When the prize is worth millions, Amir Lehavot called in a pro
In late October of this year, I got an email from Amir Lehavot asking if I’d be interested in helping him prepare for the WSOP Main Event final table. (For those of you who didn’t follow that event, Lehavot was one of the “November Nine” and ended up placing third for $3.7 million.) My role in helping him would be to analyze his opponents and him for possible behavioral patterns, i.e. poker tells.
Lehavot is a very skilled player with a great track record, mostly online but also with a few big live scores since he started playing live more in the wake of “Black Friday.” He was arguably the most experienced MTT player at this year’s final table. I was flattered to be asked to help him, considering I am not a high-stakes player nor am I well-known apart from some good press surrounding my poker tells book and blog. Lehavot said that he had decided to contact me because he had enjoyed the book, especially the way it emphasized the importance of the situation when studying poker tells.
I was excited by the rareness of the opportunity. As far as I know, nobody has been asked to play such a role in a major poker tournament before. If someone has played such a role, which is possible, they haven’t gone out of their way to publicize it.
Lehavot and I both were on the same page about the possible usefulness of such an endeavor. We both agreed that my work would most likely not be a huge game changer, but there was a chance that it could be quite useful. Lehavot believed, as many experienced players do, that the behavioral patterns of opponents can sometimes play an important role in swaying decisions in spots that are borderline. And these borderline decisions come up more frequently the tougher and more balanced your competition is. Considering the amount of money at stake, Lehavot thought it would be prudent to have someone dedicated to only that aspect of the game.
There were a few major difficulties in studying the November Nine for poker tells.
The first difficulty was that this was arguably the most skilled Main Event final table in history; eight of the nine final table players were professional-level players. While I believe that every poker player has some behavioral patterns, I also believe that most skilled live players will have patterns that are hard to spot.
The second difficulty was just one of sample size. In the ESPN footage that was broadcast leading up to the final table, there were only a few significant hands shown from each of the players. (By “significant” I mean hands involving fairly big bets.) Even in the significant hands, the editing of the episodes sometimes meant that you couldn’t see the player in question for some key moments of the hand. (For example, one key moment in analyzing poker tells is studying someone after they’ve made a big bet; in televised poker, however, they often cut away from the player who has just bet.)
The third difficulty was the four-month hiatus the players take before playing the final table. Even assuming some of the November Niners had reliable poker tells, they would have plenty of time to prepare more “unreadable” behavioral strategies. I’d have to go into watching the final table keeping in mind that the player’s behavior could be very different from what I’d seen before.
In the couple weeks leading up to the final table, I studied the WSOP footage of the November Nine players and any other footage of those players that I could find. I kept spreadsheets of possible behavioral patterns for each player. Considering the small sample size of observed hands, my notes were just possibilities of what might exist; a way to focus on behavior that could potentially be of interest during the final table.
Although I’d never done such a mission-specific job, I have analyzed a lot of poker behavior over the years. I have a growing database of interesting hands and patterns, both from televised hands and hands that I’ve played. Trying to find ways to make my database easy to navigate has necessitated creating shorthand ways to code behavioral patterns. Even though working for Amir was a new type of project, it did help that I’d already created some methods of efficiently processing this kind of information.
We’d decided that there was no value in me actually being present live at the final table. Instead, I was watching the 15-minute delay broadcast on ESPN. I had two computers set up; one to take notes on and one that was digitally recording the broadcast for quick playback if needed.
When the final table started, I posted my most relevant observations in a Skype group, which Amir could read when he visited his rail during the frequent commercial breaks. Choosing what to tell Amir was also a challenge because I didn’t want to bombard him with a lot of annoying chatter on very theoretical things. I erred on the side of only telling him the information that was most likely to have meaning, and I didn’t post very often.
About eight hours after the final table started, Amir was knocked out in third, taking home $3.7 million. Amir entered the final table second in chips but he fairly quickly became short-stacked and remained so for most of the game.
While Amir stated afterward that he believed my contribution was valuable, he only played a couple of hands that had significant post-flop action; understandably there were no situations where an opponent’s behavior swayed his decision.
I could talk for a long time about my observations on the November Nine. As expected, considering the skill level of the players and the small sample size of observed hands, there weren’t a lot of patterns that I can point out as being convincingly reliable, although there were quite a few patterns that I believe were reliable.
Below, I will describe a couple of patterns that I thought were interesting, and that will give you an idea at the kinds of things I was looking for. (If you’re interested in reading more about my observations, I’ve put up a couple of blog posts on the subject at www.readingpokertells.com.)
Loosli is a primarily online cash game player. He was very new to live poker at the start of this year’s WSOP. His pre-final-table footage showed a lot of extraneous movement of his body, arms, and hands. It’s a basic assumption of mine that whenever there are extraneous movements (meaning movements that don’t contribute to a specific, practical action) there is some kind of information present. It may be information that is difficult to interpret and use, but it is there.
It was evident Loosli had made efforts to become more unreadable during his four-month hiatus. He was much more stoic during the final table. But he still was probably the most “loose” with his body movements of all the players. One of the situations in which he exhibited extraneous movements was in the gathering of his chips before betting or raising. For many of his bets or raises, there would be a long pause between the time he reached for his chips and the time he put them into the pot. During this time he would move the chips around, stack the chips, and add or subtract chips before finally putting them into the pot.
On the final table, Loosli had KK twice and QQ once. In all three of these hands he three-bet preflop. Before three-betting in all of these hands, Loosli showed a lot of pausing and hesitation in the gathering of his chips. For instance, in two of these hands, there is a moment when he starts to reach for chips or puts a chip on his cards, so we know he’s going to be raising, and then he pauses for several seconds before continuing to gather chips.
Also, in the hand where he held AA and bet the flop, there was a moment when he looked down at his chips, making his interest in betting obvious, and then proceeded to shuffle his chips for a while before betting.
Comparatively, in the hands where he was 3-betting light or betting postflop with a weak hand, there were not these hesitations. The betting motions with the weak hands were much more straightforward and practical, with less “useless” motions.
While there weren’t many significant postflop situations in which to compare this behavior, I’m fairly confident this was a pattern for him. Making small gestures of uncertainty and hesitation when holding a strong hand is a common pattern for many players. It’s a similar but more subtle version of the shrugging gesture that you can see some amateur players do when betting strong hands. These kinds of gestures seem intended to communicate, “I’m not sure about this bet.”
The other main factor in this kind of behavior is that players betting a weak hand instinctually want to communicate confidence with their actions. This explains why a behavior like hesitation before betting, even if it’s a subconscious behavior and not a consciously deceptive behavior, is more likely to be seen when a player has a strong hand.
Farber was the only recreational-level player at the final table. He ended up coming in second. From a behavioral perspective, his two most interesting hands were the two big heads-up pots he played against Ryan Riess during the heads-up portion of play. In the first one, he made a large bluff on the river and Riess folded. In the second hand, he made a large value-bet with a rivered flush that Riess called with Queen high.
During the heads-up play, I was posting some of my thoughts live on Twitter (Amir was no longer in so I didn’t have a conflict of interest). While I wasn’t that sure about Farber’s first big bet, in the second hand, as Riess contemplated, I tweeted, “If Farber is bluffing here, I’ll eat my hat.” I was very certain that Farber had a hand this time, for two reasons.
The first reason was that Farber had much more variety in his post-bet body language when value-betting compared to when he was bluffing. If you watch the first hand, when he was bluffing the river, you’ll see that he stays continuously looking toward Riess while Riess thinks. He is very still. In a few other hands where Farber was semi-bluffing or bluffing, he also stayed looking in one direction, either at Riess or down at the table. When you look at the hand where he rivered the flush, you’ll see that he was looking at Riess for a few seconds, then looking down at the table, then looking back at Riess for a while. It’s a common behavior for some people to be more still and stoic when bluffing and more behaviorally loose when betting a strong hand. Also, post-bet eye direction (where someone looks after making a significant bet) is also something I spend a good amount of time studying. This variability in Farber’s behavior was a big indicator he was relaxed.
The other major indicator was a very genuine smile Farber exhibited after betting the flush. In the river bluff hand, Farber had a small, tight smile that didn’t involve his eyes and that was asymmetrical (Paul Ekman, the respected facial expression researcher, pointed out that smiles that were asymmetrical were more indicative of false emotion, while real smiles were more likely to be seen on both sides of the mouth). This small smile wasn’t necessarily meaningful to me; I could easily have imagined it being present when Farber was bluffing or value-betting. It could have just been something he strived for as a standard expression. And indeed in the hand when he rivered the flush this same smile was present in the same way for a while, almost as if Farber was trying to emulate that same smile from when he was bluffing earlier.
But at some point, I think the cheering and the yelling from his rail got to Farber a bit, and he showed a very genuine smile that was very different from the small, tight smile. The genuine smile was bigger and it involved his eyes more, as genuine smiles do. It was also more symmetrical, involving both sides of his mouth.
When I saw this smile, coupled with the variation I’d seen in his body language, I was almost 100 percent certain that he had a big hand.
All in all, working on this project was a big learning experience for me. I don’t pretend to be the best reader of poker behavior or a “poker tells expert;” I am just someone who is interested in the subject and enjoys thinking about it probably more than most. Most of my time studying poker tells has been invested in studying amateur players, just because studying good players doesn’t offer the same amount of financial reward in the games I’ve played. Working on analyzing the November Nine improved my approach on how to best study the behavior of good players.
To wrap up, here’s a question I asked Lehavot during our post-game interview.
Do you think that for big tournaments like this, having someone on your team who studies the behavior of opponents is a good decision? Do you feel like this will become a more common decision in the future?
Amir: I think it will, actually. As far as having a long break before a final table, there’s only one tournament a year that has this, but the tournament is so big that it makes sense for people to hire someone to do that for them when they get to that position. I think there’s definitely a market for it for this one specific tournament, and then beyond that I think for guys who are regular live tournament players, especially guys who play high rollers where it’s the same group of poker players playing it — if I was doing it, I would think doing my homework makes sense, doing the same stuff that we did: minimizing my own live tells and learning more about other players.
Even though they’re experienced, and they shouldn’t have a lot of live tells, it’s just a cost versus benefit thing; the cost is really small relative to the potential benefit. Poker players often have big egos or maybe they’ll think that it would be looked upon negatively if they chose to do something like this, but I’m sure some of them would think it’s logical and makes sense. Probably not everybody is going to have the same views as I do as to how much it makes sense.