Multi street statistics

[friidayy]

Hello, I want to extend my HUD with non-standard (e.g. River Cbet) multi street statistics, for example Call Flop Raise and Fold Turn. Are there any good available? At the moment I mainly use WSDWRF (Won Showdown when Raise Flop) as a guideline when I am the defender. Should I create my own stat for Call Flop Raise and Fold Turn? Is there any reference I can use?

[Flag_Hippo]

Take a look at 'Check Fold to T Bet After CBet & Call Flop' in the Download Warehouse.

[friidayy]

cnt_t_face_bet_after_cb_call_flop_oop reads

Code: Select all

sum(if[cash_hand_player_statistics.flg_f_cbet and lookup_actions_f.action = 'BC' and lookup_actions_t.action LIKE 'X_%', 1, 0])

What does 'X_%' exactly mean? I guess, X for check and _% for some arbitrary action?

Let's say I want to adjust their code for IP.

cnt_t_f_after_cb_call_flop_ip

Code: Select all

sum(if[cash_hand_player_statistics.flg_f_cbet and lookup_actions_f.action = 'BC' and lookup_actions_t.action ='F', 1, 0])

cnt_t_face_bet_after_cb_call_flop_ip

Code: Select all

sum(if[cash_hand_player_statistics.flg_f_cbet and lookup_actions_f.action = 'BC' and lookup_actions_t.action LIKE 'F', 1, 0])

Is this right? What does "LIKE" stand for?

[Flag_Hippo]

What does 'X_%' exactly mean? I guess, X for check and _% for some arbitrary action?

The _ is for any second action and the % covers any possible actions after that so 'X_%' means that that player checked and then took at least one other action.

Let's say I want to adjust their code for IP.

cnt_t_f_after_cb_call_flop_ip

Code: Select all

sum(if[cash_hand_player_statistics.flg_f_cbet and lookup_actions_f.action = 'BC' and lookup_actions_t.action ='F', 1, 0])

cnt_t_face_bet_after_cb_call_flop_ip

Code: Select all

sum(if[cash_hand_player_statistics.flg_f_cbet and lookup_actions_f.action = 'BC' and lookup_actions_t.action LIKE 'F', 1, 0])

Is this right?

These columns are counting the exact same thing so the opportunities column has the player only folding the turn which you don't want. Either test for different action(s) or just use cash_hand_player_statistics.amt_t_bet_facing > 0.

What does "LIKE" stand for?

LIKE is used for pattern matching when you need to test for more than one possible sequence - see this guide.

[friidayy]

Just to make sure that I understand everything correct, I wanted to validate two different formulations that should result in the same statistic.

First one (the original one):

Check/Fold to T Bet After CBet & Call Flop

Code: Select all

(cnt_t_xf_after_cb_call_flop_oop / cnt_t_face_bet_after_cb_call_flop_oop) * 100

with

cnt_t_xf_after_cb_call_flop_oop

Code: Select all

sum(if[cash_hand_player_statistics.flg_f_cbet and lookup_actions_f.action = 'BC' and lookup_actions_t.action ='XF', 1, 0])

and

cnt_t_face_bet_after_cb_call_flop_oop

Code: Select all

sum(if[cash_hand_player_statistics.flg_f_cbet and lookup_actions_f.action = 'BC' and lookup_actions_t.action LIKE 'X_%', 1, 0])

Second one:

Same as first one with one difference

cnt_t_face_bet_after_cb_call_flop_oop

Code: Select all

sum(if[cash_hand_player_statistics.flg_f_cbet and cash_hand_player_statistics.amt_t_bet_facing > 0, 1, 0])

I checked both formulations and the stats significantly differ. Why is that?

[Flag_Hippo]

There isn't anything in the last one that says the player is out of position. If you are not going to use the players actions to determine that then you can use:

Code: Select all

NOT cash_hand_player_statistics.flg_x_has_position

where x is the street.

[friidayy]

Ok, I adjusted cnt_t_face_bet_after_cb_call_flop_oop to

Code: Select all

sum(if[cash_hand_player_statistics.flg_f_cbet and cash_hand_player_statistics.amt_t_bet_facing > 0 and NOT cash_hand_player_statistics.flg_t_has_position, 1, 0])

Stats are now almost identical.
Original: 34% (34/99)
New one: 33% (34/102)

[Flag_Hippo]

The difference will be in hands where there are more than two players postflop which complicates things - you can filter for this and look at the resulting hands. For example:

Player A checks flop
Player B cbets flop
Player C calls
Player A raises
Player B calls
Player C calls
Player A bets turn
Player B?

Here player B is is out of position (not last to act) but is in position relative to the bettor. One of your columns counts this but the other does not. Things would also change if Player C bets the turn (who wasn't the flop raiser) and that could get counted in both of your columns. You can specify that there were only two players postflop or if you want to include multiway hands then that will require some tests depending on exactly what situations you want to count and which ones you want to exclude. Do you only want hands where the player has absolute position (last to act) or do you want to test for the position of the player relative to the player who raised and/or bet? Does the player who made the raise flop raise need to be the same player who bets the turn?