Win % Statistic Includes Sidepot Wins?

[TrustySam]

I was filtering out some results based on tournament type recently, and just realized that PT3 is counting sidepot wins as wins for the 'win %' statistic.

Below is a screencap and hand history :

[ img ]

PokerStars Hand #74171368553: Tournament #499550011, Freeroll Hold'em No Limit - Level IV (50/100) - 2012/01/20 10:27:54 ET
Table '499550011 885' 9-max Seat #5 is the button
Seat 1: Nugget of Oz (1845 in chips)
Seat 2: lefreak 01 (2400 in chips)
Seat 3: bolondoss (2400 in chips)
Seat 4: gsxr750k8 (2150 in chips)
Seat 5: labschie (2250 in chips)
Seat 6: Frenk1012 (5335 in chips)
Seat 7: TrustySam (6410 in chips)
Seat 8: SAYM0N (8660 in chips)
Seat 9: Shark JoKKeR (710 in chips) is sitting out
Frenk1012: posts small blind 50
TrustySam: posts big blind 100
*** HOLE CARDS ***
Dealt to TrustySam [As Ad]
SAYM0N: folds
Shark JoKKeR: folds
Nugget of Oz: raises 1745 to 1845 and is all-in
lefreak 01: calls 1845
bolondoss: folds
gsxr750k8: folds
labschie: folds
Frenk1012: folds
TrustySam: raises 1745 to 3590
lefreak 01: calls 555 and is all-in
Uncalled bet (1190) returned to TrustySam
*** FLOP *** [7d 8s 5c]
*** TURN *** [7d 8s 5c] [Qc]
*** RIVER *** [7d 8s 5c Qc] [Ks]
*** SHOW DOWN ***
TrustySam: shows [As Ad] (a pair of Aces)
lefreak 01: shows [3s 3h] (a pair of Threes)
TrustySam collected 1110 from side pot
Nugget of Oz: shows [Kc 7c] (two pair, Kings and Sevens)
Nugget of Oz collected 5585 from main pot
lefreak 01 re-buys and receives 2400 chips for $0.60
*** SUMMARY ***
Total pot 6695 Main pot 5585. Side pot 1110. | Rake 0
Board [7d 8s 5c Qc Ks]
Seat 1: Nugget of Oz showed [Kc 7c] and won (5585) with two pair, Kings and Sevens
Seat 2: lefreak 01 showed [3s 3h] and lost with a pair of Threes
Seat 3: bolondoss folded before Flop (didn't bet)
Seat 4: gsxr750k8 folded before Flop (didn't bet)
Seat 5: labschie (button) folded before Flop (didn't bet)
Seat 6: Frenk1012 (small blind) folded before Flop
Seat 7: TrustySam (big blind) showed [As Ad] and won (1110) with a pair of Aces
Seat 8: SAYM0N folded before Flop (didn't bet)
Seat 9: Shark JoKKeR folded before Flop (didn't bet)

Is it accurate/ideal to be automatically be including sidepot wins as part of the win%? It's also technically a loss as well as a win. In the event that there's both a win AND a loss, wouldn't it be more accurate to let the net chip intake/outake determine the classification of the hand as either a win or a loss?

Sam

[kraada]

There are different opinions on this - in PT3 we count it as a win if you receive part or all of any pot - as you at least won something (you didn't lose all of your wagered money). There are other cases as well where you can have won the hand but lost money - if you split the pot for example, you lose a small amount to the rake.

You could build a custom statistic to display the percentage of the hands you actually gained money on if you would like.

Create the column:

amt_won_money_hand: sum(if[holdem_hand_player_statistics.amt_won > 0, 1, 0])

Then duplicate the Win % stat, change the name/descrption as you'd like, replace cnt_won_hand with amt_won_money_hand in the value expression, change the format expression as you'd like. Apply the changes, update your cache and you'll be all set to use the stat now the way you'd like to see it.

[TrustySam]

There are different opinions on this - in PT3 we count it as a win if you receive part or all of any pot - as you at least won something (you didn't lose all of your wagered money). There are other cases as well where you can have won the hand but lost money - if you split the pot for example, you lose a small amount to the rake.

You could build a custom statistic to display the percentage of the hands you actually gained money on if you would like.

Create the column:

amt_won_money_hand: sum(if[holdem_hand_player_statistics.amt_won > 0, 1, 0])

Then duplicate the Win % stat, change the name/descrption as you'd like, replace cnt_won_hand with amt_won_money_hand in the value expression, change the format expression as you'd like. Apply the changes, update your cache and you'll be all set to use the stat now the way you'd like to see it.

Thanks for the prompt response Kraada - I saw it yesterday, but needed time to process the info you gave me.

I can sure understand why multi-player pots with a main pot and side pot have generated different ideas on how to classify them ... that's what I was re-thinking yesterday.

The thing about sidepots being treated as wins all the time, is that they're never offset by wins that are also losses that are counted as losses if you don't play cash games (the split pot scenario). Which would then necessarily overestimate win rates over the long-term.

So the alternative I listed above I guess was an attempt to try and address that overestimation - by allocating the win based on whether the net amount of chips made from the hand was positive. But even that's imprecise in that it doesn't tell the full story of what's happening with each individual hand (of there being both a win and a loss).

I decided that my ideal formula would allocate a percentage of win/loss for each hand in which there's a main pot lost, and a side pot won, based on the amount of chips put into the pot, in proportion to the amount that went into each pot. Like I guess if I were to write a program, the logic of it would go something like this (it's not code-correct, since I'm not that familiar with PT3's commands):

if [mainpot and sidepot]
and if [mainpot=lose] and [sidepot=win]
then [amt_gambled_sidepot/amt_gambled_total+count_total_wins_for_hand]
and [count_number_of_times_played_for_hand=count_number_of_times_played_for_hand+1]

So for my hand above, the variables would have the following amounts:
amt_gambled_sidepot=555
amt_gambled_total=2400
count_total_wins_for_hand=1 (number of times I've won with AA so far)
count_number_of_times_played_for_hand=2 (number of times I've played AA so far, not including this hand)

That would give:
- a 555/2400 fractional win (23% win, 78% loss) for this, the 3rd time I'd played AA
- and an overall win-rate for AA of 1.23 out of 3
- making my win % according to this formula 1.23/3 = 41% (in contrast to the 66% that PT3 displays, and in contrast to the 33% that a program allocating wins/losses based on net gain/loss would give)

Would it be possible to write a formula for PT3 that could do this? Specifically, is there a way for PT3 to get the amounts put into the pot to use in such a formula? I think I'd be really happy if I could get the win% to show like that. Thanks so much for your help - hope it was easy to understand what I was trying to convey :)

[kraada]

I understand what you're going for unfortunately the main pot vs side pot sizes are not stored separately in the database so this formula won't be possible.

[TrustySam]

Yeah, it looked like that might be the case. Figured I'd ask, just in case.

Thanks for writing me the program as an alternative - I might add that to my PT just to see how much (or how little) difference there is between it and the formula that comes with PT ... probably not much over the long-term, eh?

Thanks again for your help!

Sam

[kraada]

It depends exactly how the game you play works out - if side pots are more common it'll be different than if side pots are very rare.

If you do put it together I'd be curious to know what your results are if you don't mind sharing.

[TrustySam]

Okay ... I was going to ask you if you wanted to see the results, but I didn't want to put you on the spot in case you weren't interested haha

Here's the stats for the tourneys I was filtering out - the difference is massive ... with that sample size of 4 hands of AA :D

[ img ]

And here's the stats for all the hands I've played since I got a new computer in April or so. I didn't have a chance to look at the data in a lot of detail, but for some of the other hands like AQ, I wasn't able to find any sidepots - it looks like the difference might be because of split pots (with no chip gain or loss?). And then, what's funny is that with AA I found three sidepot only wins (and no split pots), but the statistic is showing to two decimals as being the same (didn't have a chance to check the math ... )

[ img ]

Will have to take another look at the data tomorrow ... have to go pick my parents up at the airport right now. Kind of interesting to muck around with PT in an case ... what a powerful tool! I was seeing all kind of things I never knew it could do, which was really neat to see :)

[kraada]

If you think that's great just wait - PT4 will be out soonishly and it's got a ton more power than PT3 does :)

[TrustySam]

I was just looking at one of the videos for PT4 ... looking forward to checking it out!! :)

PS Couldn't edit my last post - put the wrong screen cap in (that had a filter on). Here's the one that I meant to post - there's a little difference with AA, but it's not so much ... 3 sidepots ...

[ img ]

Sam

[kraada]

That's great to know - I didn't expect them to be common but real data is always more useful than speculation.