analyzing hh files with scripts

[tinhat]

Anyone know of any perl scripts (anything really) anyone's written for analyzing hh for stuff pt doesn't offer? For some reason when I search 'perl -"Re:"' I get lots of hits but "perl" isn't in any of the threads.

There's a lot of scenarios I'm interested in taking a look at but offhand seems like a script parsing hh could get pretty complicated so thought I'd check before wasting a lot of time trying (and I don't know anything about querying access/pgsql dbs)...

Mike

[jba]

well if you could get ahold of the source for bison's hand convertor which I heard is perl it would give you everything you need. I have no idea how you'd go about that.

you should seriously consider figuring out how to pull the info from postgres it will end up being much easier and faster

[MrMoo]

I thought about this but decided it wasn't worth it. For meaningful statistics you'll need to parse an arseload of hand histories. In the end I decided that I'd just wait until I moved Poker Tracker over to PostgreSQL and write some perl which would query the Poker Tracker databases directly.

[MrMoo]

[ QUOTE ]
I don't know anything about querying access/pgsql dbs)...

[/ QUOTE ]

http://cpan.uwinnipeg.ca/search?quer...&mode=dist

[_dave_]

[ QUOTE ]

There's a lot of scenarios I'm interested in taking a look at but offhand seems like a script parsing hh could get pretty complicated so thought I'd check before wasting a lot of time trying (and I don't know anything about querying access/pgsql dbs)...


[/ QUOTE ]

If you are capable of coding perl, you will likely find SQL very easy to get in to.

What kind of set up are you running? PokerTracker on Windows with an Access Database? or something altogether more tricky like Party & PokerTracker running in WINE using Linux native Postgres? I seem to remember you posting some Linux Poker stuff not too long ago.

If you like, I can easy get you started with some simple SQL scripts you can use to query your PokerTracker data. Give me an example of something you'd like to know, I'll try to knock up some SQL to find it.

Take a look at my post history, there are some SQL rakeback calculators, raked hand % counters filtered by Number of players + level, and some other scripts I have posted in the past.

You can always start a thread here with "Does anyone know the SQL to do this...", I, and others better than me will likely give you a good answer.

Also, PokerTracker's SQL forum is good.

My SQL skills are not great, my code is probably nowhere near optimal, but for offline analysis this isn't so important. It is you want to make an app such as PokerAce, where the SQL efficiency is critical.

Let me know what you want to know,

Enjoy!

dave.

[RatFink]

Using the postgres db (or even the Access DB) for SQL queries from PokerTracker is useful for a lot of analysis, but you will not be able to reconstruct a hand action by action.

If you want to parse hand histories you can take a look at BisonBisons code for his handconverter (written in PERL) which he released into OpenSource at: http://sourceforge.net/projects/handconverter

It is very specific to its function, and will take some tinkering to make it do whatever it is you want to do. But as long as that tinkering and eventual use complies with the GNU license he released it under, it's a good place to start.

[tinhat]

Thanks all - I did look at bison's converter which is when I decided I probaby didn't really want to try this kind of thing [img]/images/graemlins/smile.gif[/img]


Dave -

I'm a bit of a shell script jockey and dabble in perl some; know absolutely nothing about SQL except what it stands for so thought I'd try scripts first (I run xp in vmware with pgsql serving from linux). This is all for offline analysis.

I have lots of relationships I want to look at floating in my head and the only few I've written down are:

1. win %/board texture for 3-bet turns (this sounds very hard; maybe too nebulous);

2. ratio of turns 3-bet vs called and final hands for either

3. ratio of premiums dealt in blinds vs. other positions (6m; I'd define the premiums);

4. ratio of premiums vs non-premiums dealt in total (also would be nice to see the distribution w/respect to time);

5. under the second group of "player actions" in pt's "more detail" page, I'd like to be able to match any of the columns with the hands those values are derived from (that is, getting the game #s for say "bet/raised/c-r: WSD%");

6. some way (maybe not possible) to look at hands folded on flop that improved with turn card (to see if my definition of a "loose" flop call isn't loose enough).


I think pt has an untapped goldmine of relationships to look at and could try to cobble some of these together from just what pt allows directly. But that sounds awful time-consuming, intricate and having to keep track of lots of unrelated info.

It also means doing it in windows and no offense to anyone but windows' text processing capabilities are poor/non-existent or just plain retarded. So if I want to use linux I either have to use the hh or I guess bite the bullet and start trying to learn how to query the dbs...

Mike

[tinhat]

[ QUOTE ]
[ QUOTE ]
I don't know anything about querying access/pgsql dbs)...

[/ QUOTE ]

http://cpan.uwinnipeg.ca/search?quer...&mode=dist

[/ QUOTE ]

Forgot to thank you for this tidbit; didn't even think to see what kind of db support perl already has available. Thanks...

Mike

[SheridanCat]

You can take a look at hand history parsing code I wrote here:

http://www.pokergeek.com/software/

Using bison's code will probably not be where you want to go. Accessing Postgres or Access is pretty simple, really. Just look at the Perl DBI docs.

Regards,

T