Levitt's Poker Paper: It's a Game of Skill

Photo: Digital Vision

Steve Levitt has a new working paper: “The Role of Skill Versus Luck in Poker: Evidence from the World Series of Poker,” with University of Chicago colleague Thomas J. Miles.

Using data from the 2010 World Series of Poker, Levitt and Miles found that high-skilled players earned an average return on investment of over 30 percent, whereas all other players averaged a 15 percent loss. This finding has serious implications on the legality of online poker, as that debate is heavily dependent on whether the game is based on skill or luck.

The press has gone to town on Levitt’s timely paper, as the executives of Full Tilt Poker, PokerStars and Absolute Poker – three online poker  companies – were indicted by federal prosecutors last month for “tricking” banks into processing gambling profits. Check out the coverage on the Times‘s Economix blogThe Huffington Post, and the U.K.’s Daily Mail.


Is this only for Texas Hold 'Em? Do other games have similar results? Skill vs Luck depends upon the game. It would be easy to imagine poker games that are very luck-based (1-card stud or something similar). Thus, I could see a significant difference between different standard games, such as those played at the WSOP.

I am also interested at what kind of difference would make something a game of skill vs a game of luck. Apparently Levitt considers 30% vs -15% disparity a game of skill. Where does this determination come from? What if a game was 10% vs -5%?

How about sports betting? It seems like sharps could pull something like +5% and squares might be -10%.


While I agree that skill is an important part of poker, the legal question seems to be, does skill DOMINATE luck? I am not sure how to answer question, but I don't think this excellent paper answers THAT question.


@Chad: At its core, poker is decision making driven by risk vs. reward. It doesn't matter the variant (Hold 'Em, Omaha, Stud) the key is making sure that the payoff is greater than the risk. In the case of your suggested game of 1 card stud, the right decision would be to not play at all. Games that are too skewed toward chance lack the interest to be played formally and for real money. Mexican Sweat may be fun for chips or pocket change but you will never see thousand dollar bankrolls play it because the luck factor is simply too high.


This line of reasoning suggests that there is no such thing as gambling, though - one could say the same thing about roulette, lotteries, etc - the "skill" is in avoiding playing it.


In the games of chance you mention, by definition the reward never pays off for the risk being taken. Roulette has a 5.26% house edge; no matter what the player does he cannot come out ahead. Poker, on the other hand, is a zero sum game. The house makes money through taking a small percentage of each pot, but the rules of the game play favor no player over another. "Gambling" is entering a proposition wherein the outcome is determined by an element of chance. The skill comes from deciding when and when not to gamble. As you point out, with games played against the house the correct decision is always to not do it.


As I understand it, the real question (the one in the courts) has little or nothing to do with the game of poker, but whether or not the creative financial methods of poker websites constitute bank fraud and/or money laundering.
The skill level of the players have to matter too. Levitt's paper focuses on The World Series of Poker, where you would expect skill to be more dominant. But what about for the masses of relatively unskilled players that probably dominate the popular poker sites?


Table 2, row 3 of the manuscript show a priori non-skilled players had ~6x as less tourney experience as the a priori skill players. Maybe experience and dealing with human elements (eg. familiarity with the live-event pressure, nervousness, live audience pressure) explains the ROI difference?

Dave Martin

Would the results be the same if these same players had the same cards online? How much of this skill is reading your opponents in person? Does this really equate to the online game?


I don't doubt that there's at least a kernel of truth to this, but I'd be interested to know how he defines a high-skilled player. Given that the WSOP lasts six weeks and the vast majority of players are only in town for maybe one or two tournaments, how can he possibly measure their skill?

Mike B

Poker is just a repetitive social role playing game where the cards dealt define the roles people play. I believe what people, including myself, find objectionable about it is that the skills required to be good at Poker are the same skills required to be a good conman or businessman, ie fooling people for profit.

Poker is nothing I would want to ban, but I don't think a good society should be encouraging it either. A game of pure chance would arguably be more fair than what Poker actually is because in a game of luck players would win or lose depending on the cards dealt to them. In poker the skill comes from gaining social advantage over others through observation and deception. It is the world where the honest or productive lose to those who seek to take advantage of them. One can tag this as a unifying thread throughout human history where the sharks eat the guppies and appropriate all that the guppies have worked for with little effort of their own.

Frankly I would hate to see online poker embraced by the state as I feel the best thing for the game will be its complete ruination by Bots, cheating and collusive play. As soon as people realize that they cannot win most will stop playing and it will join the gambling graveyard with horse racing. Gamblers should be steered towards activities that aren't zero sum, like investing, or those that provide beneficial externalities to society, like prediction markets.


Rucksack Revolution


While observation and deception are valuable skills for a poker player, there are a lot of other qualities that can give a player an advantage, such as a deep understanding of statistics and game theory. Also, not every losing poker player is being taken advantage of. Many are well aware that they are losers and are satisfied with spending some money in exchange for entertainment (in many cases much less money than they would spend on alternate sources of entertainment).

You suggest that people be steered towards activities like investing. But aren't many investors taken advantage as well, arguably more so than losing poker players? There are many investing experts and companies who take advantage of those less knowledgeable. What about mutual fund companies that take excessive fees? Or financial advisers who work on commissions and whose interests don't align with their clients? Or a trader who has insider or expert knowledge and sells a given stock...this person has to sell it to someone else, doesn't he?

I don't really think playing poker is all that much different from attending any other sporting event. People who go to a basketball game are paying money for entertainment and this money goes to the players, owners, and other entities who gain financially. Playing poker just has more variance than attending a basketball game...although if you sometimes buy lots of beers and pretzels at the game maybe that isn't true. In most cases, though, the players are well aware of the variance and may even get additional enjoyment from it.


Mike B

You're sports analogy is correct, but not in the way that you think. Real Money Poker is less attending a sporting event for entertainment and more like investing in a career as an athlete (or singer or any other tournament style career). You can buy all the right equipment and spend all the time practicing or training, but for the vast majority of folks they will probably lose money in the endeavor.

Just like with Poker I would never ever encourage someone to attempt to make money as a professional athlete or artist. It may very well be entertaining, but it is just not make good economic sense. Now imagine how the sports model would be different if on top of the usual costs you would actually lose money for poor performance. Every strikeout or missed free throw or put would be charged against your earnings. Engaging in such an activity is simply ridiculous and anyone who does so is either highly skilled or really bad at making good decisions.

At the end of the day the telling fact is that you don't have to play for real money to get entertainment value from playing Poker. If chips were simply a way to keep score then the game would be arranged in a similar fashion to Monopoly where players would eventually go bankrupt and a winner would be declared through pure demonstration of skill. What drives people's objection to Poker is that every round delivers hits to that risk/reward center of the brain through a real money betting activity. That is why the game is addictive, that is why it is considered gambling. The problem with the law doesn't lie in its definition of poker as skill or chance, it lies with the fact that gambling isn't about either, its a neurochemical effect on the brain that impairs rational judgement in large segments of the population.

Long story short it would not be hard to modify the way Poker is played to emphasize the skill over the betting. Only problem is that would severely reduce the addictive qualities of the game and make is far less profitable. People who read this blog should be strongly in favor of free choice. Systems that are designed to trick or addict the human mind remove free choice. Divorcing real money betting from the Poker gameplay would probably solve both the legal and free choice problems.



What happened to the Pokernomics project, which was supposed to analyze millions of on-line hands played and captured electronically? That would have a much more scientific sample.


We wrote a response to the paper here:


Our summary is this: the paper is a classic case of getting fooled by randomness.


Good work Levitt, but the skill vs luck argument is a red herring as far as online poker's legal status goes.

Brian James

Here is a fairly simple proof that Online Poker is skill.

There are many low to mid stake cash players out there with super long winning streaks. Dusty Schmidt is a famous one with a 5 year plus streak of never having a losing month.

The chances of never having a losing month for 64 months if poker is chance is 2 to the 64th power. So the odds if not skill of a winning streak this long is 1 in 18,446,744,073,709,551,616.

That number is in the quintillions if you weren't sure.


Your conclusion comparing a head to head match up of poker players to that of MLB seems flawed. You discuss the poker match up being skilled vs unskilled, yet clearly in the case of MLB, it is highly skilled vs highly skilled. Anyone with a bankroll can enter into a poker tournament, fall under the "unskilled" category, and based on your analysis, has a little worse than a coin flip chance of beating the pro. Do you think I have a chance of even making contact if taking batting practice with Josh Beckett?

Yes, it is generally accepted that MLB is a skilled game, but if you were to compare a ball player skilled vs a ball player unskilled (like you did with poker), skilled would win 99.99% of the time.