 was first developed. Last season, the system made a return of 9% and thankfully, that was the system that I told most people to follow last season, so anyone who followed my advice couldn't help but have an OK season last season.
In this betting game though, there is no room for sentiment and systems 6,7,8,21,22 are taking a break from my betting portfolio this season. I obviously hope these are very successful next season (they do feed all my other systems!) but as a system in their own right, I won't be backing any of these systems.
My main portfolio then will involve systems 6-21,6-22,7-21,7-22,8-21,8-22. I've not totally decided on my staking plan yet as I'm still playing about with numbers for betting banks and so on. At the moment, it looks like I will be playing 100% higher stakes on the home bets than the away bets (thanks to Grant for that observation via the blog!) but I will be covering the draw on the higher priced away sides. I don't believe covering the draw will maximise my profits but I do think it will help my mental state a little.
Last season, there were numerous times when I was on 3/1+ away teams who were leading with 10 minutes to go who only ended up drawing the game. The swing was quite substantial in these games and I discussed it on the blog a few times. Instead of a 20pt win, I would suffer a 5pt loss or something.
Clearly, if I use DNB on these high odds teams, if they draw, I will get my stake returned and if they win, I will win less. If they lose, I'll lose no matter what way I backed them. I've not decided on the cut-off for the odds for covering the draw as yet but I'll look into this a little further when time allows. There is no right answer anyway to this question, it's very subjective.
Again, I can only stress that I don't believe this is an optimal strategy to maximise profits. This is purely a psychological thing to help keep me sane this season and will hopefully avoid losing months like I suffered in February and April last sesaon.
In terms of the new systems, I will also operate these in a separate portfolio. However, this portfolio will sit outside my main portfolio and will be played with lower stakes. I'm still working through exactly what betting bank I need and what stakes I will use but I reckon the stakes will be around 50% of the stakes on the main systems.
However, to reduce the variance further, I will be using DNB on systems involving 31-33 but for systems TOX, STOY and STOZ, these will be win only.
At the moment, it looks like the portfolio will be System 33, system 6-32, system 6-21-31, STOY and STOZ, The first 3 of these will be DNB and the last two will be win only with no draw coverage.
In total then, I'll be following 11 of the 20 systems I'll have live for next season. As I said above, I think this is too many possibly for anyone else to follow but I'm different in the sense I need to follow a range of systems to keep my enthusiasm going as the season continues. It's a long season!
With the mixture of systems and draw coverage, I'm happy this gives me something to go to war with. Importantly, I'll be going to war with a lot more confidence than I had last season, so if things do go well before Christmas, I won't need to be increasing stakes!
I will use level staking throughout the season (I promised myself this last season before I increased stakes!) and I won't tinker with the portfolio too much as the season progresses. I might change the staking depending on how the new systems shape up but I won't be using any sort of variable staking plan.
I'm still working on a few things for my own portfolios next season but once I'm done, I'll post up the results my portfolios would have achieved under my current staking plans and portfolio set up if they had been alive last season. I'm curious to see what sort of drawdown I'd have suffered in February and April when the systems went off the rails as I reckon I'd have lost a lot of money during this time!
Anyway, I hope the above gives a little insight into what I'm doing personally next season. I'm lucky in the sense I have 15 bookmaker accounts I tend to use and I could probably open another 30-40 if I went away from the more traditional bookmakers I'm using. In addition, I could use Betfair too if I wanted and lastly, if I wanted to, I could use my partner's details and open up another 40 bookmaker accounts anyway! I realise I'm fortunate in this sense but then again, I only opened my first online bookmakers account last Summer, so I'm hardly going to have any closed accounts. I only have restrictions on a couple of them as a result of last season's performance.
Hopefully after next season, I'll have more difficulty getting my bets on. :)
I wrote a good post on the subject here which was for my benefit as much as anyone else's benefit. I have to learn from my mistakes last season if I want to do better this season.
http://the-football-analyst.blogspot.com/2011/06/lessons-learnt-from-last-season.html
However, I'm in the tricky situation here of being the person who has to do all the analysis and all the results updating and ultimately, I'm really the sort of person who struggles to take an interest in things unless I have some money riding on the outcome. This sounds very unprofessional really but it's an honest assessment of me as a person. I've never paper traded a system in my life and even when I was learning to trade on Betfair a few years ago, I would be doing it with £2 stakes rather than sitting there, pretending I was backing and laying a horse.
Therefore, similar to last season, I need to probably have more bets than I should really have. If I decided to just watch my new systems this season, I know for a fact that if they started badly, I'd have thoughts about dropping them from the portfolio or trying to change them mid season as at the end of the day, I'd have no money riding on them.
Taking into account all of the above, what's my strategy for this season?
I know some people followed me very closely in the TFA forum last season and when I was increasing/reducing stakes, they were following me closely and being honest, as a result of me making a mess of things, they made a mess of things too. Therefore, I'm not going to be writing a big post on this blog stating what systems I'm playing, what % of my bank I'm placing on each system, what bet size I'm having, what betting banks I'm using and so on as I don't really want people to start this season using the same methodology as me. For example, I might change strategy after a month depending on how things go and I'd then have to tell others who started with the same banks as me to change strategy too.
I'll keep my strategy very high level and leave it to others to come up with their own strategy.
Firstly, I won't be placing any money on systems 6,7,8,21,22 this season. I firmly see these systems as feeder systems for my combined systems now (6-21 etc) and therefore, I know that the bets that only appear on one of the two algorithms are actually weaker bets than bets that appear on both algorithms.
This may sound like a fairly obvious change but system 8 is my favourite system of all the systems I've built. When I first developed system 8, I loved looking at the backtested and backfitted results as they were better than anything I'd ever seen before for a gambling system I'd created. Of course, when I started combining systems, system 8 suddenly looked like an average system but I'll never forget my excitement in the TFA forum when system 8 was first developed. Last season, the system made a return of 9% and thankfully, that was the system that I told most people to follow last season, so anyone who followed my advice couldn't help but have an OK season last season.
In this betting game though, there is no room for sentiment and systems 6,7,8,21,22 are taking a break from my betting portfolio this season. I obviously hope these are very successful next season (they do feed all my other systems!) but as a system in their own right, I won't be backing any of these systems.
My main portfolio then will involve systems 6-21,6-22,7-21,7-22,8-21,8-22. I've not totally decided on my staking plan yet as I'm still playing about with numbers for betting banks and so on. At the moment, it looks like I will be playing 100% higher stakes on the home bets than the away bets (thanks to Grant for that observation via the blog!) but I will be covering the draw on the higher priced away sides. I don't believe covering the draw will maximise my profits but I do think it will help my mental state a little.
Last season, there were numerous times when I was on 3/1+ away teams who were leading with 10 minutes to go who only ended up drawing the game. The swing was quite substantial in these games and I discussed it on the blog a few times. Instead of a 20pt win, I would suffer a 5pt loss or something.
Clearly, if I use DNB on these high odds teams, if they draw, I will get my stake returned and if they win, I will win less. If they lose, I'll lose no matter what way I backed them. I've not decided on the cut-off for the odds for covering the draw as yet but I'll look into this a little further when time allows. There is no right answer anyway to this question, it's very subjective.
Again, I can only stress that I don't believe this is an optimal strategy to maximise profits. This is purely a psychological thing to help keep me sane this season and will hopefully avoid losing months like I suffered in February and April last sesaon.
In terms of the new systems, I will also operate these in a separate portfolio. However, this portfolio will sit outside my main portfolio and will be played with lower stakes. I'm still working through exactly what betting bank I need and what stakes I will use but I reckon the stakes will be around 50% of the stakes on the main systems.
However, to reduce the variance further, I will be using DNB on systems involving 31-33 but for systems TOX, STOY and STOZ, these will be win only.
At the moment, it looks like the portfolio will be System 33, system 6-32, system 6-21-31, STOY and STOZ, The first 3 of these will be DNB and the last two will be win only with no draw coverage.
In total then, I'll be following 11 of the 20 systems I'll have live for next season. As I said above, I think this is too many possibly for anyone else to follow but I'm different in the sense I need to follow a range of systems to keep my enthusiasm going as the season continues. It's a long season!
With the mixture of systems and draw coverage, I'm happy this gives me something to go to war with. Importantly, I'll be going to war with a lot more confidence than I had last season, so if things do go well before Christmas, I won't need to be increasing stakes!
I will use level staking throughout the season (I promised myself this last season before I increased stakes!) and I won't tinker with the portfolio too much as the season progresses. I might change the staking depending on how the new systems shape up but I won't be using any sort of variable staking plan.
I'm still working on a few things for my own portfolios next season but once I'm done, I'll post up the results my portfolios would have achieved under my current staking plans and portfolio set up if they had been alive last season. I'm curious to see what sort of drawdown I'd have suffered in February and April when the systems went off the rails as I reckon I'd have lost a lot of money during this time!
Anyway, I hope the above gives a little insight into what I'm doing personally next season. I'm lucky in the sense I have 15 bookmaker accounts I tend to use and I could probably open another 30-40 if I went away from the more traditional bookmakers I'm using. In addition, I could use Betfair too if I wanted and lastly, if I wanted to, I could use my partner's details and open up another 40 bookmaker accounts anyway! I realise I'm fortunate in this sense but then again, I only opened my first online bookmakers account last Summer, so I'm hardly going to have any closed accounts. I only have restrictions on a couple of them as a result of last season's performance.
Hopefully after next season, I'll have more difficulty getting my bets on. :)
Wednesday, 20 July 2011
Example of Results for Next Season
Just a very quick post tonight. I thought it was a good question from Chris previously about how the bets would look for next season. I know it is pretty basic stuff but I personally would want to know how the bets would look before I decided if I was going to follow the bets or not.
If I go one step beyond that though, the next logical thought would be, how will the results be shown after each set of bets?
Well, if anyone followed the blog last season, you’ll know that I used a pivot table to show the results after each set of games. I don’t see any need to do anything different this season if I’m honest.
It’s always hard to know exactly what information anyone would want to see in the results. At the end of the day, we’re only interested in whether the bets won or lost but of course, next season, we’ll be tracking the profitability of DNB along with H/A, so it’s important I can show both returns for each game a system has picked.
After a fair bit of thought, I’m going to have two results tables for next season. I’ll have the systems that were proofed last season (the established systems) and I’ll have the systems that are new from this season. I don’t want the returns from these systems mixed up next season, so it’s important that I report back on them separately.
In terms of what the results will look like, here is the template that I will use. I think it captures everything I want to show for my results each week but if people want anything else added, give me a shout.
Example of Bets for Next Season
One thing I have been doing over the past few weeks or so is getting myself prepared for next season. You won't believe the number of spreadsheets I have on the go and I reckon that if I was any good at databases, all of my stuff should be kept in databases as they cope much better with large quantities of data. Unfortunately, my Access skills are nowhere near as good as my Excel skills, so I'm sort of stuck with Excel I'm afraid.
My main issue now is simply the size of the sheets as doing analysis involving pivot tables (which is how all my analysis is done) causes the spreadsheets to be quite large. It's so bad now, when I'm transferring spreadsheets between work and home, I have to send a spreadsheet at a time (even though it is zipped!) which is a farce really.
There was a comment on the blog last night from Chris asking how the bets would look like for next season. Reading between the lines, I'm assuming he means what the betting sheet will look like I post up here with the bets. He mentions if I'll be quoting AH odds along with the odds for the team to win also.
I've discussed this before but I am sticking with two ways to follow these system bets next season. I will monitor the profitability of the bets to backing the selections outright and also backing the selections DNB (covering the draw). I won't look at any other Asian handicaps although I appreciate these type of bets appeal to some people more than others.
Just on the Asian Handicap bets point, one of my other readers Mark sent me some data for AH betting for the leagues where I'm playing in. It was the AH odds for all the games for the last 3 seasons. He asked me to do a little bit of work to look at this and I spent over 4 hours on Saturday night coming up with an analysis of the performance of all my systems using these odds. I've already said to Mark I wouldn't post it on the blog but I did send Mark the analysis. I feel like this blog has enough going on without me introducing AH analysis as trying to intrepret the results is difficult.
A high level summary would be that AH betting is still very profitable with these systems but OVERALL, the returns are lower than just backing the selections to win. There were a lot of different types of AH bets and clearly, some were more profitable than outright betting for some systems but that's more to do with small samples of bets, more than anything else.
I've already looked at double chance betting and shown that it is less profitable than backing the team outright or covering the draw. Hence, there are lots of different ways of following these bets and hopefully we'll make a profit no matter what we choose to follow but all my analysis points to backing the teams to win the game is the most profitable method.
Anyway, back to the comment from Chris.
I've given a bit of thought to how my bets will look next season. It's important that I show the odds on the team to win, the odds if you were covering the draw also and of course, the bookmaker that provided these odds. The below is a template of what I will use next season. (ignore the fact I've only used one game!)
It's all fairly self explanatory I think but I'm sure it won't please everyone. As I've stated before on the blog, all of my analysis on DNB is done using the draw odds and calculating the stakes you'd place on the team to win and draw to calculate the DNB odds. I'm aware that this is likely to be lower than the DNB odds offered by bookmakers but to maximise profits, I don't see an issue with placing 2 bets instead of 1.
Of course, odds move at times and this could cause you to not be perfect and I know people don't like betting £32.38 on games with bookies as it looks like you're trying to do some sort of arbing but if you rounded every bet to the nearest £ and even ignored the odds movements, I doubt it would affect your P&L too much.
Anyway, I think this answers your question Chris. If not, let me know exactly what you were after mate.
My main issue now is simply the size of the sheets as doing analysis involving pivot tables (which is how all my analysis is done) causes the spreadsheets to be quite large. It's so bad now, when I'm transferring spreadsheets between work and home, I have to send a spreadsheet at a time (even though it is zipped!) which is a farce really.
There was a comment on the blog last night from Chris asking how the bets would look like for next season. Reading between the lines, I'm assuming he means what the betting sheet will look like I post up here with the bets. He mentions if I'll be quoting AH odds along with the odds for the team to win also.
I've discussed this before but I am sticking with two ways to follow these system bets next season. I will monitor the profitability of the bets to backing the selections outright and also backing the selections DNB (covering the draw). I won't look at any other Asian handicaps although I appreciate these type of bets appeal to some people more than others.
Just on the Asian Handicap bets point, one of my other readers Mark sent me some data for AH betting for the leagues where I'm playing in. It was the AH odds for all the games for the last 3 seasons. He asked me to do a little bit of work to look at this and I spent over 4 hours on Saturday night coming up with an analysis of the performance of all my systems using these odds. I've already said to Mark I wouldn't post it on the blog but I did send Mark the analysis. I feel like this blog has enough going on without me introducing AH analysis as trying to intrepret the results is difficult.
A high level summary would be that AH betting is still very profitable with these systems but OVERALL, the returns are lower than just backing the selections to win. There were a lot of different types of AH bets and clearly, some were more profitable than outright betting for some systems but that's more to do with small samples of bets, more than anything else.
I've already looked at double chance betting and shown that it is less profitable than backing the team outright or covering the draw. Hence, there are lots of different ways of following these bets and hopefully we'll make a profit no matter what we choose to follow but all my analysis points to backing the teams to win the game is the most profitable method.
Anyway, back to the comment from Chris.
I've given a bit of thought to how my bets will look next season. It's important that I show the odds on the team to win, the odds if you were covering the draw also and of course, the bookmaker that provided these odds. The below is a template of what I will use next season. (ignore the fact I've only used one game!)
It's all fairly self explanatory I think but I'm sure it won't please everyone. As I've stated before on the blog, all of my analysis on DNB is done using the draw odds and calculating the stakes you'd place on the team to win and draw to calculate the DNB odds. I'm aware that this is likely to be lower than the DNB odds offered by bookmakers but to maximise profits, I don't see an issue with placing 2 bets instead of 1.
Of course, odds move at times and this could cause you to not be perfect and I know people don't like betting £32.38 on games with bookies as it looks like you're trying to do some sort of arbing but if you rounded every bet to the nearest £ and even ignored the odds movements, I doubt it would affect your P&L too much.
Anyway, I think this answers your question Chris. If not, let me know exactly what you were after mate.
Saturday, 16 July 2011
Misconception!
The reason for the lack of updates on here is simply because I’ve been busy working away on the boring stuff in the background such as getting my spreadsheets all set up for the start of next season. It’s amazing to think the SPL starts next Saturday as the Summer appears to have flown by.
I only really looked at the fixture list tonight as someone mentioned in the SBC forum last night that I hadn’t said when the first bets were going to be. I’ve had a look at the fixture lists tonight and as usual, it will vary by league.
Premiership – 1st October
Championship – 17th September
League One – 10th September
League Two – 10th September
Bsq Prem – 3rd September
SPL – 10th September
I’m sure the above is subject to change and if they move any games around, it might have a small impact I guess.
It has been an interesting week or so in the SBC forum with a lot of discussion around these systems. Rowan touched on this on his blog last night but I think there is a misconception that you need to follow hundreds, if not thousands of bets to be successful following these system bets next season. I personally don’t believe this is the case.
I only really have one season’s experience of following the bets myself, so I’m not in much better shape than anyone else who decides to follow the bets next season. However, one thing I learnt from last season is that even by having a lot more bets, it doesn’t necessarily guarantee you a smoother journey throughout the season.
Below shows the graph of just following system 8-22 compared to following all of the systems from last season. As you can see, did following 2,600 more bets actually give any smoother return? I’d say no, probably not!
Obviously, the overall graph is just made up of lots of little graphs for each system and it isn’t too surprising that system 8-22 looks a lot like the overall portfolio. 8-22 includes all bets that must appear on every system, so it is going to be a good proxy for the performance of the portfolio of systems as a whole.
I followed a lot of bets last season and considering the additional workload it brought me, I don’t think it was really worth it. I should have followed a small sample of systems that represent the portfolio as a whole.
Next season, I also have all my other new systems to monitor and I’ll need to play these with smaller stakes myself. I’m not very good when it comes to paper trading and I really need to be playing small stakes to keep me interested in updating all my results and doing all the analysis as well as all the work to find the bets each week.
One thing that has struck me this week is the misconception that what I’ve built here is quite difficult to follow. I honestly believe that if people just ignore the rest of the systems and concentrate on a system/group of systems that they want to follow, then these systems are the easiest thing in the world to follow. All I’m showing is 20 different systems really but I’m not saying anyone should follow all 20 systems!
Anyway, that’s a quick update on things. We’re still 6 weeks away from the first bets, so the blog will get much more interesting from then I think!
Wednesday, 6 July 2011
TOX, STOY and STOZ
I’ve apologised before for the names of my football systems in the sense they aren’t exactly the most original. Systems 6,7,8,21,22 didn’t really get the juices flowing and being honest, I’m not sure systems 6-21, 6-22 and so on is any better.
Well, the 3 systems from the new Similar Games model are going to be called TOX, STOY and STOZ! I’m not sure this is really an improvement on the numbers theme but no one can accuse me of trying to be melodramatic when it comes to naming these systems! They are pretty boring names at the end of the day.
Putting the names of these new systems to one side, how do they look from a performance point of view?
Firstly, a common theme from these 3 new systems is the fact there don’t appear to be many bets each season. TOX only had 34 bets last season, STOY had 27 bets last season and STOZ had 49 bets last season. Clearly, this raises an issue in the sense we don’t really know if these systems have an edge or not as we don’t have a large enough sample of bets.
Looking in slightly more depth, the trend on all the systems is for less bets each season. Therefore, we’ll just need to see how many bets are thrown up this season. If it is any lower than last season, then these systems become meaningless I suspect.
Putting the number of bets issue to one side, the ROI on each system is higher than any other system I’ve managed to create! TOX has an overall ROI of 34%, STOY has an overall ROI of 41% and STOZ has an overall ROI of 28% (but nearly 50% for the last two seasons!)
Similar to what I’ve done with the other new systems for next season, I’ve looked at all the historical results in depth and looked at the P&L, drawdown, performance by league, season, odds and so on. All the results are available for everyone to look at here. Scroll down to the bottom for these newest systems.
http://the-football-analyst.blogspot.com/p/new-systems-for-201112-season.html
Similar to the other new systems next season, I suggest that we all paper trade these systems and see how do in the live environment. If they achieve returns within 50% of their backtested results, then we can possibly see a return of 20%-25% next season which would be an amazing return, even allowing for the fact we aren’t going to get a lot of bets each season. Of course, the Similar Games Model might be a crazy idea and we’ll see losses next season on these new systems. We’ll see what happens.
If anyone has any questions, please leave a comment.
Subscribe to:
Posts (Atom)