Jason Les: 'Future Is Pretty Strong For Poker Bots'

High-stakes poker pro Jason Les is a cash game star, but he just recently had his breakout year at the World Series of Poker. Les final tabled two events this summer, recording a runner-up finish in the $3,000 No-Limit Hold’em Shootout and a third place in the $10,000 Pot-Limit Hold’em Championship.

His deep runs this summer came just weeks after he and three fellow heads-up no-limit hold’em specialists beat what is currently the world’s top poker bot in a two-week long battle in Pittsburgh. Even though the academics behind the machine said it was a “statistical tie” and Les told Card Player he agrees that it technically wasn’t a complete victory, the contest did show that computers have some ways to go before being able to reach the prowess of the best poker players in the world.

Card Player had a chance to talk to the 29-year-old Les about his summer, as well as his high-stakes poker career and the match versus the poker-playing bot called Claudico.

Brian Pempus: How was your summer at the WSOP? How do you feel you played?

Jason Les: I traditionally haven’t been much of a tournament player, but over the past year I’ve been playing more. So this was the first summer that I played a lot of events. I only play no-limit, so a lot of those. I was prepared for the worst case scenario in terms of how the results would be, but I was happy with how it turned out. The second event I played I finished second, and then a week later I got third in the $10,000 Pot-Limit Hold’em. It was a relief to have these great cashes right away. I knew I was kind of freerolling the rest of the summer. I was pretty much still going to have a profit locked up. Having the results right off the bat got me excited to keep coming back and trying to go for more final tables. You go into it and you don’t think bracelets are that important of a thing, but once you get close you want to keep trying. I didn’t play much cash this summer because I was spending a lot of hours on tournaments. The hours I didn’t spend playing tournaments I was studying [poker] and trying to relax with my friends. I am glad it’s all over so I can take it easy for a bit.

BP: People often say that tournament days are really long and grueling, but with cash games people sometimes play marathon sessions when the games are good. Can you talk about how being a cash game player might make it easier to fend off the fatigue in a tournament?

JL: I typically play heads-up cash online. The hours for that are all over the place. When I am outside the country playing on PokerStars I am sleeping at all random hours of the day, because when you get good action you can’t quit. It’s not like you can wake up tomorrow and it’s going to be there. You have to take full advantage of it. This did get me prepared for playing a lot of hours [at the WSOP]. Playing 12 hours a day for a tournament is very easy in comparison to some of my sessions when there is a good cash game running. Tournaments are different though. You are there in person, and it’s not the same as kicking back at home on your computer. But for sure playing lots of cash makes the tournament grind seem much easier and not super taxing on me. It just feels like another day of poker.

BP: Can you talk about the state of the online heads-up cash games these days?

JL: I think that over time the skill required to beat heads-up cash has increased, but there is still action out there to be had. If you are passionate and enjoy playing you can get out there and get in games. A lot of people create this idea that heads-up is completely dead and there is no action, and it’s because they are setting boundaries on who they are willing to play. There is no action for the top three players, so they live with not having that much action, but there are plenty of people getting heads-up no-limit action all the time. I think heads-up is a great way to learn how to play poker. You get involved in so many different hands and get used to thinking about so many different situations. It’s a great way to prepare for all types of no-limit hold’em formats. If you want to work on your game you should play heads-up and try to challenge yourself.

BP: Are you at the point where you will play anyone at anytime? Or are there some people you shy away from especially if you are having an off day for whatever reason?

JL: At this point, I’m pretty much willing and able to play anyone besides my friends. I am not going to start a session versus Jungleman after I have been up for 24 hours or something like that (laughs). You want to be in a good mental state before you take on the top names. I am in a position to take action from pretty much anyone I can and challenge myself versus everyone out there. I haven’t been able to play that much heads-up cash in awhile and I am looking forward to getting back out there again and mixing it up with some people and having some good matches.

BP: I heard someone say recently that it might be between you and Doug Polk for the best heads-up no-limit player in the world right now. Do you agree with this? Who is better?

JL: Doug has been a good friend of mine for a long time and I will never play him. With that being said, he is without a doubt the best heads-up no-limit player. I would be lucky to be as close to as good as him. I don’t think I’ll ever be as good as him.

BP: What would it take for you to get closer to his ability? Is there anything that you can’t really teach that separates the players at the highest level?

JL: I guess what you can’t teach someone is the problem solving and the ability to create solutions to the problems you see. You can tell people your own answers to things over and over again, but that doesn’t get them thinking about things all the way for themselves. What can’t be taught can only come through experience and practicing, playing really tough opponents over and over again, playing different styles, learning how to adapt and what counter-strategies are required to be successful.

BP: Can you talk about the counter-strategies you and your team developed to defeat the poker bot Claudico earlier this year?

JL: Since Cladico wan’t adjusting, the goal was for it to play its Nash equilibrium strategy that it hoped we would not be able to beat, if we could find a weakness in it we could massively exploit that. We wouldn’t have to worry about it noticing something we were doing over and over again and adjust to being exploited. So, our style got really crazy. We were doing weird bet sizes all over the place when we found out stuff that would confuse it. We took advantage of Claudico not taking blockers into account when it did different bluffs and when we would bluff it. It would try, for instance, to represent the nut flush and I would have the nut-flush blocker in my hand. I would know that it had to be bluffing because of that and I could make pretty wide calls for that reason. On the other hand, I could bluff with blockers and it would maybe make some bigger folds. The blocker thing was mainly us catching it bluffing though. That was the big deal. There were some really big pots. Claudico used really big all-in bet sizing to really put pressure on us, and it was a good overall strategy. Its river strategy was hard to play against even though we did pick off a lot of bluffs. In certain situations when I had a blocker I knew it was representing some hands that it couldn’t have. Another thing we were able to take advantage of was the fact that it only understood a certain number of bet sizes because there is only so much time and computing power they can run through a supercomputer. They can’t have a strategy for every single bet size. They’ll go through 1/4 pot, 1/2 pot, 3/4 pot, full pot, 1.25 pot, and so on. If you ever bet an amount in between those, it would round up or down depending. What that would do is cause it to treat bet sizes a little differently than they actually were.

BP: So these big over-bets on the river from it were very polarized?

JL: Yes. I do give a lot of credit to them for creating that river component. It was polarized. It would have total air or the nuts or close to nut hands a lot of the time as well. You had to really understand theory and what was going on to play appropriately. You can’t fold every time. It was a difficult situation. There’s a $500 pot and Claudico is all-in for $19,750. You really have to think appropriately about your range and what you want to be calling with. The blockers played a big part of that. The CMU team is aware of the blocker situation and know what is going on here. While we were able to take advantage of this now, I imagine in the future they’ll have taken that into account a lot more. It won’t be as easily exploited.

BP: This really massive over-betting on the river isn’t something we see in high-stakes heads-up matches between humans. Why is that?

JL: In the match versus Claudico we were playing 200 big blind stacks. Online, it’s always 100 big blind starting. So, you can get that deep but generally you aren’t. Humans typically don’t have a strategy like that because it is so difficult to balance appropriately. Claudico could do things at perfect balance, so it didn’t have that same kind of concern. Big bets are very tough for a human to do without some sort of bias whether it be to bluff or for value. Another thing is that Claudico would slow play a lot of hands in ways that humans never would. Sometimes it would bet and sometimes it would check; you couldn’t put Claudico on a hand because it just did whatever. So, there could be a situation in which I know a human would never have the nuts because they would never slow play on this board. Claudico very well could just because that’s the way it decided to play [that hand]. It could check the nuts until the river and then just go all in. Most humans aren’t going to do that because they won’t get good value from the hand. So, in short, it is very difficult for the human mind to do that strategy with the appropriate frequencies and the appropriate hands.

BP: What do you think the future is like for these bots and when we say a bot is “unbeatable” what does that really mean?

JL: I think the future is pretty strong for poker bots. The CMU team is extremely intelligent and good at what they do. It was fascinating getting to sit down and talk with these guys and hear what they learned about poker. These guys aren’t professional players. They are computer scientists who took on this project and they made a bot, even though we won 9 big blinds per 100, that hung with the best players in the world. In the next year or two I expect these guys to make some really good progress. We could very well see an unbeatable heads-up no-limit bot. What that means, what this team is trying to do, is to get a bot that plays at a Nash equilibrium strategy where the worst win-rate it will have is 0 big blinds per 100. So, it just isn’t going to lose. Its goal isn’t to win the most. It’s not an exploitative bot that tries to figure out things about its opponent. Its goal is to play so Game Theory Optimal that it can’t lose. So, its win-rate is whatever its opponent is from GTO. It wouldn’t surprise me that if we do this challenge again in two years we see a true statistical tie. Whoever wins it is going to be so, so close.

BP: For the last question I want to return to the subject of the action at the online nosebleeds. As a professional poker player do you ever look back at the old days, like in 2009 when there were $1 million heads-up pots online, and long for those days to return?

JL: Man, I do. I have played online poker since like 2005 and I often look back at how things were then. In 2009 we were talking about 2005 being the good old days. And then in 2012 we were talking about 2009 being the good old days. It just keeps happening. In three years we are going to be talking about how 2015 was the good old days. Hopefully not. I think it is optimistic to think that things will ever return to that level, the action and the amount of people playing. But I do think there is room for improvement. The legalization of online poker in the United States will bring back a lot of energy to the game, especially if California has it. It is a huge market, a high population, a rich population, a gambling population. I think it would be great for California to have Internet poker again. You would see a lot of action return. I remain hopeful.

Related Articles

Poker Bot Crushes Humans In Historic Match

Poker Bot Has Victory Sealed Going Into Final Day

Poker Pros Now Losing By One Million Chips

Most Viewed Articles