Poker Bots: Are They More Intelligent Than You?

Poker Bots: Are They More Intelligent Than You?

Poker is a game of skill with a healthy dose of luck thrown in. That’s if you’re pretty good. For most casual online players it’s a game with plenty of upswings and downswings and a long-term Return On Investment (ROI) that will do nicely for a few extra Christmas presents.

But just as tracking software has given players the edge over their opponents, so bots are threatening to take things up a notch or three.

Where tracking software was once the ultimate tool for online players, the bots have taken things to a whole new level. And the software is getting smarter.

Better software and data availability has led to a rise in predictive methods for everything from sportsbetting to the stock market. Unsurprisingly that has now stretched to a few hands of No Limit Hold’em.

But can software ever replicate the psychology of a Will Kassouf needle or a large dose of Phil Ivey aggression? Can you really get rich playing online with a bot? Let’s take a closer look at poker bots.

How Poker Bots Work

Futurama poker bot Bender vs human Fry
Image: fox.com

Poker bots are pieces of software used on online poker sites (usually) by players who can’t beat the game normally. Bots are computers that use mathematics and player knowledge to attempt to beat real-life human players.

Bots run in the background and as standalone programs alongside the client you’re running. Primarily, they track the hands that have been played and make observations that the human eye may not be able to see.

It’s like an HUD but it actually plays for you too. Running a bot is pretty easy, as long as they are permitted by the poker site you’re playing at. They’re just downloadable programs and they come with clear instruction manuals.

There are plenty of bots for sale on the open market. They can tackle cash, MTTs or Omaha but none guarantee a winning run over the long term.

That hasn’t stopped some sites cracking down on them though. PokerStars was investigating a suspected high-profile case of bots in 2016’s Turbo Championship of Online Poker. The player was eventually found to be human.

Later it also emerged that a group of Russian players had used bots to win over $1.5 million on PLO cash games on the site. The world’s largest poker room later made changes to ban all third-party software from its tables.

What Makes Bots Tick?

Initially, bots were quite simplistic, and even if you weren’t aware you’re up against a bot, an average player would have no problem beating them. However, as time goes by and technology advances, things are starting to change.

Modern-day bots are capable of using stats and complex calculations performed by so-called solvers. Solvers play a huge role in today’s games as they represent the ultimate learning tool for the players.

But they can be abused as well.

Old robot toy
Image: Shutterstock

A poker bot can “remember” virtually countless numbers of pre-calculated scenarios, meaning it can produce a correct response in many different situations.

With that kind of base knowledge, bots are becoming a severe threat.

As Hold’em is getting closer to becoming a “solved” game, these bots are gaining more and more ground over human players.

And, while some sites are doing a fair bit to try and tackle the issue, not all bot play can be easily detected.

Bot-like software could also be used by a human player. They could have the software running alongside the poker client (or even on an entirely different computer, making any kind of detection virtually impossible), feeding them the information.

So, while you’re not technically up against a bot, you’re effectively playing against the one. It may be a human being clicking the buttons, but it is the bot that’s telling them exactly what to do.

In this scenario, the human player is nothing more than an auto-clicker, but it’s much harder to trace and detect using automated methods.

The New Generation Of Bot

Researchers are busy developing new bots that are effectively better than people. Not only do they play optimum strategy in particular situations, but they can also bet hard and, most importantly, bluff just like humans.

In recent years, the Game Theory Optimum (GTO) strategy has come a long way. This is an approach to the game that seeks the best possible plays in every scenario regardless of other players’ actions.

So, if you can play perfect GTO, you don’t have to worry about what other players are doing – you will make a profit.

Of course, if you know a bit about Hold’em or PLO, you are aware that these are very complex games, so remembering what the GTO play is for every possible situation is virtually impossible for a human.

Not for a bot, though.

Bots don’t have problems with memory, and don’t get confused. As long as they run on a computer with enough storage space and processing power, they can reach a perfect decision for every scenario.

And this is dangerous.

Image: piosolver.com

One comforting fact is that GTO is still not perfect – Hold’em hasn’t been completely solved yet – meaning bots don’t have access to a perfect and unbeatable strategy.

That said, things are changing fast, and there are better and better poker-playing software solutions out there.

Man vs Machines

The game of poker has been recognized as an excellent testing ground for AI.

Unlike chess, which used to enjoy this title, poker, especially Hold’em, has many aspects that are hard to calculate. You can fold the best hand if you feel like it, and you can move all in to win a tiny pot.

Of course, these aren’t necessarily your best options, but you are free to resort to them if you so choose.

It is this random factor that appeals to the experts. The number of choices a player has at any given point in time is virtually limitless. Therefore, developing an AI that can handle the mess and come out on top has implications that go well beyond the green felt.

Such an AI could be used for complex simulations and come up with solutions for situations with highly uncertain outcomes. It’s quite clear why such software could be highly useful in military, medicine, strategic planning, and much more.

But for those running poker bots, the goal is simple: making money off of unsuspecting players.

The latest bots have shown to be quite efficient vs. humans, and although the best of the best are still managing to hold their own, it seems like it’s getting harder for humans to keep up with the technology.

The Increased Threat Over The Years

In 2016, A bot known as Baby Tartanian 8 won the Annual Computer Poker Competition. It was believed to be the ‘best poker bot’ ever invented.

While other bots have been able to crack Limit Hold’em, with its strict and fairly basic betting patterns, Baby Tartanian 8 has been proven to beat No Limit.

No Limit has so many more facets to the game than Limit: the all-in raise, betting for value, the sheer range of moves on each turn card.

That’s why the BT8 bot created such a buzz. Its algorithms have been set up to deal with the “unknowns” of Hold’em (i.e. you don’t know your opponents’ cards).

In 2017, an Artificial Intelligence dubbed poker bot Libratus was pitched against four top poker players. Each of them played heads up No-Limit Hold’em against the AI.

When the challenge was announced, the whole world was convinced human players would claim an easy victory based on similar matchups in the past.

Image: Carnegie Mellon University

However, after 20 days of play, it turned out Libratus won more than $1.7 million worth of chips in total. A handsome and unexpected victory.

It was the first time AI was able to defeat top-ranking human players head-to-head. For the team from Carnegie Mellon University that developed the AI, this was a great victory.

For the poker community, though, it was concerning.

However, Libratus was only triumphant in one-on-one play. Many believed that science was still far away from creating an AI that could actually overcome human players in a multi-player setting, so the threat wasn’t as significant.

But then, in 2019, a new poker bot came about, called Pluribus.

Unlike its predecessors, Pluribus was designed to handle scenarios with multiple players, and it was tested in a six-max setting. Once again, researchers gathered some of the brightest poker minds to try and challenge their AI.

The AI was tested in two different scenarios. In one, it was one AI against five human opponents, and in another, five AI players against each other and one human.

In both cases, Pluribus was able to gain a significant advantage, winning around $5 per hand (which is enormous) when playing against five human opponents.

With this latest AI breakthrough, scientists have proven they’re capable of creating a poker bot that can hold its own and even come out ahead in complex 6-max scenarios. A bot such as this would represent a big challenge if it were to get into the wrong hands.

For the time being, an average bot user probably doesn’t have access to a high-tech poker-playing AI such as Pluribus. But worringly the technology is out there, and so is the proof of concept.

It’s now clear that it’s possible to create a poker bot that can beat humans in Texas Hold’em, both heads-up and multi-handed.

Beating The Bots: Teaming Up

The new generation of bots can handle many more hand scenarios than the original bots, which just had the capacity to call, fold, or go all-in. The downside is they are expensive and difficult to develop, and getting hold of one is also tricky. Aside from that, most poker sites have banned them entirely.

Carl Sampson is a trainer and poker pro who is well-known in the gambling world. For him, having a strong team of players around you is a great way of beating the software.

Cheats and colluders have existed in online poker since the first games appeared. Individual players are at a distinct disadvantage because they work and play alone, and can’t share information on suspicious behavior.

With a team of players helping each other spot the cheats – or in this case, bots – Sampson believes the bots can be weeded out.

Just like financial traders work together to improve their results, players need to start working as a collective to beat the game. Indeed, with the online game getting tougher and tougher, it’s vital that players share information and work together to plug the leaks.

There have been quite a few examples of this kind of cooperation between the players, which led to unearthing bot rings and putting a stop to them.

Even though poker is a game where it’s every player for themselves, in these types of situations, it’s vital for the community to come together to protect the integrity of the game.

Spotting bots on your own is much harder, especially on larger sites where there is significant player traffic. People running these types of bot networks have learned from their mistakes, so they make sure to regularly change the names of their bots and come up with new “players,” making tracking harder.

However, these bots usually share the same central knowledge-base, and this is probably their biggest vulnerability. With good cooperation between the players and several people capable of reading and comparing stats at a high level, these bots can still be stopped.

Spot The Bot: 6 Signs You’re Up Against A Bot

How to identify a poker bot
Image: YouTube

Identical Times For Decisions: Is that opponent taking 32 seconds on each and every decision? If so, that coincidence might be just be a bot’s programming.

Identical Bet Sizes: Just because the bets are the same time and time again, it doesn’t mean your opponent is a multi-tabling god who just doesn’t have time to talk to you.

Check The HUD Lines: Lots of poker bots feed off tracking software stats and make decisions accordingly. But it swings both ways: if you can use a HUD on your site, do your own research on them. If you find several players at the table with identical pre-flop bet stats over a long period of time, you could be under attack from more than one bot.

If you notice a player is making the same fold to your 6bb raise, make it a regular move. The bot won’t back off in fear, remember. It will just fold. Try a re-raise with any two cards for several hands in a row and see if it does the trick.

Zero Chat: Despite repeated attempts to talk to your opponent in the chatbox, you don’t get a reply.

Call Security: Most major poker sites have an integrity unit whose job is to weed out bots and put a stop to the accounts.

Make Lots Of Adjustments: If you suspect your opponent is a bot, you can make slight and regular adjustments to your play to throw them off the scent.

All these tips will definitely help you deal with poker bots. However, you should be aware that the coding of these automated players is getting better every day. Bot creators have started to add various random elements to their programs to make it much harder to track patterns.

For example, they’ll give their bots decision timers that aren’t fixed but fluctuate between point A and point B. This means the same bot can take 3 seconds to make one decision and 10+ seconds for the very next one.

The same goes for their tendencies and frequencies. Advanced bots have a way to adjust their play according to the current situation and “reads” (stats) they have on other players.

All this being said, poker bots are (thankfully) still far from perfect, and you can successfully track some of their tendencies and find a way to work around them. It may be harder than five or 10 years ago, and you may need to readjust your approach frequently, but it’s still doable.

Will Poker Bots Kill Online Poker As We Know It?

The biggest concern many players have is whether these bots will completely ruin the game. It’s hard to make predictions for what might happen in the future, of course, but it’s safe to say we’re not there just yet.

Most poker bots are focused on cash game tables. MTT players don’t have to deal with many bots, and the reason for this may be the fact that the AI behind them isn’t geared to handle the ever-changing nature of poker tournaments.

Cash game players have it harder.

It’s no secret that there are bots hanging around, even at very respectable stakes at certain poker rooms. But the operators are mainly taking a stance against these poker robots, and many of them will likely do their best to protect actual players.

Finally, the game of poker may change and will almost certainly change in the future.

Whether it’ll include bots or not is anyone’s guess, but even if poker rooms allow it, I have very little doubt in my mind that human players will come up with very effective strategies to beat the new competition.

After all, poker has always been about adapting to changes, so there is no reason to think bots will somehow kill the game we all love and enjoy so much.