value system help

[presto]

intro
this system was a long time in the making, origionally designed for football betting i am currently running it on 'american sports' at the moment - with only 2 outcomes, this simplifys it to a certain degree - but it works on the same principals and can be applyed to any form of betting with accurate ratings.

the system
this system is purely value orientated, no oppinion is put into the system, it just point's out where the value is. because of this every game is rated and given a value, no matter how small.

staking
loosely based on the 'kelly' staking plan, though in reality it has very little in common with the kelly staking plan other than value is the aim.
the 'kelly' stakes are dependant on value, and stake to a set %. the modifications i have used are:
- %stake variable on winning % chance. (ie that 2/1 shot available @ 3/1 is deffinately value - but no chance i will stake 10% of my bank on it like 'kelly' would have you do), instead the shorter the odd's the more (though sensible ammount) i would stake on it. and the larger the odd's, the less i will stake.
- there are plenty more modifications i have made but will keep it short, basically.
> bet the value
> the more value the bigger the bet
> lower the odd's the more the stake
> higher the odd's the less the stake.

..................................................

so there you have the basic outlines for what should be a profitable system + staking, value in many punters eyes is the most important factor in successfull betting, to some degree i can agree with this, but in many ways i disagree. IMO a 'banker' may not be the value, but i wouldn't oppose it for that reason.
however - in the short time i have been testing the system it's been a bit of a disaster.

starting bank £1,000
profit -£140.73
current bank £859
adverage stake £17
adverage odd's 3.46
selections 23
wins 7
losses 16
liability £398.83
SR 30.43%
POI -35.29%
% increase -14.07%

ok it's still too early to come to any conclusion, but the -35% POI is already worrying,
the odd's are to be honnest what i would expect, favouring the outsiders - i formed this oppinion as people tend to bet on the team they think will win, not pick the value,
picks - 8 / 23 were on the favs - leaving 15/23 on the outsider.
stakes - the adv stake is £17, or 1.7% of the bank. happy with this - nice and safe %. there have been 2 bets that have been way off though.
a 6% stake on a 9.4 shot (preceived true odd's 3.7) so despite the big odd's there was massive preceived value, i was alright with the stake as a very rare opportunity, the stakes have only been 2 other 3% bets.
there has been one 17p bet basically because the odd's were spot on. but stakes tend to be £5-£20.

...............................................

i have thaught up a couple of 'extra' systems based on this method:
> only bet value favs.
hopefully a more steady system

> only bet with 'X ammount +' of value.
so only the big value bets get taken, hopefully improved POI, however this tend's to mean outsiders, so lower SR.

> X ammount of value +, on favs
the favs should mean a higher SR however getting big value on fav's is rare, thus very few bets.

.................................................

i am happy to carry this system on as it is, as 'value' is something that can only be relyed upon long term, and my sample is way too small to draw any conclusion, but i am just opening this up for any discussion / suggestions early. all responses are appreciated.

[Win2Win]

Not bad. 7600+ posts to get to a decent one :)

[mathare]

How are you doing your ratings? If the whole idea of the value is based on your comparison of the odds on offer against what you think the odds should be based on ratings you need to have a decent degree of confidence in your ratings. Perhaps you can tweak the rating equations and reapply the ratings to a number of past games and see how that affects the figures. Make sure you don't tweak the ratings to look good over a certain set of games only to find they bomb when applied to another set of games though

[MattR]

Interesting you should be doing a thread on value odds as that's something I've been playing with the past few weeks on football and it's looked quite good so far, but very early days and could be no more than trialling it at a peak. Now you've posted this you've given me the inclination to post up this weekends (ready for the idea to fail dismally once posted! )

Out of interest what would pr/lo be if you'd done level stakes on those?

[presto]

with level stakes: (£17)

starting bank £1,000
profit -£124.78
current bank £875
adverage stake £17
adverage odd's 3.46
selections 23
wins 7
losses 16
liability £391.00
SR 30.43%
POI -31.91%
% increase -12.48%

value stakes:
starting bank £1,000
profit -£140.73
current bank £859
adverage stake £17
adverage odd's 3.46
selections 23
wins 7
losses 16
liability £398.83
SR 30.43%
POI -35.29%
% increase -14.07%

looks like the 'value' stakes is slightly ahead - something i didn't think it would be - as on of the value stakes was £60 (lost).

[MattR]

Are the fave value bets in profit or the underdog if seperate? Or are both showing minus if done seperately?

[presto]

the info is from another site (subscription) - look very accurate, and i apply my own ratings system to it, and number crunching is long term profitable, however the value aspect i use is something completely different to what they use (handycapping). but basically the same principle.

handycapping is also something that i will be looking into in the future.

[presto]

the odd's on are 4/4,
however 2 of thoes losses were absalute shockers, and the other 2 looked good bets to me.
the real 1.22 shocker was only a 17p bet (would have picked the outsider for a small stake - but the odd's changed a tick when placing the bet - thus changing the selection)

[presto]

a bit of a recovery with last nights 2 games.

£16.9 @3.9 on the NJ nets away to chicago, i thaught the odd's may have been higher, however a look at the stats says NJ were a good bet, as they have a very good record against the bulls, and by recent form trends. last night's game was similar to last fridays game involving these two - with the bulls getting a big lead and then letting it slip (last fri - went 18-0 up in the 1st) - last night went 15-28 in the 1stQ.

the other game was on the cleaveland @ pheonix, the odd's were near spot on - early on it was a no-bet (spot on odd's with a 0.2% overound) so the 0.2% overound put both sides as negative expectation. closer to the game the odd's on pheonix shortened from 1.29 - 1.24 (or there abouts) making cleaveland small value - for a small £2 stake. pheonix won convincingly in the end.

[MattR]

Good one on NJ then Presto. Just curious as to how the bank would be if you used a set target win on the bets, say £20 or whatever so the higher stakes would be on the favourites and you'd get a cheap bet on the underdog when they are the value.

[presto]

good idea that matt - will have a look later tonight.

[presto]

had a fantastic night last night 9/14
£216 profit - to get it back on track.
3.9 - 4.0 - 3.9 - 2.4 - 2.9 - decent stake winners
only £10 staked between 4 losers
a £25 stake did go on a 2.7 loser though :( the bet was looking a winner 1.04 shot - then it went to overtime , and lost - that's 4/4 bets that have gone to OT and lost (inc 3 favs).