Online Poker Forum at Full Tilt Poker

nilgiri

Holdem Manager has this great feature that lets me see what my EV is for a given hand. For example, here is a bad beat I had today:

Full Tilt Poker, $0.50/$1 NL Hold'em Cash Game, 6 Players
LeggoPoker.com - Hand History Converter

UTG: $48.30
MP: $119.85
CO: $122.80
BTN: $229.10
Hero (SB): $124.15
BB: $178.75

Pre-Flop: K

A

dealt to Hero (SB)
2 folds, CO raises to $3.50, BTN folds, Hero raises to $13, BB folds, CO calls $9.50

Flop: ($27) 2

A

7

(2 Players)
Hero bets $18, CO raises to $109.80 and is All-In, Hero calls $91.80

Turn: ($246.60) 2

(2 Players - 1 is All-In)

River: ($246.60) 6

(2 Players - 1 is All-In)

Results: $246.60 Pot ($3 Rake)
CO showed Q

J

(a flush, Ace high) and WON $243.60 (+$120.80 NET)
Hero showed K

A

(two pair, Aces and Twos) and LOST (-$122.80 NET)

I lost $122.80, but my EV from the hand was a gain of $29.02. So my EV difference is +$151.82 (i.e. I "should" have won $151.82 more than I did).

Over a bit more than 10k hands, my EV difference over all hands were I see a showdown (i.e. the ones that we can apply this to) is +$800.30 (this is only over 412 hands where I have seen a showdown). This is an extremely large amount of money over 10k hands. Today (after a horrible session full of bad beats) I am winning 5.37 bbs/100. Had I won the amount I "should" have, I would be at 13.26 bbs/100. That's a very large difference -- it's saying the amount I "should" have won is over 2.5x the amount of money I actually did win.

I have assumed that this difference will eventually smooth out to (statistically) $0. That is, I will, over a large enough sample, have a negligible difference betwen how much money I won from showdowns, and how much I "should" have won.

What I'm wondering is whether this is correct, mathematically. I'm wondering this in particular because over 10k hands, my EV difference keeps increasing. Is there some mathematical reason for this, or is it just coincidence, and I am correct to assume that it will eventually smooth out?

I also am curious whether there is any known bug in HM regarding the EV stat. This may be something I need to ask on the HM boards, though.

Thanks for any thoughts.

Zophar · Posted: Tue Jun 17, 2008 3:08 pm Post subject:

This is basically the concept of Sklansky dollars. The actual and expected should even out over the long term. It's cool they offer the feature.

In game, I don't know how usefull the info is, except for comfort maybe.

I'm just starting to reread the OMGitsGandolf bluff article-Gbucks, where Phil takes the concept to a new level. It's some pretty great stuff. Instead of +EV calculation vs. the specific hand, it deals more in ranges. Here's the link if you haven't read it yet:
Phil"OMGclayaiken"Galfond
http://www.bluffmagazine.com/magazine/%27G%2DBucks%27%2DConceptualizing%2DMoney%2DMatters%2E%2DPhil%2DGalfond-985.htm

nilgiri · Posted: Tue Jun 17, 2008 3:27 pm Post subject:

Cool, thanks for the link. That's interesting stuff!

On a side note, if anyone is interested in some Sklansky dollars let me know. I'm selling 'em real cheap!

live4freerolls · Posted: Tue Jun 17, 2008 3:39 pm Post subject:

Nice article, now I need to start saving up my sklanksy bucks for a trip to equitytown imo.

Riddim · Moderator Joined: 04 Dec 2005 Posts: 8172 Location: Sweden

I don't know whether there's a bug in the HM software or not, but being down 8 buy-ins in all-in EV after 10k hands doesn't constitute an extremely large amount of money. It just means you've lost 4-5 all-ins or so more than average. I mean in this specific situation you're going to lose about 38% of the time, so it's not at all out of the ordinary and you still end up $150ish below EV when you lose just because you're a bit deep. Now throw in some lost flips and a couple of 80/20s and you're up to $800 already.

nilgiri · Posted: Tue Jun 17, 2008 3:55 pm Post subject:

I had another terrible session this afternoon. All told I'm down about 8 buyins over 1300 hands/3.5 hours today. I seriously hope the math is correct and I'll be seeing that almost $1k in lost equity eventually... Does anyone have higher winnings than EV? Way in the beginning you can see that my winnings were higher at one point. But they seem to steadily diverge in the wrong direction.

HuJwang · Posted: Tue Jun 17, 2008 8:28 pm Post subject:

Over my last 44,000 hands, I'm 27 buyins below all-in equity. So your downswing really isn't much to complain about Laughing

I don't think HM has an EV calculation bug, but you can check it by enabling the "EV difference" and "equity%" columns in the hand view, and calculating it yourself.

Peepers_au

Riddim · Moderator Joined: 04 Dec 2005 Posts: 8172 Location: Sweden

Peppers, you can still hold up more often than average and win more flips than average.

telebob · Posted: Wed Jun 18, 2008 3:53 am Post subject:

Peepers, I think you're misunderstanding the graph. We're looking at each hand's EV versus the actual result. This includes the times we get in bad and suckout as well as the times we get in good and lose, the times we win flips, the times we lose flips. And of course it counts the times the best hand holds up too.

One would hope that we get our chips in good more often than not, but that has no effect on the graph.

Oh yeah, does anyone know how to make custom graphs on PT3?

EDIT: Yay, this is my three-of-a-kind post!

Peepers_au · Posted: Wed Jun 18, 2008 4:38 am Post subject:

I understand what you're both saying and what the graph represents.

My argument was that the number of times you'd get your money in as the favourite and lose would probably outweigh the times you'd get your money in as the underdog and suckout. Not because of variance, but because of the skill of the player and recognising a bad situation and folding.

While that wouldn't apply so much to moderate/weak hands vs. other moderate/weak hands that go to showdown, it certainly should in a strong hand vs. strong hand situation.

Over infinite hands, would you expect your hands won at showdown to be above, below or exactly 50%? If it's not exactly 50% then I don't see how you can think your expected earnings and EV earnings should be the same.

Riddim · Moderator Joined: 04 Dec 2005 Posts: 8172 Location: Sweden

Peppers, you're misunderstanding either the graph or the math behind it. The graph represents how close to our expectation we are. That isn't going to change depending on whether we get it in good or bad most of the time. It's not easier to run above expectation by getting it in bad. If it was, we should be getting it in bad more often.

Peepers_au · Posted: Wed Jun 18, 2008 9:26 am Post subject:

nilgiri · Posted: Wed Jun 18, 2008 11:09 am Post subject:

Thanks guys. Hu, this post was spurred by a downswing, but I'm not complaining (honestly). It just made me rethink whether my previous assumptions were correct.

	Full Tilt Poker is licensed by the Kahnawake Gaming Commission. © Copyright 2004 - 2005 FullTiltPoker.com., All Rights Reserved