Do you need any luck in poker or is it all down to skill? We sent one of our intrepid reporters to find out

Is poker more about skill than luck? Paul Cheung speaks to the man who’s using science to highlight the gulf between expert players and poker novices…

Phil Hellmuth famously once said that if it weren’t for luck, he’d win every time. As quotable quotes go, it’s one of the Poker Brat’s best. Because although he might be exaggerating a little, he knows that even the best players need the cards to cooperate in order to win at poker. What Phil is saying is that the more ‘correct’ decisions you make in poker the better you’ll do – and that the best players (like him) always make the best decisions.

The trouble is, until now no one has produced anything more than anecdotal evidence to suggest this is the case. Could the likes of Phil Ivey, Daniel Negreanu and Mr Hellmuth simply have been running hot for all these years?

Well, thanks to the efforts of Joey St. Germain at this year’s World Series, we now have an answer. The 29-year-old Ph.D from Florida State University may not have been racking up bracelets like Jeff Lisandro and Phil Ivey at the 2009 World Series, but tucked away in a small meeting room at the Rio he was making an equally important contribution to the game…


Four years ago, St. Germain decided that he wanted his upcoming Ph.D to combine two of his great passions: sports psychology and poker. ‘I did sports psychology when I was 21 and I’ve played [poker] since I was six years old for baseball cards and pennies. I’m not a professional but I love playing it.’ The focus of his study would be the differences in decision-making between expert, intermediate and novice poker players, an area often explored in sports but never formally investigated in poker. ‘It’s one of those things where you assume there are differences but you wonder where those differences are.’ Ultimately, he wanted to find out what makes a pro a pro.

His plan was to find 15 players from each of the three different levels (45 people in all) and analyse their play and thought patterns. Finding novices and intermediates would be straightforward. ‘They were just your average person who was good enough and willing enough to give themselves to this study.’ Novices were judged to be those with less than 50 hours’ playing time under their belts, while intermediates would have between 500 and 3,500 hours. Neither had to be a ‘winning’ player.

Finding experts, however, was a more difficult task and St. Germain concedes that he would have taken flak however he did it. ‘For the experiment to be perfect, you’d have to get the 15 best players in the world, but if you ask lots of different people who they think the ten best players are, the list would probably end up with around 100 people. There would be similarities like Phil Ivey but you could get into a semantics argument on what’s an expert and what’s not.’

For confidentiality reasons, St. Germain can’t reveal the names of the pros he used, but he confirmed that there were WSOP bracelet winners, a WPT Player of the Year and some former world champions. ‘All of them ended up having 8,000+ hours of experience each.’ Getting 15 top pros to travel to Florida was wishful thinking, so St. Germain made the trip to Vegas during the World Series where he knew the pros would be. ‘I did it during the first four days of the Series, which was tricky, because it’s such a busy period. I went through every medium I could think of: PokerStars, Full Tilt, Two Plus Two, etc.’


Each player would play 60 computer-generated hands and be required to verbalise their actions and why they were doing them. ‘We didn’t just want to find out if there were differences in the quality of their decisions,’ he says, ‘we wanted to know their thought processes as well.’ To make things more interesting, he decided to impose two further conditions. First, the computer players would play in different ways, i.e. all loose-aggressive or all tight-aggressive.

In addition there would also be timed and untimed hands. Untimed, the players had as much time as they needed, whereas they had 15 seconds for each timed decision. The 60 hands were then split into three lots of 20 hands, each encompassing the different playing styles. Half of those 20 were timed, the other half were untimed.

St. Germain admits that 60 hands is statistically quite low, but he had to be pragmatic. ‘In a perfect world we’d have them play a million. But we thought, “What is enough hands where we think we can find statistically significant, reliable, valid results and balance that with the real-life component of how much time people will give me without getting bored.” After that we thought there would be diminishing returns because they’re not playing for real money.’

Each hand would play out like a cash game, but there would be no accumulation of funds and the computer players had no ‘memory’ from the one hand to the next. The only information that was on screen for the player would be their cards, who was still in the hand, and how much money they had behind. ‘Some programs do one thing 85% of the time and something else 15% of the time,’ he explains. ‘But in terms of this study, everything we could control was controlled.

If you and me played the hand exactly the same way, the computer players also needed to play exactly the same.’ By imposing these strict rules, the study would sidestep the obvious criticism that the game wasn’t ‘realistic’ enough. The point wasn’t to try to mimic a real game with humans and thousands of pounds on the line. It was simply to find out how the various player types rationalised their play.


Unsurprisingly, there was a huge gulf between the success of the experts and the novices. When it came to tackling the different playing styles, the novices played every hand the same way, whereas the experts and intermediates adapted. In terms of pure profit, this led to the experts making an average of $18.22 per hand, whereas the novices managed to lose almost the same amount. This was also reflected in the EV scores (how much they could expect to win if luck wasn’t a factor). Experts enjoyed a positive expectation of over three times that of the novices.

What was more surprising, says St. Germain, was the comparison between an expert and intermediate player. ‘The differences between experts and intermediates was nowhere near as wide as between experts/intermediates and novices,’ he says, pointing to the average profit of $14.63 per hand for intermediates and -$17.26 for novices.

However, while experts and intermediates largely reached the same decisions, it was the explanations of how the two groups reached it that highlighted the knowledge gap. This is why the intermediates’ EV scores were somewhat lower than the experts. ‘The intermediates aren’t thinking about the future as much as the experts,’ says St. Germain. ‘For the experts, the hand was one long thought process, whereas for the novices and intermediates, every decision was a separate one. For the novices, it was like every street they played was a new hand.’


The timed conditions threw up some expected results but also some strange anomalies. Although they were thinking at a more advanced level, the experts made their decisions quickly. ‘The literature backs this up,’ says St.Germain. ‘The expert players have all this experience, so they’re just adapting the current situation to a situation they’ve seen before. The novices had the most trouble with the timed hands and would say stuff like, “God, I’m glad the timed condition is over, I’m nervous and sweating.” ’

Strangely, though, one area where the experts fell down was on the river – but only in the timed hands. In fact the novices recorded the highest EV on this street, even though it is usually considered the most difficult. St. Germain believes he has a good explanation for this apparent anomaly. ‘If there was a tricky spot on the river, the experts would re-run the whole hand in their head.

But in 15 seconds, they didn’t have time to go through the normal processes they usually go through, so their performance levels dropped. The novices just played the game.’ St. Germain makes the point that in reality this wouldn’t be a problem. ‘The only reason why we did a timed condition was because of other timed conditions in other sports psychology studies. There aren’t many real-life situations that would cause a poker player to play a hand every 15 seconds.’


Although the study didn’t throw up any revolutionary ideas, it did support some aspects of poker thought processes that we always suspected. Novices tend to think only about their hands and things not necessarily related to poker. Intermediate players take it a step further and think about the hand, pot odds and position. Pros automatically take all these factors into account but mainly focus on what their opponent is thinking. St. Germain now hopes his findings can help the poker industry as a whole. ‘I guess the biggest practical application in the US is getting online gambling legalised, as the question remains: is poker a skill game? As far as I know, this is the first scientific experiment that shows poker is in fact a skill game.’

PokerPlayer magazine features 24 pages of expert poker strategy every month HERE

Pin It

Comments are closed.