Submit your questions for heads-up expert Will Tipton and win his new book

He will also be giving away five ebook copies of his new book Expert Heads Up No Limit Hold'em: v. 2: Strategies for Multiple Streets. It is available to buy in Europe from the end of May.

All you have to do to have your question answered by Will is to post it in the thread below. We will pick out the best questions for Will and five of which will be picked out to win a copy of his new book.

Will is a poker player, author and PhD at Cornell University. He has been playing poker since 2007 and is a regular in the high stakes heads-up games online.

Buy Expert Heads Up No Limit Hold'em: v. 2: Strategies for Multiple Streets

Read a sample of Will's new book

A traditional way to characterize opponents is in terms of two variables: aggression and tightness. Generally speaking, we call a player aggressive if he tends to use a lot of bets and raises relative to his use of checks and calls, and the opposite of aggressive is passive. As for the second variable, a player is loose if he plays a lot of hands in many spots and tight if he does not. This generally starts pre-flop, where a loose player folds fewer hands at his first decision point, and continues throughout the hand. Since there are two choices for each of the two variables, there are four total combinations possible: loose and aggressive (LAG), loose and passive (LP), tight and aggressive (TAG), and tight and passive (TP).

These categories give us a way to characterize and model an opponent’s play. Once we observe an opponent for some time, we can categorize him, and knowing a player’s type gives us some predictive capabilities. It gives us some idea of Villain’s ranges in various spots, and thus how he might respond to our actions. However, when it comes down to it, this system is very limited. It simply cannot convey much information about players. How could it – there are only four categories. If that is all we can learn about an opponent, it is better than nothing, but we need to get in the habit of trying to do better to find success in modern heads-up games.

This system is likely a carryover from other hold ’em formats such as full-ring. When playing a 9-handed table, and especially when playing multiple such tables and potentially dozens of opponents simultaneously online, it is challenging to remember much about all of them. Thus, it is reasonable to try to keep just the two bits of information in the four-category system for each opponent. Additionally, when playing long-handed as opposed to heads-up, we generally have much less and much lower quality data to work with for any particular opponent. Players fold many more hands pre-flop, so we are likely to play out hands with any particular player very rarely. We can watch how he plays against other people, but that might not be how he plays against us, so those reads are potentially tainted. Worse still, whenever other players join or leave the table, our opponent’s strategy could be affected, and that could change the ranges he plays against us.

In heads-up play, we have Villain all to ourselves. We get untainted information that we can use to build a detailed understanding of his strategy, since we need not worry that it will change constantly due to the table composition. Thus, the kind of characterization we can do in a HU format is not only deeper; it is fundamentally different than what can be achieved in other formats. The dynamics at heads-up tables are in some sense simpler than when we have three or more opponents. However, they are simpler for our opponents, too, so to beat the game, we have to delve deeper – beyond the traditional categorization system.

The traditional way may be better than nothing, but we should approach our matches planning to develop a much more detailed understanding of our opponent’s play as quickly as possible. The four-category system can be rather confining, so I prefer to avoid it entirely if possible. For example, suppose we are playing a LAG. What does that really mean? Maybe it means that Villain has wider value-betting ranges and correspondingly larger bluffing ranges. If so, we should simply say that. A more specific description might tell us that he has wide value and bluff ranges on early streets, but he gives up a lot on the river. Drilling down further still, perhaps he barrels a lot on early streets, but gives up his bluffs on specific types of rivers – those where draws miss, or those where draws get there, or those where the board pairs, etc.

In order to exploit an opponent, we do not want to figure out generally how he plays – we want to figure out exactly how he plays. When characterizing an opponent heads-up, we have the opportunity to delve into the sorts of considerations listed in the previous paragraph. The goal, again, is to figure out, i.e., model, Villain’s play in a way that has predictive power. In other words, we want to use what we have seen in the past to predict how he will play in future.

When it comes to this sort of learning and pattern recognition, mathematical and computational approaches are possible. However, simple techniques that can compete with the human brain are rare. So, in practice, opponent modelling for use at the tables takes focus, logic, and experience. We must focus intently on Villain’s play. We need experience to identify which of the myriad pieces of information are important, and logic to see what Villain’s play in one spot says about his strategy in others. However, there are some big ideas we can use to lend structure to this creative process.