Creating a report to filter for play vs specific players

[Stally]

I used to have PT3 reports that looked at how people play against other groups of players based on the player icons. Now that feature doesn't exist, I was thinking I could just filter for a position being among a list of specific players using an OR function. After searching, I came across this:

( 0 < (SELECT count(hhps2.id_hand) FROM holdem_hand_player_statistics hhps2 WHERE hhps2.id_hand = holdem_hand_player_statistics.id_hand AND hhps2.id_player = (SELECT p.id_player FROM player p WHERE player_name='name')) )

The above piece of code seems to be PT3 only - I would also like to make an alteration to it to include multiple names (is this just a case of using player_name='' OR player_name=''... etc), and finally to list those players as being in a specific position.

[kraada]

Change it as follows for PT4:

( 0 < (SELECT count(hhps2.id_hand) FROM cash_hand_player_statistics hhps2 WHERE hhps2.id_hand = cash_hand_player_statistics.id_hand AND hhps2.id_player = (SELECT p.id_player FROM player p WHERE player_name='name' OR player_name='name2' [OR ....])) )

You keep adding as many OR options as you want where I have the [OR ....] - the []s aren't necessary of course.

[Stally]

how do i add a filter so that the particular players are in a certain position?

[kraada]

Change it to:

( 0 < (SELECT count(hhps2.id_hand) FROM cash_hand_player_statistics hhps2 WHERE hhps2.id_hand = cash_hand_player_statistics.id_hand AND hhps2.position = X AND hhps2.id_player = (SELECT p.id_player FROM player p WHERE player_name='name' OR player_name='name2' [OR ....])) )

Where X is the position number you want - 0 is button, 1 is CO, 2 is HJ and so on (though 8 is always BB and 9 is always SB).

[Stally]

Thanks! I appreciate the help

[Darvini]

I wasnt able to get this to work in PT3 so hoping very much that it can be done in PT4.

A filter for all those occasions where the hero acted against a player with certain HUD stat value at the moment of action. For example: filter all those occasions when hero made a pf3B against a player with "Fold to pf3B" -stat has a value of over 60% at the moment of the pf3B. Now if the "at the moment of action" is too difficult, then it can be dropped out. But would really like it to be there!

This filter would allow to test all the HUD stats for usefulness. For example if the Hero would not pf3B win (everyone folds to 3bet) with a significantly different percentage against opponents that have a high "fold to pf3b" stat, then against opponents that have low value, then there would be no point having that stat on the HUD. Maybe one needs only the "raise first in pf" or even just VPIP for some useful info on how likely a 3bet will get through.

[kraada]

Theoretically this can be done, but it would require a reasonably high level of SQL knowledge plus a desire to wait - reports like this can take a considerable amount of time to run. I'll put this specific report on my list to build for you, but realize that it's going to take a considerable amount of time to generate the data you're looking for.

[Darvini]

Thank you.

I hope you can appreciate how vital this filter is. Without it HUD is only a hypothesis without verification from tests. Its like a complex tool that has many features but we have little or no idea how the features affect the outcome. Even if it takes a long time for the report to run its still better then the alternative which is testing HUD stats manually by choosing certain hands which you only pf3b against an opponent that has folded over 75% to pf3b. You do this for a month or so and look from report what is your pf3b win rate for that hand.

[Darvini]

Would tipping make that filter appear faster?

[kraada]

I've built two reports and attached them here. Both show matrices of position vs position for 3betting. One shows all 3bets, another shows 3bets versus hero. Set Active Player to some opponent and change between the two reports to see the differences.

Make certain to check the sample size on the versus hero report.

I just pulled up the player that I had the most data on (besides myself) in my last 60,000 hands of data and the cell with the highest sample size was 7.