The response has been overwhelming. I think we’ve got over 7 million poker hands already. (For those of you who sent hands, but don’t have your Freakonomics book and t-shirt yet, be patient, they are coming, but we haven’t been able to keep up with the back orders.)
Other than being buried in data, my only problem is that everyone who is sending in hands is a good player! This makes sense. If you aren’t a good player, you tend to quit. So the majority of people who amass huge hand histories are winning players. Also, I suppose it is more satisfying to send your hand histories to me if you are poker star than a poker dud.
But to really tell the good players from the rotten, I need some more losers!
Just to show that I am serious, I started playing online poker a few weeks ago, and tracking my hands. I’m determined to make it to 10,000 hands so I can be a data point in my own study, no matter how much it costs me.
So if you are as bad as me at poker, or know someone like that, please send them to www.pokernomics.com. My initial look at the data makes me think that I can give players (especially bad players!) some useful insights on how to do better.
25 Comments
Isn’t this selection bias? Bad poker players stop playing. People who start out bad but have the perseverance (and money) to keep playing, inevitably will improve — or, go broke, and end up in the first category.
— DossyI’m a pretty bad poker player (I like to think unlucky), but unfortunately I haven’t tracked my hands. Maybe if I had, I’d have become a better player. Oh well.
— AnonymousNow I don’t know about winning/loosing hands, but it seems to me that online poker might lead into a very freakonomic realm: the development of Artificial Intelligence. Computers are good at calculating odds, counting cards better than Rain Man, say. But they’re not good at chatting across the table. If a computer ever passes the Turing test (a program that can converse in a manner indistinguishable from a human) you better believe it’ll be playing poker online.
— 3612It seems to me that this would be a good application of a modern computational experimental approach, like genetic programming. It would be pretty straightforward to breed a variety of bad poker-playing agents, to offset the selection bias mentioned above.
— Bill TozierHi Steven -
Unfortunately (for you), my own hand history DB won’t help. But thought I might point you to the old IRC hand history info. Those games were significantly better than today’s medium and low limit play for real money and Michael Maurer created a huge database of hands.
http://games.cs.ualberta.ca/poker/IRC/
Good luck
— Tim OgilvieThat’s a tough challenge. You need to find a body of people who are interested enough in online poker to track their hands, are wealthy enough to play losing poker for over 10,000 hands, and yet are drawn by the lure of a free T-shirt.
Anyone willing to drop the money on tracking software is probably already a decent player. What ways are there to get crap players to start tracking their hands? Joint promotion with Poker Tracker? - Buy Poker Tracker at a bargain price of $35; submit your first 10,000 hands to the Freaks and get a chance to win an ipod (everybody loves ipods right?)
Funny that when you started promoting this, the first reaction of many people was that good players would never give away their databases.
— angrymanangryman: what about free online poker sites? If Freakonomics becomes their “partner” — and, the Terms of Service and Privacy Policy says “your information won’t be shared with third-parties, only partners” (ha, ha) — then you could get a fair sample of winning, and losing, players and their hand records.
But, then there’s another selection bias: people who play poker for free, instead of for real money.
— DossyFree poker will give skewed results… You always play different when playing for matchsticks than when playing for dollars.
— Jason PerdueThe currently operating poker sites I have to imagine will keep their data secret at all costs. Are there any failed online poker rooms looking to salvage a few bucks? Surely among all the new entrants someone has gone belly up. Can you buy data at a bankruptcy sale?
The “online poker community” the people who read poker blogs, mags, websites, etc. is too good. It sounds like you need a bunch of newbies to do exactly what you’re doing - start playing now and get 10,000 hands. Jason is right that free poker is different from real poker, but only because people have no incentive to win. What about a contest for UofC students with a prize good enough that they will want to play well? Starting September 10, students have 2 weeks to play 10,000 hands at the lowest limit on Poker Stars (if memory serves, Stars has the smallest limits). Top 3 performers win an ipod, and everyone who submits gets entered into a draw for another ipod. Minimal financial risk for students, enough incentive that they won’t just play like idiots, and good players won’t bother competing because, as anyone who plays poker knows, all good players already have ipods. Promote it in the school paper.
— angrymanPerhaps this is a little off topic, but the current issue of Wired Magazine has a very interesting story on pokerbots, online here. The title: “On the Internet, Nobody Knows You’re a Bot”.
Winholdem (a pokerbot) has an interesting site here, with a forum.
— Eric Gallowaywhat if bad poker players are what make people good?
— justakimAnybody who continues to play poker, even though they are lousy, is a hopeless romantic. Take the fact that all the data you are getting is from good players and run with it.
There is a big difference is playing technique between better than average players who show a tidy profit, and exceptional players who can make a living playing poker. Your research shoud be directed toward determining the difference between better than average players and great players.
Forget about the lousy players. Fortunately, there’s no hope for them.
— Michael GiesbrechtI like the Freakonomics a lot. Forgive me for being naive. What is the point to promote the poker game, to me, it is a way of gambling?
— A-Time-To-TalkSince I use a Mac, I play at pokerroom.com (Java applet) which does not save information in Poker Tracker’s format. I use MacPokerPro. Is there some tool to convert to Poker Tracker format out there? Or is there a place where the Poker Tracker format/API is published so that I can convert my hand history (10,000+ hands) myself and send it in?
— AnonymousPokerbots will kill off online poker. Who wants to play against a machine? Certainly not a newbie. Can you imagine sitting down at a chessboard and facing Big Blue?
— AnonymousProbably the first program to pass the Turing Test will be a pokerbot. Now there’s a monetary incentive to beat the test!
— AnonymousI’m guessing that if pokerbots become prevalent, the industry will find a way to adapt to exclude them.
— AnonymousThis is a great idea, Steven. I’d like to learn to play poker, too. Could you tell me what poker sites you are playing bad poker at and which rooms you play in? It sounds like a great way to learn!
— AnonymousThe teachers are coming over tonight to play lousy poker. Too bad you couldn’t track our hands — we’re not so hot at no-limit hold’em but we don’t cheat, and our test scores are high. :-)
A friend recently wrote something up on bad poker. I hope he doesn’t mind my passing on the the link.
— Princess LeiaOne thing to keep in mind when looking at poker data - there are essentially four different styles of play - for free poker, low limit poker, high limit, and no limit. There’s a number of poker books that talk about the differences - and the strategies that work for, say, no limit, don’t work for low limit poker.
So I’m worried that you’re going to get in trouble even if you do analyze good data, since you’re looking at three (or more) different strategic scenarios…
— Jim Driscollangryman, not everyone would have to play 10,000 hands. Just as many as you bother to keep track of would add to the data. Levitt is playing 10,000 as a testament to his dedication. But 10,000 is an arbitrary number.
— penxvAgree with Jim above. Harrington’s on Poker is an excellent reference for no limit hold’em strategy.
— Princess LeiaConcerning my previous comment: Developing a pokerbot is a difficult problem in artifical intelligence, contrary to the impression given by the article in Wired. Otherwise the developer of Winholdem would probably be a lot wealthier.
Here’s a good story from the New York Times on poker that talks about pokerbots.
The Computer Poker Research Group at the University of Alberta, has been working on the problem for more than a decade. See here for much more.
— Eric GallowayActually, if you want examples of bad play then the free sites are exactly the place to look. However, it is unlikely that anyone would play similarly on a for money site. Seems like there was is a web site out there where you could look at other player’s statistics. I am not sure how they are getting their data, but maybe there is a way to get the data anonomously, so that it is not attributable to any actual person. Perhaps you could identify poor players through the statistics accumalted at the site to which I am referring?
With regard to the previous post suggesting you find a bankrupt online casino, that would be the day.
— AnonymousThere was a post related to this on the twoplus two poker forums a few years ago (see link below). An analysis was performed using hand data to show the EV (Expected Value) for each basic playing style (Tight-Aggressive, Loose-Aggresive, Tight-Passive, and Loose-Passive).
http://archiveserver.twoplustwo.com/ubbthreads.php/ubb/showflat/Number/549460
— jgorsegner