Forums

We want to share an important update with our Betfair Community users.

Over the coming months, the Betfair Community platform will be winding down. The final active day for the platform will be December 21st. From December 22nd, users will no longer be able to access the platform.

Your discussions, insights, engagement, and, most importantly, feedback have helped shape the platform and guided innovation. Your feedback continues to be important to us, and we will continue to listen via our dedicated feedback channels, as well as through product testing and innovations.

To share your feedback and experience, visit: https://support.betfair.com/app/home

General Betting

There is currently 1 person viewing this thread.
Homer Jay Simspon
10 Jan 11 19:47
Joined:
Date Joined: 20 Aug 05
| Topic/replies: 182 | Blogger: Homer Jay Simspon's blog
There have been some very interesting threads on this forum lately and I've been inspired to do some serious research on Kelly staking and modelling football games using maths as my guide.  I'm going to try and and generate some profits using what I've developed over the last week or two and hopefully learn a bit more as time passes. Some of the questions that I want to answer for myself are:

1. Is Kelly staking optimal?
2. Does my model work?
3. Are Betfair football odds very accurate, or not?

So, without further ado I will get the ball rolling. The bank is £2,500 and I will use half Kelly stakes to begin with. I will only be looking at league games as I haven't yet figured out how to model form between sides in different leagues anywhere near accurately.

Bet 1 Hercules v Atletico Madrid

My model threw up almost precisely the same match odds as available on here (which gave me some sort of comfort, I think!) but the prices were quite a bit out of line on the Over/Under 2.5 goals market.

I have overs at 1.72 (58.13%) so 2.10 looks very attractive.  Being conservative I've used 55% in the Kelly calculator and it suggests a bet of £315.18 including an allowance for my commission rate, so that's £157.60 at half stakes.

Post your reply

Text Format: Table: Smilies:
Forum does not support HTML
Insert Photo
Cancel
Page 1 of 3  •  Previous 1 | 2 | 3 | Next
sort by:
Show
per page
Replies: 81
By:
Lori
When: 10 Jan 11 20:49
Good Luck Homer
By:
FINE AS FROG HAIR
When: 10 Jan 11 20:56
Nice start Homer.
That indeed was a good call.
And all the pressure off by half time as well.
By:
Homer Jay Simspon
When: 10 Jan 11 21:48
Thanks very much guys - it won't come as a surprise that both of you contributed a great deal to the threads I mentioned earlier from where I got the inspiration to work this maths thing out properly.

That was remarkably straightforward for the first bet - I'm sure plenty will not be so comfortable. A profit of £167.30 on the bet, so the bank climbs to £2,667.30 and that will be the basis for the next stake.
By:
Homer Jay Simspon
When: 11 Jan 11 18:09
Enthused by early success I ran all tonight's league 1 and 2 games through my model with some very interesting results. When I say interesting I think worrying would be a more appropriate word as the outputs threw up some major variances to the available odds. This either means that the markets are very wrong in places (unlikely) or that my model has some major flaws (highly probable). The resulting Kelly stakes are of course astronomical and way beyond my acceptable levels of risk. Of course the model could be right, and over the last week or two the test results have been very encouraging and profitable on paper, but some of these outputs look far too exaggerated to be plausible.

This is work in progress and I will put up the bets nonetheless, as maybe I will be pleasantly surprised by results. One immediate question has come to mind ... at full Kelly if you stumble across 1 or 2 exceptional value bets and they don't oblige then your bank would be decimated. It seems like there is a flaw in the staking methodology somewhere, albeit mathematically sound. Either that or such value bets never actually exist and when they do it's a sure sign that the model is weak - perhaps that's the flaw. Any thoughts on this would be welcome!

Ok, reverting to paper bets for these bearing in mind the issue above. All are half Kelly stakes:

Bet 2 Bristol Rovers v Walsall Over 2.5 Goals £648.50 at 1.84(!)
Bet 3 Peterborough v Brentford Over 2.5 Goals £407.42 at 1.70
Bet 4 Stockport v Rotherham Over 2.5 Goals £201.01 at 1.80

And the variance on these is so big that I won't even put any stakes up, but for the record there appears to be value here:

Bristol Rovers Draw at 3.70
Oldham Win at 3.80
Crewe Win at 3.25
Stockport Draw at 4.10

Apologies for the long post!
By:
DFCIRONMAN
When: 11 Jan 11 18:41
GL with thread Homer....Think FOOTBALL is so very hard to bet using KELLY ......but it does make you think harder re your stakes ....so is a better way to bet than LEVEL STAKES IMO...
By:
kenilworth
When: 11 Jan 11 20:54
I do my own ratings and 90% of the time my prices are close
enough to the market not to bet, which means I get one bet
in ten. Whenever I have a big difference I have another look.
GL.
By:
Lusitano71
When: 11 Jan 11 21:42
impressive stuff HJS especially the draws part, nice one

hope you ended up making the bets [;)]
By:
Lusitano71
When: 11 Jan 11 21:55
he had the over2.5 on both of them TM so the only results possible were the 2-2 and 3-3 although i don´t believe he went for the 3-3
By:
jonnyg
When: 11 Jan 11 22:14
good stuff Homer
By:
Homer Jay Simspon
When: 11 Jan 11 22:16
DFC, thanks for the good wishes. For me where you have 2 or 3 outcomes it feels much easier to model the probabilities than horse racing, where there are so many variables and possible outcomes. That's possibly down to my low level of maths skills though!

Kenilworth - that's what I expected to happen but I seem to have 4 or 5 out of 10 at the moment.

The Management - as it turns out 2-2 and 3-3 would have been a good bet in both the drawn games but I didn't have any bets tonight as I was concerned about the huge variance between my model and the prices here. Turns out I should have gone with them, but I'm just pleased to see some encouraging results at this point.

Lusitano - thanks very much for the kind words.
By:
Homer Jay Simspon
When: 11 Jan 11 22:33
Well, what a good evening that was ... on paper!

The bank would have risen to £3,623.36 with a profit on the 3 overs bets of £956.06 to half Kelly suggested stakes. However that would have meant risking 47% of the bank on three games, and unless any Kelly advocates can tell me otherwise that can't be a wise staking strategy long term. Maybe it's a case of having the balls to get the money down when such good value arises? Anyway I am gaining some faith in the model which is probably the most important factor at this early stage and I imagine that fractional Kelly stakes that are a smaller % of the total bank will have to be the way forward (1/4 or even 1/6 Kelly).

The match odds results were also very profitable thanks to those 2 draws, but there is no way I could have staked them with the £2,500 bank as well as the O/U2.5 bets at half Kelly.

More questions than answers tonight but that's all part of the fun!
By:
tobermory
When: 11 Jan 11 23:03
I did one of these threads..was a fkin disaster Silly .Sure you will do better  Homer.
By:
kenilworth
When: 11 Jan 11 23:55
Kelly is all about getting an edge of about 10% and anything bigger than that
you have to question how you calculate your edge. Professional compilers don't
get things wrong by much, especially with overs/unders and draws, and anyone
who thinks their price on two matches can carry so much as between £680+ and £201
will soon find out different, especially as the market had them both around
1.80 to 1.84. However, I wish the OP all the best and look forward to what comes
next.
By:
Trevh
When: 12 Jan 11 01:04
Apologies for dampening the enthusiasm, but I don't see anything here that luck does not explain. GL though Homer, but I can't see how guessing which teams will win/lose or have overs/unders is remotely possible long term. The BF odds are derived from the opinions of a huge amount of gamblers, and only those with the largest banks can shift them out of kilter to representing a value position, and even then not by much.
By:
The Investor
When: 12 Jan 11 01:48
I've seen odds out by more than 10% with large amounts available, though not recently. Guess those guys have gone bust Laugh
By:
Homer Jay Simspon
When: 12 Jan 11 09:52
Tobermory - I don't see this being any different, after all there are much brighter people than me out there who can't find an edge in football. But thanks for the good wishes!

Kenilworth - I agree entirely with your point around questioning how the edge is calculated when the variance is so large, something I mentioned before the matches last night which prevented me from putting any money down.

Trevh - whilst I agree with your point around luck, to be fair I am not really guessing. I've built a mathematical model which, although is no doubt very basic compared to some, is producing probabilities on all outcomes and where the variance is large (admittedly way too large) I am posting the bets. I think that's a bit different to guessing, but in time it may well be proven to be a flawed model. I have always shared your view on the BF market being accurate on football without really testing it for myself properly. I will say, however, that the BF market on horse racing is definitely not always accurate, despite being a representation of the views of many.

The Investor - the only good news is that I won't be going bust with this whatever the outcome, especially if I don't have the balls to put the money down!
By:
jonnyg
When: 12 Jan 11 10:02
Bristol Rovers Draw at 3.70
Oldham Win at 3.80
Crewe Win at 3.25
Stockport Draw at 4.10

I will be following this thread with great interest.

Can you not exploit this further.

1. Bristol Rovers were as low as 1.21 when 2-1 up. The strategy that i use was saying that Walsall were likely to score again so homer bet In Running as well.

1a Stockport 3-1 up so lay them im running at a short price.

2. I may be clutching at straws but you had Oldham win and Crewe win. Does this not mean that odds of draw were lower then expectation of draw so lay draw?

3. looking at your over bets the home teams overs are now 100% (peterborough) and 69% and 70% ie expectation of the game being overs.

4. Hercules game was screaming overs as i poinyed out on a thread. This was based on probability of early goals.

So chaps what do you think? In the hercules game unders were as low as 1.8 pre ko because i assume the market maker wanted to put some weak traders in the red pre ko.
By:
Homer Jay Simspon
When: 12 Jan 11 11:48
Hi jonny, thanks for input. Until I've got some idea on how the model is going to work, which will mean a lot more selections and testing, I won't be ready to try anything more adventurous. I'm keeping an eye on your SOT stats though and enjoy watching the games unfold and reading your thread (by the way I would pay particular attention to what Bayes said - he knows his stuff)

On the Oldham game I had the draw very close to the market so it wouldn't have been a lay. I did have the Torquay draw as slight value, but Crewe's value was much bigger which is why I put that one up. The thing is there will always be losers so I'm not worried about finding a winning bet every time, just a matter of beating the odds over a long series of bets.

Interesting that your stats also pointed to overs in the Hercules game - and that Atletico 1st half goal stat is amazing really for Spanish football.
By:
Homer Jay Simspon
When: 12 Jan 11 18:50
Before I continue I should warn anyone against backing these selections, my model is by no means proven yet and based on the points made by some experienced guys above, it will almost certainly prove to be inaccurate.

The bank sits at £3,623.36 and there are 4 stand out value bets tonight according to the model. Stakes are still half Kelly to keep things consistent, which still means a risk of 45.4% of the bank tonight:

Bet 5 £377.22 Blackpool Win at 4.00
Bet 6 £443.36 Blackpool Over 2.5 goals at 1.98
Bet 7 £453.80 Hamilton Under 2.5 goals at 2.24
Bet 8 £370.58 Hamilton Draw at 5.60
By:
jonnyg
When: 12 Jan 11 18:55
great stuff homer and ggod luck.
By:
jonnyg
When: 12 Jan 11 18:56
Blackpool overs at home this season=100%
By:
Homer Jay Simspon
When: 12 Jan 11 19:24
Thanks jonny - let's hope they keep up the 100% record!
By:
jonnyg
When: 12 Jan 11 19:34
Celtic Under 2.5 goals away is 62% with 4 0-0 ht draws so i can see that is a good bet at odds against.
By:
jonnyg
When: 12 Jan 11 21:33
well done homer.

with sot and your modelLove

if celtic equalise you have a full house
By:
Lusitano71
When: 12 Jan 11 21:35
now it must hold Laugh
By:
Lusitano71
When: 12 Jan 11 21:40
Celtic scored 1-2 Sad
By:
jonnyg
When: 12 Jan 11 21:41
this is thread of the year and it is only JanuaryPlain
By:
Lusitano71
When: 12 Jan 11 21:41
sorry forget that [:D]
By:
jonnyg
When: 12 Jan 11 21:47
what does a £100 acca pay?
By:
zero-zero in san siro
When: 12 Jan 11 21:48
3 out of 4 up - come on Blackpool! What a thread!
By:
jonnyg
When: 12 Jan 11 21:48
9934.84 i think
By:
zero-zero in san siro
When: 12 Jan 11 21:55
Acca up jonnyg! Well done Homer! Now, please describe your prediction model!
By:
Lusitano71
When: 12 Jan 11 21:57
amazing i cant believe it, i almost never follow predictions but today i did, i made all 4 bets (but half the stakes of HJS so i guess it was 1/4 kelly Happy) but i also traded along the way which went great, but i must admit that if Celtic scored and eventually wins i would've made a ton

anyway great stuff again HJS, congrats [;)]
By:
jonnyg
When: 12 Jan 11 22:05
well done again homer.
By:
kenilworth
When: 13 Jan 11 00:40
Take the money and run away as fast as you can...
By:
Trevh
When: 13 Jan 11 00:47
Nice start Homer, well done, but I think your followers are mad if they think this is any more predictable than the lottery! I'm quite suprised tbh, just goes to show why tipsters can make money when even seasoned gamblers believe :)
By:
I.quit.my.country
When: 13 Jan 11 01:12
no ken he should take the profit, ditch the kelly nonsense, use level stakes and watch the bank grow a little week by week.

The beauty of compounding and all that.

Level staking % of bank: it's not exciting and you need incredible patience, but it works a treat. Ramping up and down has no benefit LONG-TERM plain and simple. (I quit 2011)

we all compile ratings and books....if kelly is telling you to stake so high, your book is way out.....trust me

or can u look into the future?

kelly imo is just an excuse for people to gamble

no1 can be 100% accurate without the benefit of hindsight

if its not a 3pt max bet, then don't play at all - this is my view on ramping up and down

and when you hit 10% ROI week after week staking 3% of bank every bet.....

why would you want to lower the ROI by using a simple maths formula with an irish name?
By:
tobermory
When: 13 Jan 11 01:13
Thats a good nights work Homer Happy

What i'm not clear on is when you are betting on games taking place at the same time , and the bank is say 3,000 , and bet 1 is recommended to be 10% (£300) , if bet 2 is also 10% do you still count the bank as 3K and bet £300 on that too? Myself i would then consider the bank to be 2,700 and bet 10% of that.
By:
Lusitano71
When: 13 Jan 11 01:28
Trevh you take yourself too seriously man, chill out!

you dont get fat eating a lot in just one day
Page 1 of 3  •  Previous 1 | 2 | 3 | Next
sort by:
Show
per page

Post your reply

Text Format: Table: Smilies:
Forum does not support HTML
Insert Photo
Cancel
‹ back to topics
www.betfair.com