mathare's gamblog

[mathare]

Interesting stuff Mat. Are those systems using both unders and over bets? Or just one or the other? If both does one perform better than the other?

All three of the systems I have come up with to date are actually just unders systems as I haven't been able to find anything profitable on the overs. I might have a sniff of one for the Championship though if my early analysis is anything like.

I would think part of the problem finding a consistently performing system for these leagues is the fact a quarter of the teams change year on year, but like you mentioned in an earlier post that can also play into our hands because it also makes it much harder for the bookies to be accurate too.

While it's true that the line-up in these divisions can change significantly each season most of the time the teams being relegated/promoted are going into a division they have some recent(ish) history in. So a team from League One may make a brief appearance in the Championship but will probably slip back to League One in time. There are exceptions of course, but unless a team has sufficient history in the division in question they aren't selected by the system. And I'm not going to quantify what sufficient history is as it varies by system.

Out of interest I have just run the League One systems over the results for last weekend. 1a (30) would have picked three matches to back under 2.5 goals in (Southend v Southampton, Carlisle v Norwich and Yeovil v Brighton) which would have given us just the one winner, Carlisle v Norwich going under at 19/20 so a loss of 1.05pts. System 3a would have picked the Carlisle and Yeovil matches so would have lost 0.05pts. I won't be including these in the official figures when I start posting qualifiers on the forum though.

[mathare]

12th October 2009 (part two)
I have just been revisiting some of the conclusions I drew about the League Two system and am now considering running an unfiltered version of it alongside the filtered version I mentioned yesterday. Why? It's simpler for one but also the filters affect the max odds bets reducing the profits slightly so it's not clear what they would do to the middle-of-the-road odds I expect to get. The worst case scenario is better for the filtered version but the best case scenario is better for the unfiltered version so it might be worth running the two side by side in a live test when I am not sure where on the scale my odds will lie.

I also did a season-by-season breakdown of the unfiltered system to average odds and both the filtered and unfiltered version to maximum odds. When looking at the unfiltered system to average odds we see 06-07 is still a freak year providing all the profits with two years showing slight losses (05-06 and 07-08). But switch to maximum odds and all years show a profit. Not only that but while 06-07 still provides a lot of the profits, 08-09 weighs in with a good chunk too giving us two good seasons and two not so good ones. The same is true for the filtered system viewed in terms of maximum odds, although now 07-08 is actually a slight loser. So if we're getting closer to the maximum odds than the average odds we are probably better off using the unfiltered version and for that reason I will run both systems for League Two in the live test so we can see if there is any difference over the coming months.

[MattR]

Will be interesting to see how these pan out 'live' Mat, look forward to seeing it progress.


A thought that just crossed my mind concerning under overs is that the Premier with a current average of 3 goals per game will probably with pretty reasonable 'certainty' (if that isn't a contradiction :D ) come back down by the end of the season to around 2.5/2.6. With that in mind you would think that unders would have to start coming in with much greater regularity than they have been. Much in the same way that we had 5 draws from 10 games on the last set of fixtures after just 4/66 previously. In other words can we take advantage of statistical correcting?

[mathare]

In other words can we take advantage of statistical correcting?

I'll be honest and say I'm not sure. I ought to look at seasonal goal averages in the Premiership and how they have changed throughout the season to get a better idea of this but you can be sure the bookies will be doing the same. But this could go one of two ways - the goals could keep coming and we could have a record season for overs or there will be a correction and things will revert back to normal with more unders than we have had so far to bring that average back down. The bookies can't price up both eventualities on a weekly basis so it will be interesting to see whether they start to lengthen the unders odds expecting the goals to keep flowing or trim them expecting a correction. What we have seen so far this season amounts to little more than a statistical blip though as not many games have been played. It would be interesting to see how other seasons looking in terms of goals per game at this stage and what they settled down to eventually. A job for tomorrow that one I think

[mathare]

A job for tomorrow that one I think

I've cracked and done it now!

I've attached the stats as images to keep the tables looking nice. I have calculated end of season averages and also the averages after 76 games, which is the number played in the Premiership so far this season.

This season's goals per game ratio is high at 3.013, especially in the light of last season's 2.479 over the whole season. That is the second lowest average number of goals the Premiership has seen with 06-07 being the lowest. After 76 games this season is still by far the highest average goals per game figure but if you look at those seasons were the average number of goals after 76 games is greater than the overall average since the start of the Premiership (94-95, 95-96, 97-98, 99-00, 00-01, 03-04 and 08-09) all of these bar one (99-00) ended up with a lower average at the end of the season than they had after 76 games. Therefore I'd say the average number of goals is going to go down between now and the end of the Premiership season but I don't know how much. It was 2.592 before the start of the season so let's assume it drops back to that long-term average. That means we'd be looking at another 756 goals this season (or thereabouts) giving an average of 2.487 goals per game, not a lot below the 2.5 line offered by the bookies so how much value can we expect there? Not a lot I'd say.

So really I think all we can say is that the average will come down but how far down no-one knows and I'm not sure I'd be looking to back a glut of unders as that average could drop slowly over the course of the rest of the season without seeing huge clumps of unders.

[MattR]

Interesting Mat. Like you say the probable reduction could be gradual.


Just checked the 4 seasons I have saved on my hard disk and interestingly both home and away teams are averaging more to date this season. Not sure what betting use that would have anyway, asian handicaps perhaps, but I guess it's kind of what you'd expect to see given the higher % of home wins. Would imagine that would translate across most leagues.

2009/10 1.75-1.25
2008/09 1.40-1.08
2007/08 1.53-1.10
2006/07 1.45-1.00

[mathare]

15th October 2009
I've been feeling a bit rough the past few days so haven't had time to analyse the Championship under/over data till now. But here are the findings anyway, and it's nowhere near as promising as for the lower divisions.

There are two systems of interest, one of which I have selected despite the fact that in it's basic form it doesn't produce a profit when settled to average odds. So why have I picked out a system that makes a loss? Because when you settle the bets to maximum odds it produces over 60pts of profit so there must be something in it. Bearing in mind that the odds I expect to get will lie somewhere between the average odds and the maximum odds and there ought to be a profit in the system but how much remains to be seen. So let's take a look at this system in a little more detail.

System 2a (10) is a system for backing over 2.5 goals, unlike the lower league systems which were all unders. As I said it makes a loss (-13.75pts) to average odds in it's raw form with a ROI of -2.00% but to maximum odds we see a profit of 62.38pts with a ROI of 6.18% - much better. Average that out and we're looking at something like 25pts over four and a bit seasons which is only something like 6pts a season, less than I was after from a system but still not too bad I suppose. Anyway, the raw system isn't the best form for this system - filters can be added to cut out some of the losing bets and improve the profit levels to -11.75 (-1.87% ROI) for average odds and 69.95 (8.72%) for maximum odds. That increases the midway profit to something like 30pts with an approximate ROI of around 3.5%. That's maybe 7pts a season which is getting a bit better. The system will produce around 150 bets a season too I reckon, on average. Unfortunately though the seasonal breakdown doesn't look too good. The graphs don't show this very clearly at all but the best years for this system seem to have passed with this season (09-10) and last (08-09) being pretty much breakeven and all the profits coming in the first three years. Because of that I want to monitor this one to see how it goes and as such is will be paper traded live on the forum until the end of the season as I simply don't trust the figures to be reliable enough to add it to my portfolio before that point.

The other Championship system, 4a (30), is again an overs system. In terms of profit there is little to choose between the system when settled to average odds and the maximum odds version with both producing 15-20pts of profit but when we use maximum odds we nearly half the ROI indicating we're picking a lot more losing bets. Indeed the SR for the max odds system is 4% lower than that of the average odds version. But it doesn't make sense not to take the best odds on offer so clearly we need to add some filters here to improve the basic system. When we do that things look better, increasing the profit of the max odds to 50.79pts (8.56% ROI) while to average odds we have a profit of 19.49pts (9.28% ROI). Fewer bets with this one but we might still squeak somewhere around 80 or so each season. Again, not really up to the standards I was looking for from a football system but the profits are there or thereabouts again averaging around 8pts a season if we assume the odds will be midway between the average and maximum odds I have recorded. As with the previous system the seasonal breakdown indicates the early years (05-06, 06-07 and 07-08) were best but the downward trend observed earlier isn't quite as pronounced here. Even so, this is another one to paper trade on the forum till the end of the season where the trends will hopefully start to shine through a bit better and I can decide how to proceed with this system.

I have attached the graphs again, blue is the average odds line and red the max odds line as always.

Things are certainly progressing with these systems and I will be starting a new thread on the forum with any qualifiers for this weekend as and when I pluck them out, which will probably be tomorrow when the odds are out.

[mathare]

15th October 2009 (part two)
As I thought a few days back, there is nothing doing with the Premiership for under/over systems. I have just looked at the data again, in more depth than I did a week or so back and in a more systematic manner and the systems just aren't profitable to average odds and only produce a few points profit each season to maximum odds. I reckon the odds compilers have got these markets sewn up pretty tightly and there isn't much value to be had most of the time. I think the lower leagues are where it's at for profitable football systems, at least for under/over 2.5 goals anyway

[mathare]

16th October 2009
I have now set up a thread in the English football betting area of the forum to hold the system qualifiers for the various under/over 2.5 goals systems I have found over the past week or so. You'll be able to follow the progress of these systems in that thread.

I have considered applying the same lines of thought as I used on the English leagues to the Conference and the Scottish leagues as at least a few of the big bookies have priced up these markets it seems but I'm not sure I am up to any of that today. It can wait till the weekend or early next week.

[John]

Have been following this thread with a lot of interest Mat, as I really the idea of football systems, more-so at the minute than horse racing systems due to my greater interest in football. Plus I think there are less parameters in football to fiddle with when creating systems. You don't have as much stuff to tinker with.*

I'll be watching your footy systems to see how you're doing.

*Note the very technical language of this post.

[mathare]

Thanks for the interest John. There are fewer parameters in most of the football markets I'd say, at least fewer important parameters. With a horse it's not clear (to me at least) how much weighting one should give the weight, jockey, trainer, ground etc when trying to assess how a horse will perform in a given race. Plus a football team is playing one other team so the possible results are fewer compared to a race that will have a field of around 10 runners or so. Football teams also have a lot more 'events' than a racehorse. A horse generally runs only a few times a season whereas a football team will play around 40-50 games giving you much more data to work with. The one thing football lacks compared to horse racing though is an SP, I really miss that. But there are mroe markets for football so more opportunities to look for value while still keeping things pretty simple.

[John]

I've had a bit of a thought with regard to your under/over system ideas as well. As it's a very simple idea of mine, chances are you will have probably tried it out already. But how does backing under 2.5 goals fair with teams playing each other in the bottom half of a league? Or even reducing it further and say the bottom quarter of a league? Teams in the bottom half who play each other doesn't always produce a very exciting game, and by the same token they sometimes/frequently (trying to choose my words carefully here...) have fewer goals in them.

First thing's first though are the exceptions to the rules you have to deal with. Let's suppose Pos. 14 (H) plays Pos. 22 (A) one could argue that Pos. 14 would win the match. This could still be a dodgy qualifying bet though, since Pos. 14 could be a much stronger team/in a good run of form, compared to the team sat in Pos. 22. They could win 3 or 4-0. So you could rule this out of your criteria. However, if Pos. 22 are the home team (against Pos. 14 (A)) one could again argue that the teams are in closer proximity ability-wise to each other so there's more chance of a deadlock. So then this could be a more justifiable betting proposition.

But just going back to my initial point, let's say Pos. 17 play Pos. 18, one could argue there'd be very few goals. Assuming that both teams are in poor form, morale is low, neither team are really setting out to win the game. 0-0, 1-0, 1-1 could easily be a very likely scoreline.

A (possible) drawback for this is that there would be less matches to bet on. Not really a drawback, depends how you look at it I guess... plus you would also have to investigate previous meetings between clubs A and B and how they've performed against each other in the past. You could use a percentage rule here whereby supposing they've both met 10 times previously: if 5 or more of those matches resulted in the scoreline total being <2.5 goals then it could qualify as a bet.

My English is rubbish today but I hope I've made clear what I mean.

[mathare]

I've had a bit of a thought with regard to your under/over system ideas as well. As it's a very simple idea of mine, chances are you will have probably tried it out already. But how does backing under 2.5 goals fair with teams playing each other in the bottom half of a league? Or even reducing it further and say the bottom quarter of a league? Teams in the bottom half who play each other doesn't always produce a very exciting game, and by the same token they sometimes/frequently (trying to choose my words carefully here...) have fewer goals in them.

First off I'm not sure I agree with your hypothesis here (are the teams down in the depths of the league because of poor attack or poor defence? If the latter there could be a lot of goals if both sides have a tendency to concede too easily) but then it's an idea that could be tested, but for the fact it involves a lot of work.

How do you know a team is in the bottom half/quarter of the league? To automate the back-testing of this you need to effectively calculate a league table after each set of matches, compute a position for each team and then match them up according to your criteria. Way too complicated for the systems I am looking at right now.

First thing's first though are the exceptions to the rules you have to deal with.

The best thing to do is start simple and look at filters (exceptions) after that. But the basic system itself ought to look reasonable before considering tweaks to it.

[John]

If both teams' defences are poor though, it doesn't automatically mean that their attack is going to be good. It all depends if the poor defence is a comparison to the attack or not, or whether the forwards are just as bad as the defenders.

My point about back-testing the scorelines was less geared to their league positions at the time, and more geared towards the sum of the scorelines. If 7 from 10 matches between Teams A and B resulted in less than 2.5 goals being scored, then that alone could be a good enough qualifier for a bet without even taking their respective league positions at the time into consideration.

But then as you say, that'd be more like you using your opinion and not adhering to a set of rules which takes it away from a system and starts to turn it into a method.

[MattR]

Don't want to hijack your thread Mat, so just a quick one here

John, why don't you test that on the forum, the basic idea of under 2.5 for all games between the bottom (half or quarter you decide) of the league. Take it from the current league position. I would suggest waiting at least 6-8 games into the season, so now's a good time to start. On the face of it there's some merit to it as a weak defence could be the fact of playing the better teams although a weak attack could be down to the same reason

[John]

Certainly an idea that I'd like to test at some point Matt, but I don't know if I have time at the moment to do it. I'll see...

I'm not sure which league(s) to go for either. On assumption, the lower league teams would be more profitable because the teams are a lot poorer. But then it's proportional/relative to two bottom Premiership/Championship teams playing against each other.

[mathare]

21st October 2009
It's been a while since I had anything to say in here. I haven't got that much to say today but I thought I ought to make another entry anyway.

The under/over systems have started up now and had their first few games. It was a fairly inauspicious start to the live paper trade but there is plenty of time to turn it round. And of course it's not real money that is being lost - that's the whole point of a paper trade. I'm confident that things will pick up in time as the work that has gone into these systems stands up to scrutiny. I am confident that they will come good despite a losing opener. There will be no big winners to boost the numbers though as under/over bets are for prices at or around evens so it will be a steady climb back to profitability I should imagine.

I have taken a few days off from the football systems to let my brain (and body) recover somewhat. In fact I have dug my PS2 out the loft, bought Gran Turismo 4 second-hand and have been playing that. Awesome game. I enjoyed GT3 but didn't realise how much better GT4 was!

Today I am back to gambling related activities though. I have started to realise recently that the spreadsheet that I use to record all my bets in could do so much more for me than it does already so I have set about updating it to get this information out of it. When discussing results on the forum it is much easier to talk about things in terms of points won or lost rather than actual cash amounts as everyone stakes differently so one of the things I have set about doing is working on getting the spreadsheet to output that kind of data in a nice format to make posting my results easier. It's complicated slightly by the fact that sometimes I don't get the full stake matched when using the exchanges. If my system stake is £10 and I only get £8 matched on a 2/1 winner then I have only won 1.6pts not 2pts and I need to have that reflected in my data. I'm getting there but I need to work out how best to display this data in the current structure of my workbook. I am also looking at simplifying some of the sheets I have in there and combining some of them to reduce the overall number as a few of them are so similar in structure that they could easily be combined. None of this really matters to anyone other than me of course but that's what I am up to. Plus it's keep my brain active and concentrating on gambling matters. I hope that once I am armed with the information I desire I can better decide whether anything should be dropped from my portfolio.

[John]

I haven't got that much to say today but I thought I ought to make another entry anyway.

And you still manage to write 500 words... :D

I have dug my PS2 out the loft, bought Gran Turismo 4 second-hand and have been playing that. Awesome game. I enjoyed GT3 but didn't realise how much better GT4 was!

I love car racing games; GT4 is awesome, I'll agree with you there. Have you played 5? I don't have a console, in fact I've never owned one and the pleasure I get out of playing on other people's (consoles, that is) makes me think I should perhaps buy my own.

[mathare]

I love car racing games; GT4 is awesome, I'll agree with you there. Have you played 5? I don't have a console, in fact I've never owned one and the pleasure I get out of playing on other people's (consoles, that is) makes me think I should perhaps buy my own.

It was only the adverts on TV for GT on the PSP that made me think about getting my PS2 out. I bought the GT3 edition so the console came bundled with GT3. I had planned on just playing that (GT3) but I can't find it - I may have traded it in years back when I used to play my PS2. So I ordered GT4 from Amazon marketplace instead. It was less than a fiver with postage so I was happy to pay that. And it's amazing. The B-Spec mode is an ace feature; it's like having the Stig driving for you.

Never played GT5, nope. Never even played on a PS3, or an XBox. And I've only played a Wii once or twice. I'm not a big gamer as I don't have the co-ordination to use the controllers that well. I have to use digital acceleration/brake on GT4 as I can't get that right in analog mode although I use analog steering. But things like Grand Theft Auto etc are beyond me as you need to use most of the buttons at once for some of it and I can't do that. Similarly I can't use keyboard and mouse together for PC games

[John]

I must start using Amazon Marketplace more, you can pick up some great bargains as you rightly have done. I'm in a similar situation to you as far as modern consoles go. I've only played on a PS3 a handful of times round a friend's house. Mainly FIFA 2009 [which is so realistic it's unreal] and some sort of boxing game which I kicked his ass at... you're repeatedly pressing buttons and trying combinations in boxing games and it's quite exhausting to play... achy hands!

I can't say I suffer from the mouse/keyboard combinations you mentioned. An example of this being playing Battlefield 1942 and Battlefield 2, both great war games where you use the keyboard to run/strafe and the mouse to pan around. It does take some getting used to though.

[mathare]

Y'see I can't play those footy games as I can't master any of the moves. I can't switch players, tackle, get the ball and then drop my shoulder to beat a man before putting a cross in and getting the striker's head on it. It takes way more coordination than I have so I end up fouling the opposition and when I do get the ball I kind of run at the other side and get tackled really easily so resort to longshots that never go anywhere near the goal. I don't have the deft touch required. I'm better with simple controls like point and click games such as Championship Manager, although I haven't played that for years now.

[scoobydoo]

Y'see I can't play those footy games as I can't master any of the moves. I can't switch players, tackle, get the ball and then drop my shoulder to beat a man before putting a cross in and getting the striker's head on it.

It sounds like I need to play you instead of the computer then Mat! :D

[John]

Y'see I can't play those footy games as I can't master any of the moves. I can't switch players, tackle, get the ball and then drop my shoulder to beat a man before putting a cross in and getting the striker's head on it. It takes way more coordination than I have so I end up fouling the opposition and when I do get the ball I kind of run at the other side and get tackled really easily so resort to longshots that never go anywhere near the goal. I don't have the deft touch required.

Does that remind you of any current managerless football teams?

[mathare]

Saturday 24th October 2009
Since I last made a proper entry in this gamblog I have spent some time tidying up and reformatting my betting records. I now have all profits calculated in both cash terms and a number of points. What's more my stats macros now pull out the profits in points too so I know how many points profit I made on any system, in any given month, in a particular year etc. I have also cross-checked all my stats figures to make sure they add up properly and I am pleased to say they do. This has put me in a very positive mindset. I've not finished the work I wanted to do on this spreadsheet but the work I have done so far has put me in a good position to add in the other features I had planned, which is nice. I think the new structure is much better too and am feeling very positive about many things now, despite recent downturns in my general wellbeing (but I don't want to dwell on that here).

Depending on how I feel over the next couple of days I should be able to finish off the improvements I had planned and will then be able to get much more meaningful information out of my spreadsheet, and get it easily too. Maybe some of that will find it's way in here as I will be able to analyse my portfolio much more readily.

Hopefully, once all that is done, I will find the self control to continue with some other projects I have been working on from time to time, one of which is very exciting but I don't want to reveal too much at this stage as it's still embryonic and will take quite a while to reach the stage where I am happy to announce any details.

[mathare]

Saturday 24th October 2009 (part two)
Nothing definite here but after an initial glance I'd say W2W lays and the Nine Or Less system will soon be dropped from my portfolio. The lays have been terrible recently and I have never really got the bank off the ground so it's a prime candidate to be dropped. Nothing official yet though but watch this space