Football Musings

[Bobbyjoe]

Your thoughts please.......
I have been betting on Draws for just over a year, just shy of 500 bets.
I have found that my average win odds are exactly the same as the average of all odds bet. I only bet over 3.14 odds and the highest is about 4.00. All bets are placed on Betfair and I stake more on the higher price Draws than I do on the lower price Draws. This gives me a 10.5% roi as opposed to 8.7% level stakes. Can I read anything into this as usually the average price of a winner would be less than the overall average price of all bets placed. Go easy, I have a dream!!

[andy28]

In one of Peters video's he suggests it is the other way around, you should bet more on the shorter priced bets. So for you say you want to win $10 if the odds are 3.15 you should bet $4.65 to win $10, If the odds are $4 then bet $3.33

[wearthefoxhat]

Your thoughts please.......
I have been betting on Draws for just over a year, just shy of 500 bets.
I have found that my average win odds are exactly the same as the average of all odds bet. I only bet over 3.14 odds and the highest is about 4.00. All bets are placed on Betfair and I stake more on the higher price Draws than I do on the lower price Draws. This gives me a 10.5% roi as opposed to 8.7% level stakes. Can I read anything into this as usually the average price of a winner would be less than the overall average price of all bets placed. Go easy, I have a dream!!

My view is, you should be looking to have a bigger bet when your tissue book % price for the draw is higher than the live odds book %.

Book%.png

------

Then, stake covering to win £x with your edge (8%). The way you could achieve this is to calculate the CTW against your tissue price and not the live price.

CTWEg.png

-------

I've attached a spready showing this. Not sure how you are with excel. (It's quite simple to use though)

Column C enter the Live prices.
Column D enter your tissue prices.
Column G enter 1 if Win.

Cell G3 (Cover to win) Enter the desired stake.

The win stake column returns 0 if there is no edge %

The concept is, the more your tissue price beats the live market, the more profit you make long-term.

-------

Value CTW.xlsx

[Bobbyjoe]

Thank you for your reply's and especially to wearthefoxhat. Having been interested/betting for over 40 years, my personal experiences have led me down many different approaches to staking and value. I have concluded that in regard to finding value, I personally are unable to find a method of predicting the probability of an individual outcome with any great accuracy. It therefore doesn't make sense to stake according to an individuals price. In fact over the years I found I made more profit, by betting on those with the least value according to my figures. People say don't bet if there is no value, which is true to a degree, unless of course every bet you make is a winner, then you don't need value, just a large safe. I now operate the same way as a bookmaker, I am interested in the overall returns on a large sample of bets not trying to find value in one selection. I know this doesn't sit well with the majority of successful bettors, but it is working for me, so far!! I know my sample of bets have value, I just wait for it to show up in my stats rather than find that elusive value in individual bets. Sorry to bore you all.Good luck to everyone...

[andy28]

The big match coming up I have my prices for Arsenal/City, Arsenal 2.08 and City 3.66 so looks like currently City is value

[wearthefoxhat]

The big match coming up I have my prices for Arsenal/City, Arsenal 2.08 and City 3.66 so looks like currently City is value

Nice one.

----

Mine lean more towards Arsenal. Seems 1-0 / 2-0 and the Under 3.5g show up well for value purposes.

I've added in an "Elo adjustment factor" to take into account an overall teams' strength/weakness in that league.
(Elo data taken from Sinceawin)

PD Model.png

-----


My value range expectation is higher for match odds. Although Arsenal show as around an 11% edge, there's no bet in that market.

Arsen v MC.png

[wearthefoxhat]

Exploring alternatives to Poisson D as it does have a tendency to skew the draw and the outlier score-lines. ie: 2-6, 7-0, 4-4...etc

Results have been okay though, but it's good to learn new stuff to run along side it, so I asked my ole mucker GPT for alternatives (Free version). Looks like it put some thought into this one....

GPTFooty.png

[wearthefoxhat]

Cracked the Zero Inflated Poisson issue. (ZIP)

ZIP.png

[wearthefoxhat]

If I were just straight value-backing, when the alert fired, the recent Championship games would have done well.

The staking is designed to put more on the better the value prices and is calculated against the tissue price not the live odds available and set up as a minimum of 1pt to a maximum of 20pts.

Championship.png

[TupleVision]

If I were just straight value-backing, when the alert fired, the recent Championship games would have done well.

The staking is designed to put more on the better the value prices and is calculated against the tissue price not the live odds available and set up as a minimum of 1pt to a maximum of 20pts.


Championship.png

Are you using Kelly Criterion out of interest? I use dynamic Kelly in my model (although I'm yet to decide if I will use this as everything is still in the testing phase) that is to say as I'm laying I don't want too large liability on larger odds so I use 25% Kelly if < 1.99, 12% Kelly if under 8.0 and 6% Kelly for all else calculated on a sliding scale based off the perceived edge.

[wearthefoxhat]

If I were just straight value-backing, when the alert fired, the recent Championship games would have done well.

The staking is designed to put more on the better the value prices and is calculated against the tissue price not the live odds available and set up as a minimum of 1pt to a maximum of 20pts.


Championship.png

Are you using Kelly Criterion out of interest? I use dynamic Kelly in my model (although I'm yet to decide if I will use this as everything is still in the testing phase) that is to say as I'm laying I don't want too large liability on larger odds so I use 25% Kelly if < 1.99, 12% Kelly if under 8.0 and 6% Kelly for all else calculated on a sliding scale based off the perceived edge.

No, not Kelly criterion.

With my staking sheet, once the alert is triggered, it then checks to see if there's an edge (my value), then it suggests a points stake between 1pt and 20pts. So far, it has only suggested one 20 pts coup, Newcastle (Southampton v Newcastle 1-3) Market odds were 1.46 my tissue was 1.02

The points can be framed according to the bankroll. ie: 20pts = 10% 1pt = 0.50%

Of course, the more accurate/reliable the model is, then all should be good.

[TupleVision]

If I were just straight value-backing, when the alert fired, the recent Championship games would have done well.

The staking is designed to put more on the better the value prices and is calculated against the tissue price not the live odds available and set up as a minimum of 1pt to a maximum of 20pts.


Championship.png

Are you using Kelly Criterion out of interest? I use dynamic Kelly in my model (although I'm yet to decide if I will use this as everything is still in the testing phase) that is to say as I'm laying I don't want too large liability on larger odds so I use 25% Kelly if < 1.99, 12% Kelly if under 8.0 and 6% Kelly for all else calculated on a sliding scale based off the perceived edge.

No, not Kelly criterion.

With my staking sheet, once the alert is triggered, it then checks to see if there's an edge (my value), then it suggests a points stake between 1pt and 20pts. So far, it has only suggested one 20 pts coup, Newcastle (Southampton v Newcastle 1-3) Market odds were 1.46 my tissue was 1.02

The points can be framed according to the bankroll. ie: 20pts = 10% 1pt = 0.50%

Of course, the more accurate/reliable the model is, then all should be good.

Makes sense

[wearthefoxhat]

Small sample, but here's how the ZIP (Poisson) qualifers have performed when backing to win value-bets, outright.

Totals.png

[TupleVision]

Any reason League Two has negative ROI, just variance do you think? Results look promising though, is this taking value bets at halftime only?

[wearthefoxhat]

Any reason League Two has negative ROI, just variance do you think? Results look promising though, is this taking value bets at halftime only?

No, it's the value-bet taken pre-off to show that the set-up is value to begin with. The HT approach is preferred going forward, as it confirms the stats pre-off.

The league two results could be down to variance, or, the teams' dependence on the fitness levels of key players. The sample size for all the leagues are small at this stage to really tell.