PokerTracker

I have an observation of Preflop Positional Awareness-stat. I'm getting weird 0.00 values that I don't think are proper. This is all from a datamined hands database that has only 6-max PLO hands.

Just as an example, I have a player with 17305 hands. His RFI from different positions are:
OTB-3 -> 0.55 (9/1638)
OTB-2 -> 1.40 (30/2143)
OTB-1 -> 4.09 (65/1592)
OTB -> 10.42 (108/1036)

So AFAIU, this means that he is very positionally aware with his raising, even to extremes.

However, Preflop Positional Awareness shows 0.00 for him! I thought that this might be some calculation overflow or somesuch thing so I created two new stats to see what values go into expression for Prelop Positional Awareness:

PFPosAw1 = (cnt_p_rfi_btn / cnt_p_open_opp_btn)
PFPosAW2 = (cnt_p_rfi_ep / cnt_p_open_opp_ep)

Nothing fancy there, just the two terms from the Preflop Positional Awareness separated to check out their values. For this player the values are (I changed the decimal count to 3 on those stats):

PFPosAW1 = 0.190
PFPosAW2 = 0.005

So PFPosAW1 / PFPosAW2 = 0.190 / 0.005 = 38

So, for this player I would expect Preflop Positional Awareness to be 38.00. Instead it is 0.00.

Any idea why is that ? My hunch is that the source for this problem could be your code that processes the formulas for the statistics. Maybe it misbehaves with very small numbers as the denumenator. Maybe it interprets 0.005 as 0.00 and thus outputs just 0.00.

Check this out for yourselves, just add Preflop Positional Awareness in to some player report and look for 0.00 values for players with decent number of hands and anlyze the calculations for them. Maybe utilize the PFPosAW1 and 2 to get a view on the separate terms.

I don't see any open bugs in our system for this particular issue, but I agree that this sounds like something strange might be going on here.

Please create a support ticket containing a link to this thread and a couple of screenshots of a player showing the RFI by position and his preflop positional awareness and we'll definitely make sure this gets looked into.

Done: https://www.pokertracker.com/contact/ti ... _id=128936

Much obliged!

Did this ever get solved?
The current way of calculating this is not optimal. If you play 5-max tables, the stat will always be 0.

I feel like I've asked this before... looks like the column wasn't thought out thoroughly. If you look at the bolded part... how can someone be 6 off the button in a 4-6 handed game? Same for 7-8 handed.

cnt_p_rfi_ep:
sum(if[cash_hand_player_statistics.flg_p_first_raise AND ((cash_hand_summary.cnt_players between 4 and 6 and cash_hand_player_statistics.position between 3 and 5) OR (cash_hand_summary.cnt_players between 7 and 8 and cash_hand_player_statistics.position between 4 and 7) OR (cash_hand_summary.cnt_players between 9 and 10 and cash_hand_player_statistics.position between 5 and 7)), 1, 0])


Also: doesn't flg_p_first_raise also count raises after limpers, not just open raises? But it gets divided by open_opp. So someone who isolates a lot will get a higher value... which may be true actually, but not sure if that was intented, and might give wrong impression on actual preflop opening ranges.

It's true isolations would count but if you isolate a ton in EP at full ring something strange is going on - remember you still need to be in EP.

In our current definitions for short handed, "EP" is only UTG 6-handed. At 5-max there is no EP, only MP, CO, BTN, SB, BB - so there is nothing to compare. If we add in MP everywhere then that will dilute the stat considerably for all non-5max situations.

If you're playing 5 max or shorter, you're probably going to be wanting to just look at the player's positional stats - there aren't that many of them to compare and looking at the RFI stats for the player for the 4 valid positions should give you a pretty good idea if he's adjusting or not.

kraada wrote:It's true isolations would count but if you isolate a ton in EP at full ring something strange is going on - remember you still need to be in EP.

It's the same for button.
This is a pretty useful stat to give a quick idea on how the ranges change... obviously I have per position percentages in a popup. I'll revert to my old posaw stat that does some weights for the positions also, and takes care of the 5-max problem. I've been avoiding it since it needs custom columns... hopefully the rfi-columns per position is something you consider to be cached.

Here's the stat if anyone should be interested, the extra columns are straightforward to build

Code: Select all

((cnt_p_rfi_co / cnt_p_open_opp_co)+ 2* (cnt_p_rfi_btn / cnt_p_open_opp_btn))
/
((cnt_p_rfi_hj / cnt_p_open_opp_hj)+ 2*
(if((cnt_p_open_opp_ep > 0),(cnt_p_rfi_ep / cnt_p_open_opp_ep),(cnt_p_rfi_hj / cnt_p_open_opp_hj)
)))