Will this catch on? Care to share?

[Matthew]

A great way to know if something is going to work is to backtest........

This is only possible with data......I can't begin to think how many of us,
particularly in this systems and strategy area of PL, spend time playing
around with data, calculating ratings, testing this that and the other......

Some people may feel protective of their hard work......but would you feel
so defensive if you had been helped by the others on here?

The number of spreadsheets, text files, access databases lying around
deemed useless by members must run into the 100s.....1000s maybe.

Wouldn't it be great if we all shared the info, different stats, different
takes on similar topics, widely available information - but in a form that is
useful.

We all help each other with advice - why not start helping with your
redundant or current data.

I'll start:

Something I have been playing around with is Corners in Football. With no
odds data readily available I've aggregated the betfair data to find
weighted odds ((volume matched * odds) / volume matched) for all the
EPL matches offered in 2005. This results in 102 games.

File Layout:
League - E0 for EPL
Date - Date of match
Home Team
Away Team
9 or less - betfair "weighted odds" for this market
10 - 12 - betfair "weighted odds" for this market
13+ - betfair "weighted odds" for this market
avg(home team corners in last 3 home games) - expected home team corners
avg(away team corners in last 3 away games) - expected away team corners
Actual Corners - The number of corners that actually happened in the match

What have I found out so far?

If the expected home team corners + expected away team corners > 12
then backing the 13+ market would result in the following from the 102
matches in this dataset.

At £5 level stakes

No of bets:___33
Stake:___£165.00
Return:__£204.45
Profit:____£39.45
Yield(%)___23.9%

I have not taken into account betfair commision.

I'm hoping this inspires others to share as I think this is something that
would add to this already awesome community.

Yes some will leech all the info and never contribute - do I care? Probably.
But this is very much in its infancy and I can see a group of the more
data-orientated types on here sharing between each other and possibly
not in such a public way as this.

Lets help each other beat the bookmaker.

Care to share?

[Hodgey]

Nice post Matthew and nice idea too.

Good luck with those corners - looks very promising indeed.

As a by product from my Laying Correct Scores thread, I am building up a record of Betfair odds on all Premiership matches this season for 12X and O/U. If anyone needs/wants the prices then feel free to ask.

[georgej]

I've been tinkering about with the horseracing betfair data in excel for backtesing purposes. I have all the betfair data saved into workbooks and i have another testing workbook that will go through each of my selections and tell me what the last Win or inplay odds were for that horse were before the race started. It is over 1GB of data and takes about 5 sec per seelction - which i can live with if i set it up to run before going to work. Got 1 or 2 bugs to work out, but once i get it working will hopefully be in a better position to judge what systems may or may not work - and why.

Just going to work on that last bug now. Then maybe try to look at some proper database software.

[Matthew]

Originally Posted by Hodgey

Nice post Matthew and nice idea too.

Good luck with those corners - looks very promising indeed.

As a by product from my Laying Correct Scores thread, I am building up a record of Betfair odds on all Premiership matches this season for 12X and O/U. If anyone needs/wants the prices then feel free to ask.

WOO HOO.......at last some interest.

I was beginning to think it was gonna be a lonely old forray.

Does anyone have any ideas of somewhere we could store the data?


I think it is the one thing this forum is missing -what better way to get
people developing systems and trying their hand than having the data
there for them on day 1. The more people that try creating systems and
testing them properly, the better.

Anyone else out there with anything to share?


georgej
- sounds good - are you offering to share this when you have
bugfixed and found the most appropriate database software?

[georgej]

georgej
- sounds good - are you offering to share this when you have
bugfixed and found the most appropriate database software?

That was my intention...
But it looks like there are a few discrepencies in Betfair data that will make it difficult. ' is either removed from a name or a space added to the name. Found one called "Franks Quest" where the ' was removed and there were 2 spaces between the words rather than 1.

Not sure i can make anything useful out of it as i need to keep going into the code to bypass certain discrepencies that crop up.

Also not able to clearly identify the win market. Most go inplay, so that is easy enough to spot, but the ones that don't i need to look up manually.

So far this is beyond me how to fix, but hopefully i'll get a few new ideas to play with. A decent database system should enable me to specify a date (which excel has been trying to convert to US ) and from there hopefully typing in part of a word and the remaining matches are displayed...
Ok - got a few more ideas now

Good idea on the sharing though Mathew - be it able to ask for specific advice that people have readily available or will make available and can be expanded by others. Not sure of how the layout may take shape, but worth considering.