Stats testing

[WRC]

Hi,

I was going to do some stats testing, making sure things add up and are displayed correctly so I've been poking around the database. There's obviously the compiledplayerresults table with general data contained which can be used for working out most stats. But I can't work out how stats are calculated by position, can I receive a point in the right direction please. For example, how would I calculate VPIP from the small blind?

Thanks.

[udbrky]

I'm not sure what you're asking. Are you asking how to do it by calculating raw information from the postgres database?

[WRC]

Yup, I'm just trying to validate imports and stats.

[udbrky]

Your request has been passed on.

[udbrky]

The data is not stored in postgres files in 2.0.

We have done thorough stat testing and continue to do so. Is there a stat you feel is not correct?

In the future, data may be stored differently so as to allow more free access to it. Currently it is not.

[roy.goncalves]

The data is not stored in postgres files in 2.0.

We have done thorough stat testing and continue to do so. Is there a stat you feel is not correct?

In the future, data may be stored differently so as to allow more free access to it. Currently it is not.

Just to follow up on this, the best way to do any stat testing is just load the same sample in HM1 and HM2 and load the appropriate stats in a report.

As far as accessing the data, we have done things differently in HM2. We will be building an interface into this at some point and you'll see exactly how deep the data is but for now most of that is hidden except for the compiled stats which you have seen. Compiled stats are only used in the opponents tab for quick lookups

rvg

[WRC]

Thanks for that - after poking around I guessed this was the case. I noticed the CSV files in the Players folder and I'm guessing it's related to this. I'm guessing dumping to a file rather than writing to a DB is where most of the import speed improvement came from.

[roy.goncalves]

Yeah, the csv file format will be released at some point and part of the import speed improvement is due to this but really the main reason we went this route was due to performance on the stat analysis - we wanted incredibly deep statistical info with very high performance. Once we release some of the documentation for this the advantages will become more evident.

rvg