Coding the Wheel

Optimizing VP$IP

Saturday, August 27, 2011

In response to the VP$IP riddle we posed on Twitter, the PokerTracker team hooked us up with an explanation of the upcoming changes to the VP$IP stat, along with some superfluous superlative scatter graph awesomeness.

But first, the answer to said riddle:

Raise!

That is, in order to maintain a VP$IP of 100%, you must RAISE every time you're in the big blind and the table limps around to you. The reason, in a nutshell, is that the big blind, being a forced bet, isn't considered "voluntary".[1] But be aware that this behavior is changing in PokerTracker 4.

Upcoming Changes to VP$IP

The canonical formula for VP$IP is about as old as the Giza Necropolis, give or take a few years. Not marble nor the guilded monuments...

(cnt_vpip / cnt_hands) * 100

But at least according to Twitter, the longstanding formula for VP$IP will be changing in PokerTracker 4.

To be sure, this is a subtle change. The amended formula for PT4 looks like this: 

(cnt_vpip / (cnt_hands - cnt_walks)) * 100

The only difference is that hands in which you get a walk in the big blind are ignored for the purposes of calculating VP$IP. This could make VP$IP more intuitive, as most VP$IP-related confusion centers around the blinds (which is why we chose an edge-case situation in the blinds for our VP$IP riddle).

VP$IP: Old vs. New

Assuming the VP$IP enhancement shows up in the final PT4 release, the million-dollar question is how will the change affect VP$IP values in the games you play? Will this rule induce any wrinkles to the VP$IP statistic such that what used to be a 24% VP$IP is now a 32% VP$IP, requiring some sort of adjustment on your part?

As it turns out the answer is not much, and probably not.

The PokerTracker team provided links to three graphs showing a pretty tight correlation between VP$IP in PokerTracker 3 versus PokerTracker 4. The graphs were generated from a pool of 800k hands and 69k players with qualification requirements of 100 hands, 250 hands, and 500 hands respectively.

The first graph shows before-and-after VP$IP for players who had more than 100+ recorded hands (a fairly lenient sample size, then again VP$IP converges relatively quickly):

VP$IP in PT3 vs. PT4 (100+ Hands). Scatter graph showing VP$IP stability between PokerTracker 3 and PokerTracker 4. Vertical axis is VP$IP in PokerTracker 4. Horizontal axis is VP$IP in PokerTracker 4. Sample of 68k players, 800k hands.

The second graph shows the same data restricted to players who had 250+ recorded hands. As you can see, the outliers are already started to dwindle.

VP$IP in PT3 vs. PT4 (250+ Hands). Scatter graph showing VP$IP stability between PokerTracker 3 and PokerTracker 4. Vertical axis is VP$IP in PokerTracker 4. Horizontal axis is VP$IP in PokerTracker 4. Showing players with 250+ hands from a sample of 68k players, 800k hands.

Finally, the third graph was generated using a solid 500+ hand requirement. Here there's only one visible outlier.

VP$IP in PT3 vs. PT4 (500+ Hands). Scatter graph showing VP$IP stability between PokerTracker 3 and PokerTracker 4. Vertical axis is VP$IP in PokerTracker 4. Horizontal axis is VP$IP in PokerTracker 4. Sample of 68k players, 800k hands.

PT4 hasn't been released yet and we're not privy to the details of that release. But assuming similar patterns across common game variants, the data suggests that you probably won't need to recalibrate your VP$IP ranges when upgrading from PT3 to PT4.

[1] The big blind is voluntary in the sense that you can refuse to pay it, gather up your chips, and storm off into the night. But in the context of a game you've agreed to participate in, it's mandatory.

Tags: Hold'em Manager, poker strategy, PokerTracker, VP$IP, poker

17 comment(s)

"That is, in order to maintain a VP$IP of 100%, you must RAISE every time you're in the big blind and the table limps around to you."

Why can't you check, then make a bet (or call small blind's bet) on any flop?

jsn on Sunday, August 28, 2011

Ok I just realized VP$IP was only for preflop. All these years I was assuming you could VP$IP on the flop.

jsn on Sunday, August 28, 2011

so... if the new formula is cnt_vpip / (cnt_hands - cnt_walks) x 100... what happens when everybody limps around to the blind and he raises? Not counted as a VPIP?

Anonymous Coward on Sunday, August 28, 2011

Its pretty crazy how many of these stats are way more complicated than you think. I've been playing poker for years but now am implementing a stats tracking system for http://www.pokercloud.com and realizing how much random stuff like this you have to consider.

Is it a 3bet if a guy raises, some-one calls and you squeeze. Its definitely a squeeze but also a 3bet? Everyone handles all these corner cases completeing differently. Good news is in the long run I don't think it makes much of a difference.

PokerCloud on Sunday, August 28, 2011

What makes the problem even worse: the sites won't standardize on a hand-history format. The stats already had inconsistencies, and some of the more obscure stats are downright buggy, and not having a consistent HH format just makes the problem worse...

Anonymous Coward on Tuesday, August 30, 2011

Now I'm curious. What is poker cloud and why do you need to define 3bets?

Anonymous Coward on Friday, September 02, 2011

A web service to store/analyze hand histories?

Anonymous Coward on Monday, September 05, 2011

so now you can have a vpip > 100 ?

internet on Sunday, August 28, 2011

My read is that max VP$IP would always be 100%. cnt_vpip would have to change so cnt_vpip can never be greater than (cnt_hands - cnt_walks) but I doubt > 100% vpip will be possible. Too confusing

D.B. on Sunday, August 28, 2011

It's a little early to speculate on the details, but all such wrinkles can be solved by defining cnt_walks cleanly. For now we just wanted to share the graphs the PT team put up on Twitter. When we get our hands on the PT4 public beta we'll update.

Coding the Wheel on Tuesday, August 30, 2011

In the VPIP formula cnt_vpip expands to 

sum(if[holdem_hand_player_statistics.flg_vpip, 1, 0])

When the hand history is imported PT3 sets a flag for each player (flg_vpip) based on if PT3 deems the player voluntarily put money in preflop. If he has, the flag gets a "1". If he didn't, it gets a "0". Voluntary just means: whatever PT3 says it means.

Anonymous Coward on Tuesday, August 30, 2011

this is interested!

kevin117 on Tuesday, October 04, 2011

A serious looking iron designed to be amazingly forgiving and explosively long, the JPX-800™ is the first model to bring Mizuno’s Global JPX technology platform to Left Handed TaylorMade R11 Driver the United States. The unbelievable distance of the JPX-800 irons is achieved by combining MAX COR Technology, Hot Metal™ Face construction, and an exotic pocket cavity to hit the maximum allowable limits for energy transfer. Left Handed TaylorMade R11 Fairway Wood Mizuno MP-59 Irons Titleist VG3 Irons Left Handed Callaway RAZR X HL Irons

wholesale golf clubs on Friday, November 04, 2011

Hello! Sorry for being off topic, but how hard do you think would be for a poker site to find out if i bought hand history or not ?

andrew on Sunday, December 04, 2011

Good site! I truly love how it is simple on my eyes and the data are well written. I’m wondering how I might be notified whenever a new post has been made. Have a great day!

federaldrugs on Monday, December 19, 2011

for a few minutes. When Putnam came up to his party he said "we drive them." Now let it be

Coach Factory on Tuesday, December 27, 2011

For now we just wanted to share the graphs the PT team put up on Twitter. When we get our paroxetine dosage hands on the PT4 public beta we'll update.

kash on Wednesday, February 08, 2012

Use the form below to leave a comment.

Site Robot on just now!





Subscribe to Coding the Wheel over email or through any RSS reader. Coding the Wheel has been published since 2008.

Home | About | Contact | Subscribe Copyright © 2011 | All Rights Reserved