Importng a custom Leauge with expected Goals as results

[papadas]

Hello Alex,

what would hapen if i import a custom league to the Application but instead of importing the results i would import the expected goals after each match.

I have a Dataset where i have the expected Goals of each after the end of the match. Is there a problem for the Application to parse float numbers instead of integers?

BR
Alex

[Alex]

Existing version is not able to parse floats as results of matches. However it is possible to implement such a feature in some custom release.
Another thing, what is the sense of such manipulation? All real past games have integer values of results. This data is used by OW to compute odds for future events. Intermediary floating values for home and away estimated goals are also calculated by the program. What would we compute if we take some external data with floating values? By default, the program does not process future "results" or estimations, neither float nor integer. Also, Tournament analysis tool computes various scenarios of future games and standings using just integer possible results values.

[papadas]

Hello Alex,

thanks for your response. The Idea behind my suggestio is that if we feed ODDs wizard with a expected goals this would perform better in terms of predictions, since expected goal describe a lot better one teams ability in terms of attack and defence and would like to experiment. Will be cool if some time in the future will be able to take such input.

Have 2 off topic questions:
1) The ODDs archive contains the ODDs that are calulated by OW PRE-game or POST-game. I guess these odds are calculated using a default depreciation period for each league . Is this statement correct?
2)The Perormance analysis Results (e.g http://www.newhavensoft.net/stats/Italy.htm) are calculated using the 2015 algorithm and the default Depreciation period for the leauge? Is this correct?

Thanks in advance

Alex

[Alex]

Ok, I agree that it would be better to use floating values as results of past games if they really exist. But they don't! Suggested expected goals are also predictions. How can we use one predictions to improve another predictions? I guess it is impossible.

Now you can try the following trick: multiply all expected goals by 10 and then truncate to integer. All real past results must also be multiplied by 10 so that you can compare two approaches. Use hockey algorithm. I suppose that only past performance could be slightly improved, but not future predictions.

1. Archive odds are usually calculated at Saturday morning. So all coming weekend odds and all further odds are PRE-MATCH. All past odds appear to be POST-MATCH.
2. Yes, that is correct.