PFR doubts!

[kraada]

Yes those should work properly.

[andrenalina]

Thank you all guys for the attention, you helped me a lot.

Here is the columns and stats as I put in my PT:

Columns

Name: cnt_p_fold_pfr_first_action
Expression: sum( if[ lookup_actions_p.action = 'F' AND holdem_hand_player_statistics.flg_p_face_raise , 1 , 0 ])
Descripition: Number of times the player folded to a preflop raise.

Name: cnt_p_call_pfr_first_action
Expression: sum( if[ lookup_actions_p.action LIKE 'C%' AND NOT holdem_hand_player_statistics.flg_p_limp , 1 , 0 ])
Descripition: Number of times the player called a preflop raise.

Name: cnt_p_raise_pfr_first_action
Expression: sum( if[ lookup_actions_p.action LIKE 'R%' AND NOT holdem_hand_player_statistics.flg_p_first_raise , 1 , 0 ])
Descripition: Number of times the player raised a preflop raise.

Name: cnt_p_total_pfr_first_action
Expression: ( cnt_p_fold_pfr_first_action + cnt_p_call_pfr_first_action + cnt_p_raise_pfr_first_action )
Descripition: Number of times the player folded, called or raised a preflop raise.



Stats

Name: Preflop Fold to Raise First Action
Description: Percentage of times the player folded to a preflop raise when did not act yet.
Vaue Expression: (cnt_p_fold_pfr_first_action / cnt_p_total_pfr_first_action ) * 100

Name: Preflop Call Raise First Action
Description: Percentage of times the player called a preflop raise when did not act yet.
Vaue Expression: (cnt_p_call_pfr_first_action / cnt_p_total_pfr_first_action ) * 100

Name: Preflop Reraise First Action
Description: Percentage of times the player raised a preflop raise when did not act yet.
Vaue Expression: (cnt_p_raise_pfr_first_action / cnt_p_total_pfr_first_action ) * 100

[andrenalina]

Hello again! I was studying the stats and columns and decided to construct a stat that counts the times the player folded or raised a preflop single raise as first action in position and that wasn't a steal. So I constructed the follow columns:

cnt_p_sr_first_action_ip_def_action_fold: Number of times the player folds to a single raise as first action in position that wasn't a steal.
sum( if[ holdem_hand_player_statistics.flg_p_3bet_opp AND lookup_actions_p.action LIKE 'F' AND NOT(lookup_positions.flg_blinds AND holdem_hand_player_statistics.flg_steal_att), 1, 0])

cnt_p_sr_first_action_ip_def_action_raise: Number of times the player 3bets a single raise as first action in position that wasn't a steal.
sum( if[ holdem_hand_player_statistics.flg_p_3bet AND NOT( holdem_hand_player_statistics.flg_p_limp AND lookup_positions.flg_blinds AND holdem_hand_player_statistics.flg_steal_att), 1, 0])

cnt_p_sr_first_action_ip_def_opp: Number of times the player faces a single raise as first action in position that wasn't a steal.
sum( if[ holdem_hand_player_statistics.flg_p_3bet_opp AND NOT( holdem_hand_player_statistics.flg_p_limp AND lookup_positions.flg_blinds AND holdem_hand_player_statistics.flg_steal_att), 1, 0])

And then construct the follow stats

Fold to PFSR First Action In Position: Number of times the player folded to a preflop single raise as first action in position and it wasn't a steal.
( cnt_p_sr_first_action_ip_def_action_fold / cnt_p_sr_first_action_ip_def_opp ) * 100

Raise PFSR First Action In Position: Number of times the player folded to a preflop single raise as first action in position and it wasn't a steal.
( cnt_p_sr_first_action_ip_def_action_raise / cnt_p_sr_first_action_ip_def_opp ) * 100

Is that ok?

[andrenalina]

Other stat I want to construct is the "Limp and Fold to a PFSR OOP" (Percentage of the time a player limps then folds to a preflop single raise out of position). I want to diferenciate that situation from the other when the PFR is at the blinds, when the player limps and calls a PFR and he's in position on the flop. I made the columns:

cnt_limp_fold_oop: Number of times player limped then folded out of position.
sum( if[lookup_actions_p.action LIKE 'CF' AND NOT(holdem_hand_player_statistics.flg_p_4bet_opp), 1, 0])

cnt_limp_face_raise_oop: Number of times player limped and faced a raise out of position.
sum( if[holdem_hand_player_statistics.flg_p_limp and holdem_hand_player_statistics.flg_p_face_raise AND NOT(holdem_hand_player_statistics.flg_p_4bet_opp), 1, 0])

And then the stat

Limp and Fold to PFSR OOP: Percentage of the time a player limps then folds to a prelop single raise out of position.
( cnt_limp_fold_oop / cnt_limp_face_raise_oop ) * 100

But the problem here is that the stat will count raises from the blinds, when the player has position on the flop, can I put some restriction to who the raise come from? I want the stat to count only the raises from players that has position against the player on the flop. Is that possible? To construct stats in reference to the position after the flop is easier, I saw an exemple, but in this case.

[kraada]

Fixes below:

cnt_p_sr_first_action_ip_def_action_fold: Number of times the player folds to a single raise as first action in position.
sum( if[ holdem_hand_player_statistics.flg_p_3bet_opp AND lookup_actions_p.action LIKE 'F' AND NOT(lookup_positions.flg_blinds), 1, 0])

The only time you can have a 3bet opportunity and fold to it directly is if you are facing a raise for the first time. We cannot pick out the times you are on the button and face an open from the cutoff with 100% accuracy, so we cannot be certain that you weren't facing a steal. We can be certain you were in position though since you're not a blind. You can never have a chance to steal the blinds here, so that check was actually detrimental.

cnt_p_sr_first_action_ip_def_action_raise: Number of times the player 3bets a single raise as first action in position that wasn't a steal.
sum( if[ holdem_hand_player_statistics.flg_p_3bet AND NOT( holdem_hand_player_statistics.flg_p_limp OR lookup_positions.flg_blinds), 1, 0])

You want it to be neither the case that the player limped, nor that they were in the blinds, nor that they had a steal attempt. So we've changed that AND to OR. You cannot have 3bet and made a blind steal so that check was removed.

cnt_p_sr_first_action_ip_def_opp: Number of times the player faces a single raise as first action in position that wasn't a steal.
sum( if[ holdem_hand_player_statistics.flg_p_3bet_opp AND NOT( holdem_hand_player_statistics.flg_p_limp OR lookup_positions.flg_blinds), 1, 0])

Same comments as the last column. The stats will then work fine.


cnt_limp_fold_oop looks good. One minor comment about cnt_limp_face_raise_oop - if you limp and face raise and reraise and then get shoved on, that won't count as an opportunity here - when you face a 4bet, you also have a 4bet opportunity as all 5bets are considered "4bets" in PT3 (really it should be thought of as "4+ bet"). Depending on your game that may or may not be a problem, but I wanted to point it out.

You can also add a check that holdem_hand_player_statistics.position > holdem_hand_player_detail.val_p_raise_aggressor_pos to check to make sure that the last person who wasn't you that raised preflop was in position relative to you. val_p_raise_aggressor_pos holds the position number of the last aggressor preflop. Note that this means if the button 3bet and the SB 4bet val_p_raise_aggressor_pos would be 9. You'd prefer to do it this way because if there's no flop there's nothing to compare.

[andrenalina]

Actually I want to construct theese stats, but its a question of time



Any comments?

[kraada]

Any comments?

That's a lot of stats

If you take it one at a time over the course of a week or something, it won't feel that bad.

[andrenalina]

I put the check NOT( holdem_hand_player_statistics.flg_steal_att ) to not count the cases where the BU faces an open raise from CO, a steal Attempt (that case would be considered in the Steal HUD). If you dont check this it will be counted, right?

In the others I dont want to count the same case, where the CO open raise, independent of the hero action, this situation should not be counted.

I think its better to add a check that exclude only the cases where the CO open raises. Will it work? How cam I do this check? I dont know how to check the action of the other player, is it possible?

For the last 2 stats its better to not check holdem_hand_player_statistics.flg_p_4bet_opp and consider just Limp and Fold OOP, the intention was not to count the cases where the player limps, gets raised and somebody else 3bets, but i would have problems in that cases you said for exemple, but it will work properly for my needs. I will only put the check holdem_hand_player_statistics.position > holdem_hand_player_detail.val_p_raise_aggressor_pos, so it will be

cnt_limp_fold_oop
sum( if[lookup_actions_p.action LIKE 'CF' AND holdem_hand_player_statistics.position > holdem_hand_player_detail.val_p_raise_aggressor_pos, 1, 0])

cnt_limp_face_raise_oop
sum( if[holdem_hand_player_statistics.flg_p_limp and holdem_hand_player_statistics.flg_p_face_raise AND holdem_hand_player_statistics.position > holdem_hand_player_detail.val_p_raise_aggressor_pos, 1, 0])

Right?

And about your comment, its not a task that Im on a hurry! It will take the time it's necessary. While im doing this im learning a lot about stats and columns

[andrenalina]

I dont now if you understand my needs.

My intenton with theese stats is to help the cases where theres a limper (or more) and we want to raise, its necessary to look at limper's limp and fold OOP and Fold to F Cbet when Limped OOP (I need to change this too), the fold to PFSR and 3bet of the players behind and the stats about Steal Attempt for the blinds. The same for the case where hero 3bets, they're complementary.

In your opnion, its that fine? I think if we have the specific stat of a move separeted from the others its better to avaliate the situation, right?

[kraada]

flg_steal_att is true if and only if the player in question made a steal raise themselves. flg_blind_def_opp is for when a player faces a steal while in a blind, but there is no database flag that you can check to eliminate (or pick out) times you are on the button and face an open from the CO - there just isn't the data for that sort of thing.

With regards to the limp_reraise columns you also want to add a check that a player does not face a 4bet unless you want this case:

UTG limps, BTN raises, BB 3bets

to count (but if you're happy with that, then go for it).

The idea behind these stats is fine, though in practice I suspect it's a bit much for analysis at the table. Personally I just show a player's limp/fold percentage (position independent) in 10% chunks as a color (so 0 - 10% is red, 10-20% is light red, etc.) on another stat. Though it's entirely possible I'm missing a goldmine of useful data here