Random Probability

[mattio]

I am hoping someone with some mathematical probability knowledge will be able to help me here. I am looking at a possible favourites lay system but I need to know if the testing results I have are at least a semi-accurate indication of of the future. The system I am looking at considered 1000 races over the past 4 months and selected 70 races to lay the favourite, if I were to randomly select 70 races in a group of 1000 what would be a probable win percentage of favourites in those 70 races?

Cheers,

Matt.

[The Ocho]

I'm no rocket scientist (or mathmatese) but wouldn't it still be the long term favourite winning percentage of around 30% - especially in the long term?

[moeee]

Its not Rocket Science OCHO.
You are correct with 30%

Sometimes people try to make simple things complicated.
Like the Origin of the Universe.
Something made it.
Lets call it GOD.
Too easy.
The Rocket people can have their Big Bangs and there Primordal Soup.
I got a Life that needs Living.

[UselessBettor]

70 is too small a sample size ans saying 30% is too simplified. what if I chose 70 races where the favourite was $1.50 or less. I would expect at least 50% strike rate not 30%. What if I chose 70 races randomly but they all had favs at $4+ prices. In that case the strike rate should be 25% or less.

The better option is to determine how many winners you should have got from those races. Its simple enough to work out . 1/odds = chance of winning. Add up all of the favs using this formula and it tells you how many winners in those 70 races were expected. This doesn't account for commissions so you should adjust for those too if you using tab prices. For betfair you don't have to worry as commission comes out after the winning bet so the odds on display are generally to 100%.

[The Ocho]

And when you take all those various 70 races into consideration, I'm betting (in the long term) it will come out at around 30%. You haven't seen the thread about the wisdom of crowds then, UB?

[Dale]

Hey Mattio.

Probably looking for a quicker answer than this but if it was me id look at it like this.

Id work out the strike rate of your favorite in those 1000 races then wait till your 70 selections grows to 1000 and see if it stacks up.

Worse or the same throw it out, significantly better youre onto something.

[moeee]

And when you take all those various 70 races into consideration, I'm betting (in the long term) it will come out at around 30%. You haven't seen the thread about the wisdom of crowds then, UB?

Some people you can rub their noses in it and they still won't see or smell it.
Its why we have Winning Punters and Losing Punters.

The ONLY reason a Punter Loses is because he is doing something wrong.
And until they realize and understand what it is , and make it good , then they will continue to fail.

[mattio]

70 is too small a sample size ans saying 30% is too simplified. what if I chose 70 races where the favourite was $1.50 or less. I would expect at least 50% strike rate not 30%. What if I chose 70 races randomly but they all had favs at $4+ prices. In that case the strike rate should be 25% or less.

The better option is to determine how many winners you should have got from those races. Its simple enough to work out . 1/odds = chance of winning. Add up all of the favs using this formula and it tells you how many winners in those 70 races were expected. This doesn't account for commissions so you should adjust for those too if you using tab prices. For betfair you don't have to worry as commission comes out after the winning bet so the odds on display are generally to 100%.UB the total odds of all those was 264.60 (NSWTAB) so is the formula (1/264.60)*117% to account for the TAB takeout?

[woof43]

Results for 10 groups of randomly selected races (70 consecutive races)
Losers
51 0.729 0.271
44 0.629 0.371
48 0.686 0.314
50 0.714 0.286
42 0.600 0.400
53 0.757 0.243
49 0.700 0.300
47 0.671 0.329
49 0.700 0.300
43 0.614 0.386
0.68 0.32
In the groups selected the longest expected Losing run for the fav worked out to be 15, but 1 group had 16 consecutive losing favs.

[mattio]

Hi woof43,

Thanks for that, would you mind explaining what the groups of numbers mean please? I am guessing the last number in the row is the win percentage but I am not sure about the others.

Cheers,

Matt.

[1annandale1]

Not sure this helps Mattio

These are some actual figures from a laying system i have been doing.
All were favorites on betfairat sometime in betting

From 20/3/2010 to 20/4/2011

Bets 2013

Accidents 510

Av price 3.7

Laying to liability of $2000

More than 90 % were actually Layed

The balance were all documented(They werent layed because of phone or internet problems)

[woof43]

first column is the number of losers in the group of 70 races, next column is that losing % and third as you stated is the win %.
the two figures at the bottom are the averages

[Shaun]

Not sure this helps Mattio

These are some actual figures from a laying system i have been doing.
All were favorites on betfairat sometime in betting

From 20/3/2010 to 20/4/2011

Bets 2013

Accidents 510

Av price 3.7

Laying to liability of $2000

More than 90 % were actually Layed

The balance were all documented(They werent layed because of phone or internet problems)

75%sr is not very good in laying.

[mattio]

1annandale1 - thanks for sharing your results mate, looks like you get quite a bit of action there.

woof43 - ok mate, that clears that up for me now thanks.

In the 4 months of testing I have done so far the results are as follows, these are based on win betting the selections as they are TAB prices but you can see how it could transform to a lay system.

70 selections
5 winners
$3.80 avg price

7% win S/R, 93% lay S/R
-79% LOT win betting
Longest run of outs 32

It's only early days as it is such a small sample so I will continue to test for a few more months.

[moeee]

70 selections per month?
My Goodness - you are very patient.
I know I couldnt survive on a single wager every other day.

[mattio]

No moeee this is over a 4 month period. I have multiple backing systems I use that give me enough bets so I thought I would see if I can get a few lay methods to go along with it.

[UselessBettor]

UB the total odds of all those was 264.60 (NSWTAB) so is the formula (1/264.60)*117% to account for the TAB takeout?
Mattio,

You should do it for each fav. ie 1/$2 + 1/$3 = 0.83333

But from your figures I assume is over 70 races so we can assume odds of 264.6/70 = $3.78 per fav. This means there were 1/3.78 * 70 = 18.52 winners.

If these are tab prices I would be very careful about laying them as betfair prices are not always X% higher. Sometimes they are lower, mostly higher, and sometimes the favs are different.

If they are tab prices then its likely the number of winners should be 1.15% higher which would be 21.3 winners from the 70 races.

Again its based on the assumption this is 70 races for the figures you gave me.

Also my maths may be wrong so always double check it.

[UselessBettor]

And when you take all those various 70 races into consideration, I'm betting (in the long term) it will come out at around 30%. You haven't seen the thread about the wisdom of crowds then, UB?
Long term yes 30% is probably right .... But 70 races is not long term. It could be in a very large range of 10% - 50% strike rate over only 70 bets.

Using the right methods at the right times for testing can save a lot of heartache and give a much better understanding of when a short term swing above or below the long term strike rate is really what you are seeing.

[mattio]

Mattio,

You should do it for each fav. ie 1/$2 + 1/$3 = 0.83333

But from your figures I assume is over 70 races so we can assume odds of 264.6/70 = $3.78 per fav. This means there were 1/3.78 * 70 = 18.52 winners.

If these are tab prices I would be very careful about laying them as betfair prices are not always X% higher. Sometimes they are lower, mostly higher, and sometimes the favs are different.

If they are tab prices then its likely the number of winners should be 1.15% higher which would be 21.3 winners from the 70 races.

Again its based on the assumption this is 70 races for the figures you gave me.

Also my maths may be wrong so always double check it.

Thanks UB, I will be testing for a lot longer before placing any bets and also taking into consideration the price difference. I was mainly looking to see if the method was able to find less winners than what could obtained simply by random chance. Would you mind giving me your email, I would like to run something past you.

Cheers,

Matt.

[Bhagwan]

Hi Mattio,
I feel you are taking the right approach by using multiple lay plans.
Just make sure that one does not double up the bet when there is 2 or more selecting the same Horse.
Because it does not mean that it has twice the chance of falling over.


Coming back to your original question as to what the SR of winners would be , the general answer is approx 20%

I think you will find it rather challenging, long term to get it much lower than that 20%.

The general SR of Favs is 30%
Take out all the odds-on shots <=2.00 & its 25%.

Here is an exercise you may like to try .
Grab 10 lots of 10 races in any order = 100 races
Get 10 of your mates to select just 2 Horses to fall over from those 10

I can almost guarantee that it will average out that 2 from 10 will still win instead of lose = 20% win SR

Its amazing how many can win in spite of terrible form or barrier draws ect,.

If you can secure only 20% winning at a max price of 4.00
Then that leaves us with an acceptable profit of 20%+ POT.

You now laughing like a Hyena.

[moeee]

Hi Mattio,
Coming back to your original question as to what the SR of winners would be , the general answer is approx 20%


I'll have a guess too.
I'll go the other way.
Quite a number of those selections would be false Favourites.
So my guess is the general answer is 40%

Go Go Mattio!!!