General Betting
Welcome to Live View – Take the tour to learn more
Start Tour
There is currently 1 person viewing this thread.
10 Feb 11
12:15
Joined:
Date Joined:
20 Jan 04
Hello, I haven't posted on here for a long time. I used to play the horses but now I have turned my attention to football.
I have been working on a system for 3 months now involving the scores at half-time. I based my original ideas on the percentages of likely outcomes obtained from over 20,000 historical games. The only thing I didn't have, of course , were the prices available on Betfair at the time, so I have been collecting this data for the last 3 months. So far I have collected odds at half time for 1800 matches and my system would have made 1477 bets ( my philosophy is if you cast a big net you will catch the big fish when they come along). I have been tweaking the parameters as I go along to keep the profit at it's maximum, and at the moment it stands at 1477 selections, 423 winners, 28.6% SR, 505 pts profit (after deducting commission).
My questions are, how long should I continue to trial before I go live?, and when I do go live, how much of a dip should I expect in my results based on the fact that my system has been taylored to fit historical events? Is it possible that the next 1477 selections could produce a loss? Also, would you continue to tweak the parameters after going live in order to maximise the system's profit, even though your actual profit will be less, as you are tweaking after the event.
I hope I haven't bombarded you with too many questions, but I would like some reassurance that I am doing the right things in order to make a long term profit on here. I would really appreciate any advice.
I have been working on a system for 3 months now involving the scores at half-time. I based my original ideas on the percentages of likely outcomes obtained from over 20,000 historical games. The only thing I didn't have, of course , were the prices available on Betfair at the time, so I have been collecting this data for the last 3 months. So far I have collected odds at half time for 1800 matches and my system would have made 1477 bets ( my philosophy is if you cast a big net you will catch the big fish when they come along). I have been tweaking the parameters as I go along to keep the profit at it's maximum, and at the moment it stands at 1477 selections, 423 winners, 28.6% SR, 505 pts profit (after deducting commission).
My questions are, how long should I continue to trial before I go live?, and when I do go live, how much of a dip should I expect in my results based on the fact that my system has been taylored to fit historical events? Is it possible that the next 1477 selections could produce a loss? Also, would you continue to tweak the parameters after going live in order to maximise the system's profit, even though your actual profit will be less, as you are tweaking after the event.
I hope I haven't bombarded you with too many questions, but I would like some reassurance that I am doing the right things in order to make a long term profit on here. I would really appreciate any advice.
Show More
28% winners 505 points profit go for it now , re dips dont go live keep it too yourself that way only you know how selections were chosen .. good luck you wont need it
Re: average odds, the way my spreadsheet is laid out makes it awkward for me to calculate an overall average, but I should point out that my overall system is made up of 13 smaller systems. The one with the most selections (437) has an average of 2.8. The one with the least (19) has an average of 400.
Zipper, when I said 'go live' I meant 'start putting money on'. There's no way I could publish my selections on here anyway as there are far too many and not enough time I'm afraid. Thanks for your coments though.
Zipper, when I said 'go live' I meant 'start putting money on'. There's no way I could publish my selections on here anyway as there are far too many and not enough time I'm afraid. Thanks for your coments though.
Gotta go live because it's so hard to know which bets would actually get matched by yourself and which would be matched by other people who may or may not have a similar method for deriving the prices.
Continue tweaking until you're happy with it.
Then trial for another 300 games.
Then go live in smallish.
Once live, the dip may not happen straight away. I have been running a system for the last 600 games, after testing on 1200 games beforehand. Since going live the 1st 300 went perfectly to plan. The last 300 have seen an unprecented drawdown ~ less than it made in the 1st 300 live games, but still very concerning. I am currently asking a statistical analyst whether this drawdown is terminal, or within the bounds of reasonable expectancy. It's a long long slog.
Then trial for another 300 games.
Then go live in smallish.
Once live, the dip may not happen straight away. I have been running a system for the last 600 games, after testing on 1200 games beforehand. Since going live the 1st 300 went perfectly to plan. The last 300 have seen an unprecented drawdown ~ less than it made in the 1st 300 live games, but still very concerning. I am currently asking a statistical analyst whether this drawdown is terminal, or within the bounds of reasonable expectancy. It's a long long slog.
^good advice, imo. it needs to demonstrate itself over a significant number of games after you stop tweaking it.
alternatively, go live now for notional stakes.
alternatively, go live now for notional stakes.
terminal means when building the system you probably (over)fitted the curve ie. you have deluded yourself and the system is no good.
It's akin to deferred failure.
To the OP, it's definitely time to go live.
As Lori says, you have at least to start finding out exactly at what prices you can actually get set. Theoretical prices are misleading on this and establish a false profit expectation.
Start with min stakes for a month if you have capital constraints.
To the OP, it's definitely time to go live.
As Lori says, you have at least to start finding out exactly at what prices you can actually get set. Theoretical prices are misleading on this and establish a false profit expectation.
Start with min stakes for a month if you have capital constraints.
if he's been retrospectively tweaking his system to optimise it in the light of past results, it's definitely not time to go live unless for notional stakes as a way of keeping score; it's time to test a stable version.
imho
imho
Absolutely spot on Viva.
Stop tweaking, settle on what you have developed, and start to use use real money for real results.
Stop tweaking, settle on what you have developed, and start to use use real money for real results.
Nah you just don't like what I'm saying.
I make real money, not pretend money.
Btw I went through exactly the same stage)s) as the OP has done.
Farted around for ages on things.
It got to the point where my wife for God's sake said either go for it or stopping wasting time on it.
Good advice in the end.
I make real money, not pretend money.
Btw I went through exactly the same stage)s) as the OP has done.
Farted around for ages on things.
It got to the point where my wife for God's sake said either go for it or stopping wasting time on it.
Good advice in the end.
I very much doubt it. imho you spout more rubbish than anybody. difficult to imagine that translates into successful betting. i could be wrong, but then i don't have your confidence ...............
WSO, have you tried chi test? I use a similar test for identifying statistical strength of a system.
T-test of a mean yield
1. Calculate the yield for all bets. (wins, losses)
2. Calculate the mean yield and std. Error of the mean. =AVERAGE(D1:Dn) (that is just an average of your wins or loses)
=(STDEV(D1:Dn))/(SQRT(COUNT(D1:Dn))) (std. Error of the mean)
3. Calculate the prob. Of achieving negative yield.
=NORMDIST(0;E1;E2;1) (where E1 is mean yield and E2 is std. Error)
The result of that 3. should be less than 0,05. Less is better.
T-test of a mean yield
1. Calculate the yield for all bets. (wins, losses)
2. Calculate the mean yield and std. Error of the mean. =AVERAGE(D1:Dn) (that is just an average of your wins or loses)
=(STDEV(D1:Dn))/(SQRT(COUNT(D1:Dn))) (std. Error of the mean)
3. Calculate the prob. Of achieving negative yield.
=NORMDIST(0;E1;E2;1) (where E1 is mean yield and E2 is std. Error)
The result of that 3. should be less than 0,05. Less is better.
I was once stood in a queue behind a retrospective tweaker. Made my eyes water [:p]
losersarechasers
In your opinion I spout rubbish.
But you're never really specific as to what that rubbish is.
I am perfectly willing to be corrected on all I say, if the corrections make sense.
Why not give it a try.?
Could be fun ( for both of us ).
In your opinion I spout rubbish.
But you're never really specific as to what that rubbish is.
I am perfectly willing to be corrected on all I say, if the corrections make sense.
Why not give it a try.?
Could be fun ( for both of us ).
Btw I am extremely confidemt that I know more than most, but certainly not all, on here.
Maybe you're one of the few whom I can tip my hat to.
Let's hope so.
Maybe you're one of the few whom I can tip my hat to.
Let's hope so.
Thanks everyone for taking your time to post your advice, I really appreciate it.
I think first of all it is time to stop tweaking. You can tweak all you like after you have made your bet and lost, you can turn a losing selection into a winning one or a 'no bet', but it doesn't get you your money back. I need to see how the system performs when I just leave it alone, so I won't be touching it again.
As for getting matched, to start with my stakes will be quite low as my wife will only let me have a starting bank of £400 and from what I've seen so far prices available seem to remain pretty static during half-time, which is when I will be placing my bets. Besides that, my system has been based on prices available to bet on, not me assuming I will get matched if I request specific odds.
Sun Wukong, I am very interested in the test you refer to but I am no mathematician and can't really understand it. I'll have a go though and see what I come up with, thanks for that.
I am thinking at the moment that I should leave it alone until the end of Feb and if it continues to look promising then go live with minimum stakes at the start of March.
I think first of all it is time to stop tweaking. You can tweak all you like after you have made your bet and lost, you can turn a losing selection into a winning one or a 'no bet', but it doesn't get you your money back. I need to see how the system performs when I just leave it alone, so I won't be touching it again.
As for getting matched, to start with my stakes will be quite low as my wife will only let me have a starting bank of £400 and from what I've seen so far prices available seem to remain pretty static during half-time, which is when I will be placing my bets. Besides that, my system has been based on prices available to bet on, not me assuming I will get matched if I request specific odds.
Sun Wukong, I am very interested in the test you refer to but I am no mathematician and can't really understand it. I'll have a go though and see what I come up with, thanks for that.
I am thinking at the moment that I should leave it alone until the end of Feb and if it continues to look promising then go live with minimum stakes at the start of March.
Do you have "back-testing" results covering MARCH last year?
With CHAMPIONS LEAGUE/ EUROPA LEAGUE games and other end of season matches where CL / relegation positions in leagues are being fought , together with cup competitions etc....it is a very busy time for top clubs and players.
European Championship fixtures played in MARCH etc too.......therefore players may not perform to high level in club fixtures.....
No idea what leagues you are betting on .......just indicating that be wary on selecting matches during busy time for clubs and players etc etc
GL
With CHAMPIONS LEAGUE/ EUROPA LEAGUE games and other end of season matches where CL / relegation positions in leagues are being fought , together with cup competitions etc....it is a very busy time for top clubs and players.
European Championship fixtures played in MARCH etc too.......therefore players may not perform to high level in club fixtures.....
No idea what leagues you are betting on .......just indicating that be wary on selecting matches during busy time for clubs and players etc etc
GL
If he has data for over 20,000 historical games, I would expect he's well past last March.
But good sage advice nevertheless.
Have you personally noticed that Mch results generally show large variances from the norm.
But good sage advice nevertheless.
Have you personally noticed that Mch results generally show large variances from the norm.
from what I've seen so far prices available seem to remain pretty static during half-time
---------
not sure about that, actually. I've seen quite substantial moves in the course of HT, and the price you manage to get on at may well be key to your system's success or failure.
---------
not sure about that, actually. I've seen quite substantial moves in the course of HT, and the price you manage to get on at may well be key to your system's success or failure.
Of course his system may be robust enough to work at any " reasonable " prices.
Some systems can and do operate like that.
Some systems can and do operate like that.
it could do. but eg the other day I was watching a match where one team's odds fluctuated between 4.1 and 4.6 during half time. most systems would stand or fall on where they got in that kind of range, I think.
Here we are giving away edges again, or encouraging growth in liquidity which ever you look at it, but yes Presidente it is more likely than not that a half time price will drift over the course of 15 minutes.
Not giving much away there Trevh.
I'm still looking at the 0-0s.
Must be blind though, or not really seeing what I'm looking at or for.
Time decay or lack of I think.
I'm still looking at the 0-0s.
Must be blind though, or not really seeing what I'm looking at or for.
Time decay or lack of I think.
Frog, I think it would be quite difficult seeing the decay of the volume via the website with a manual refresh, but if a match kicks off with say 10k on a single tick it may well take several minutes for the backers to eat their way through it, depending on match activity and who wants to back.
Of course prices move at HT You've got the pre off players making a return and the gamblers dumping their winning or losing bets.
Ok, I may have got the wrong impression on whether prices move very much, probably because I don't really have time to study any one particular match, it was just an impression I got as I work through my method.
All I have done so far is take what I see at the time that I see it, working my way through the in-play coupon and entering the odds available into my spreadsheet. No doubt my system has made some selections where I could have got better odds, and sometimes I have got better, it's just been down to luck at the time I looked at that match. During busy periods, though, like on a Saturday afternoon, I haven't got time to mess about, so this is the way I will be making bets too. You win some, you lose some, hopefully it all evens out in the end. Of course, if a perceived weaker team is 3-0 down at half-time and offered at 200-1 to win I will obviously try to get matched at 500 or more...you have to chance your arm sometimes.
FAFH - If march results generally show large variances from the norm, then this will be an ideal time for me to start :-)
All I have done so far is take what I see at the time that I see it, working my way through the in-play coupon and entering the odds available into my spreadsheet. No doubt my system has made some selections where I could have got better odds, and sometimes I have got better, it's just been down to luck at the time I looked at that match. During busy periods, though, like on a Saturday afternoon, I haven't got time to mess about, so this is the way I will be making bets too. You win some, you lose some, hopefully it all evens out in the end. Of course, if a perceived weaker team is 3-0 down at half-time and offered at 200-1 to win I will obviously try to get matched at 500 or more...you have to chance your arm sometimes.
FAFH - If march results generally show large variances from the norm, then this will be an ideal time for me to start :-)
White Sock Offender, I am not a mathematician as well. Sorry if what I wrote was unclear, I am not a native speaker. Basically you have to sort your wins and losses into Excel spreadsheet, e.g. in a column you have -20, -15, +1 (win), +1 (win). That is 1. from my post.
2. =AVERAGE(D1:Dn)
is the formula for the mean yield, basically an average of those same wins & losses (the D1:Dn is just that same area of your wins and losses, from cell D1 to D45454, wherever you have your data, of course D is just an example)
=(STDEV(D1:Dn))/(SQRT(COUNT(D1:Dn)))
this is a formula for Std. Error of the Mean
3. =NORMDIST(0;E1;E2;1)
is the most important formula, you put that in one cell where E1 is the mean yield calculated in 2. and E2 is std. Error of the mean. If you are still in the mist, I suggest you read this page, there are also more simpler tests:
http://www.punterslounge.com/forum/f23/chi-square-test-76084/
where all is explained (more clearly?)
The end result should show how robust your sistem is in the long run or how much the results were achieved by chance.
2. =AVERAGE(D1:Dn)
is the formula for the mean yield, basically an average of those same wins & losses (the D1:Dn is just that same area of your wins and losses, from cell D1 to D45454, wherever you have your data, of course D is just an example)
=(STDEV(D1:Dn))/(SQRT(COUNT(D1:Dn)))
this is a formula for Std. Error of the Mean
3. =NORMDIST(0;E1;E2;1)
is the most important formula, you put that in one cell where E1 is the mean yield calculated in 2. and E2 is std. Error of the mean. If you are still in the mist, I suggest you read this page, there are also more simpler tests:
http://www.punterslounge.com/forum/f23/chi-square-test-76084/
where all is explained (more clearly?)
The end result should show how robust your sistem is in the long run or how much the results were achieved by chance.
If march results generally show large variances from the norm, then this will be an ideal time for me to start
Fwiw, March results don't stand out in any way in my footy records.
Fwiw, March results don't stand out in any way in my footy records.
Thanks for your time and patience Sun Wukong. I have tried your formula and after a lot of messing about with my spreadsheet, and between yourself and Punterslounge I have managed to come up with the following........drum roll................0.04067.
I have no idea what this means, but I assume it's OK?
Trevh - So your advice would be to wait as long as possible at half-time before placing my bets, as I will get better value in the long run?
I have no idea what this means, but I assume it's OK?
Trevh - So your advice would be to wait as long as possible at half-time before placing my bets, as I will get better value in the long run?
If your backing in the C/S market, White Sock, then yes without any doubt. As the first half is drawing to a close, the MMB will be offering [say] 4.2 and 5.0, sometimes a larger spread than that. Backers will be backing the 4.2 at poor value in the dying seconds of the first half, then layers money will enter the market while the price is static and jump in front of the Bot, pushing the price out. It wouldn't be unusual for the 2nd half to kick off at 4.6 and 5.2 in this example.
I don't know what you're doing or how you go about it, but I would want to be asking for 5's in the above scenario, and would never never take what's on offer at 4.2/4.6. The difference between taking and requesting a price makes the world of difference long term.
I can't speak for other markets but I would imagine the drift may follow suit.
I don't know what you're doing or how you go about it, but I would want to be asking for 5's in the above scenario, and would never never take what's on offer at 4.2/4.6. The difference between taking and requesting a price makes the world of difference long term.
I can't speak for other markets but I would imagine the drift may follow suit.
White Sock Offender, that basically means that there is a 95.954% possibility that your system will behave like it has, that is produce those results. But have in mind you sort of 'aftertimed'.
I would start placing real bets but with minimum stakes - couple of cents or pennies or whatever. That way you can check if all your bets will get matched. I would do this for a year or two or for 2000 bets, my advice would be at least a year. If after a year the results stay or are better it can be assumed that the trend will continue, that is if parameters affecting the system stay the same.
If you are too anxious you could use a medium bank (amount you could lose without an impact on your life) and build it slowly.
One more thing: discipline. What does it mean? With a strike rate of 28% there is almost a certainty of a 14 losing bets in a row, more unlikely 20 bets in a row and a 1% to lose 30 bets in a row. If you used your life savings and reached losing streak of 15, 20, would you have the discipline to continue?
I would start placing real bets but with minimum stakes - couple of cents or pennies or whatever. That way you can check if all your bets will get matched. I would do this for a year or two or for 2000 bets, my advice would be at least a year. If after a year the results stay or are better it can be assumed that the trend will continue, that is if parameters affecting the system stay the same.
If you are too anxious you could use a medium bank (amount you could lose without an impact on your life) and build it slowly.
One more thing: discipline. What does it mean? With a strike rate of 28% there is almost a certainty of a 14 losing bets in a row, more unlikely 20 bets in a row and a 1% to lose 30 bets in a row. If you used your life savings and reached losing streak of 15, 20, would you have the discipline to continue?
SW - Thanks again, I had a good 'gut feeling' about this system but it's nice to know it's backed up with some statistics.
I don't worry about long losing streaks as I am making so many selections that they are swallowed up without me having time to worry about them. I have a cumulative profit graph, and the only thing that concerns me is the trend, but I know that I can go hundreds of games in a slightly downward direction before the next 'step' up. As long as the steps up continue to be bigger than the downward slopes, I'll be Ok.
You're right though, it takes discipline to continue with the same stakes and same method when you can see the graph heading downward instead of up, but it makes it easier if you are confident you are doing the right things, which I now am.
I would say that most of the prices I have recorded at half-time have been early ones, so based on Trevh's advice, the system should be even more profitable than it is showing at the moment. I'm looking forward to going live with this, starting small, and hopefully building it up to levels where it can generate a nice second income...it's nice to dream isn't it?
If it doesn't work and I lose my starting bank then hey ho, at least I gave it a try.
I don't worry about long losing streaks as I am making so many selections that they are swallowed up without me having time to worry about them. I have a cumulative profit graph, and the only thing that concerns me is the trend, but I know that I can go hundreds of games in a slightly downward direction before the next 'step' up. As long as the steps up continue to be bigger than the downward slopes, I'll be Ok.
You're right though, it takes discipline to continue with the same stakes and same method when you can see the graph heading downward instead of up, but it makes it easier if you are confident you are doing the right things, which I now am.
I would say that most of the prices I have recorded at half-time have been early ones, so based on Trevh's advice, the system should be even more profitable than it is showing at the moment. I'm looking forward to going live with this, starting small, and hopefully building it up to levels where it can generate a nice second income...it's nice to dream isn't it?
If it doesn't work and I lose my starting bank then hey ho, at least I gave it a try.
note that what trev says is about the CS markets. can't comment on that as I don't do them. but for MO markets, there's no set direction prices move in at half time.
but theory would suggest that the longer HT goes on, the more "correct" the prices will become. so if you think you've found an edge, it may be that the beginning of the break is the better time to get on.
but theory would suggest that the longer HT goes on, the more "correct" the prices will become. so if you think you've found an edge, it may be that the beginning of the break is the better time to get on.
Thanks Viva, I had missed that trevh was talking about correct score, which I don't do. While I am still testing I will now try to monitor what happens to prices during half-time to see when it is best for my particular system.
Thanks Sun, your help and advice is much appreciated, I wish you all the best too.
Post Your Reply
Please login to post a reply.
Wonder
Instance ID: 13539