BackTesting Strategies Using OddsWizard

[odds software]

I know some are just mad gamblers who are happy to bet on anything without
having any idea of past performance.

I guess I myself prefer to be more analytical.

Long term strike rate / roi etc is best know in advance and based on a good sized historic data set.

Looking at the question of betting 1x2 on soccer matches one needs two key things in order to do that.

1 - a list of past qualifiers of the method
2 - a list of historic odds

#2 I have for many leagues and bookies.

#1 however I am not so sure off.

How can one get a list of historic 1x2 true odds predictions from oddswizard when the various settings are applied?

My ideal rough logic might be to

A - for individual leagues get from OW a list of historic matches and calculated true odds based on odds line settings chosen within the program.

B - I then compare in a spreadsheet these odds against my records of available odds and use the findings to try and create a "method" based upon OW true odds output.

eg first test might be ..profit from betting all in excess of true for league X
Further tests might include looking at value for home away draw etc.
Also possible to build in and test various factors to account for degree of acceptable value available above true etc.

Running OW manually every week for a year or two seems a very long and tedious way to go about building up such a historic record.

Anyhow thoughts from all appreciated.

Cheers
Mick

[pswirly]

You're right about it being tedious Mick.

I've done it for nearly six months.

Result aren't good either.

[AimHigher]

Mike,

Have you checked out the odds archive? http://www.newhavensoft.net/oddsarchive.htm. It doesn't provide all available leagues but you might find some that you want. It also doesn't seem to (at least on my computer) show the applied trend hints which might or might not be critical to you.

Another thing you can do which would be tedious but at least you wouldn't have to wait for a year or two is to go into a league, manually enter e.g. the last round of games as upcoming fixtures (left side below the headers that say N, Home team, Away team, Kickoff etc) and then in the history of completed games (top right side) you would delete the completed games for that round and then have OW calculate the odds without the data from the last round. That way you are getting the odds for an already completed round but you are only using data that was available at the time to calculate the odds. You can go back round by round as far as you have enough data for.

You can then choose to export the odds to a Excel, round by round. I know that is a lot of work but at least it is possible without waiting for upcoming games to be completed.

Regards,

Aim

[odds software]

Thanks for your feedback guys.

Seems getting data to make semi educated betting decissions upon is far from easy.

Odds archive

#1 does not contain any data other than that of default settings.
#2 is badly structured for such a purpose with spreadsheets on a week by week basis instead of grouping all data on an individual league into one long term spreadsheet.

Admin re organising that historic data into easier to use league by league data spreadsheets would make it much much easier to at least get some historic default settings data to work from.

Still of no use for any analysis outside standard settings.

A possible brighter future for users might be to include into the software programatic calculation of past odds
where unique settings are applied.
The percentage of people who would be prepared to go through it all manually I suspect might fall below 0.5%

Take Pswirly there who has been manully recording it for 6 months only to find his settings are not producing positive results.
Would it not have been better if he could of gotten several years of historical "true" odds data for several past years at the click of a mouse?
What is he to do now?
Spend a futher six months monitoring the effect of adjusting settings?

Do understand the software makes no claims as being able to do such stuff.
I feel myself however it could be significantly improved if a bit more emphasis were placed
on making it easier for clients to get historic oddsline data out of it.

Just my tuppence worth as someone who does not like betting blind but instead prefers
to determine if a long term historic edge is there..or not before having any interest in betting.

I am no software tech expert but with all historic match result data available
I would suspect such a thing to be doable if will was there to do it.

Picture someone asking..

Should I

A - bet totally blind having no idea about profit or loss from past history
or
B - examine past history to determine what edge if any is there before making any betting or bank management decissions

I think most including Alex himself would suggest #B the more shrewed route to go down.
He might even go further and suggest that having a decent data set of historic results to work from was an absolute core necessity of any educated betting strategy approach.

If he does believe that to be true then he might think it a good idea to build more into oddswizard in future to facilitate such thinking.
( easiest place to start perhaps isthe archive for default settings historic odds grouping data not by week but by league )

Cheers
Mick

[jazzfish]

Hi Mick,

The points you raise are spot and I have spent many hours attempting to answer your main question.

The first thing to say is that the archive odds can be used to back test results (I have done it with the help of another OW user) but the problem is that many of the fiar odds are in fact archived after matches have taken place (you will notice that in OW it's often possible to look at odds of a game once it's finished). Such odds are adjusted to take into effect the result so if you use them to back test you will find this factor unfairly improves the performance and profitability of OW.

I have tested around 8,000 selections (line odds > fair odds) and believe it's possible to gain an edge. Possible but very hard work and you will have to be patient. The first 5,000 selections performed well but there was then a 3,000 selection period which was not profitable. Who wants to make 3,000 selections for no profit?

Happy to discuss further if you require. Please PM me with your email address and questions.

Regards

JF