Sunday, December 15, 2013

R is for Reading. And Reduction.

Long time readers of this blog know that a few years ago I put together a little system called REDi, which I frequently use to analyze what specific actions one should take in a hand of poker. It's basically a method that breaks down the process of hand evaluation into four discrete steps. The letters R, E, D, and I represent the four discrete steps of Read, Estimate, Decide, and Implement, respectively. Not surprisingly, these four steps also line up with the same basic OODA method (Observe, Orient/Evaluate, Decide, Act) that our military uses for battlefield decision making (poker is war, after all):

The REDi process is simple but effective. We Read our opponents and game situation to determine what probable hands they hold. We then Estimate the strength of our own hand against that range, determine how much if any fold equity we have, and calculate our degree of pot commitment. Once we've done these two things, we Decide on the line we should take in the hand (e.g., bet for value or as a bluff, check for pot control, call on a draw, etc.). Finally, we Implement that decision in a manner that maximizes our expected value (e.g., we determine the largest bet we can make and get called by a worse hand when on a value line, or the smallest bet that can get a better hand to fold when we're bluffing).

For a simple example, let's say we were seated at a full-ring NL game, and a tight break-even ABC player in early position opens for a 3x raise. Effective stacks are 100bb deep. The action folds to us on the button. We have a pair of threes. The blinds to our left don't look interested in the hand and are probably going to fold. What do we do? Answer: employ REDi:
  1. Read. Our opponent is normally tight. He's also a straightforward player, and probably understands position. He's in early position and opening at a full ring table where stacks are fairly deep. Our read is that he has a relatively strong hand. Call it AQ+ and 88+. 
  2. Estimate. Against a strong range like this, our pot equity with a pair of treys is pretty low-- probably around 30%. Similarly, our fold equity is pretty tiny; in fact, he's probably only folding out the bottom of his range if raised. If we call, the stack-to-pot ratio is going to be around 13, so we're not going to be pot committed. With these stacks, we also theoretically have a maximum of ~100/3 = 33:1 implied odds.
  3. Decide. With IOs like this and this type of opponent, this looks like a good situation to set mine. We'll fold on the flop if we don't hit and get bet into. Our line is therefore a one-street draw to a set.
  4. Act. Call.
  5. Repeat 1-4 on Flop, Turn, and River as required.
Got it? Good. Because in today's post, I want to delve a little more deeply into the first and most important of these four REDi steps: Read. Why? Because you cannot win long-term at poker if you're not accurately reading the hand ranges of your opponents, that's why. Let me rephrase for clarity:

Question: How can you possibly decide whether you're ahead (and therefore should take a value line) or behind (and therefore consider taking a bluff or draw line) unless you have accurately read your opponent?
Answer: You can't.

That's right, you can't. You can't estimate your pot or fold equity. You can't make any kind of reasonable line decision. You can't choose an appropriate bet size. In other words, reading, so to speak, is fundamental to making good poker decisions.

As you move up into higher stakes (and therefore play against better opponents) you also need to accurately predict not only the hand range of your opponents, but also what lines they are trying to take. Do they think they have the best hand and are trying to extract value from us, or do they think they're behind and are they trying to bluff us, or are they on a draw, and so on…

So how do we make these kinds of reads? To a beginner, it almost seems like magic when a professional like Negreanu calls out what their opponent has before he or she turns their cards over. I still remember when I was just starting out in poker, watching my own poker coach tell me that a villain he was playing online probably had AQ...and then being amazed when that villain turned over exactly the predicted hand. I was in awe.

Now that I'm a relatively advanced player, this type of hand reading doesn't seem like such mystical magic anymore. In fact, I now know that there is a logical and systematic process to hand reading that can be learned by pretty much anyone willing to do the work. Is this process accurate 100% of the time? No, of course not. But it's nevertheless a pretty damn good approach, and by using it correctly, the majority of time you will be able to put your opponent on a fairly narrow range of hands.

So how does this process work? To answer, the first thing we have to recognize is that hand reading is a reductive process. We start with gross, big picture assumptions, and then as we gather and analyze information, we reduce, or narrow the possible hands our opponent can hold.

I discussed this a little in a previous post via the graphic that showed how we refine our reads by adding in successively more detail and information. This technique is known in engineering and management circles as "progressive elaboration", and it's perfectly applicable to poker, too:

Uh, okay Bug. You showed us this before in a previous post. What's the big deal? Well, for starters, we can take the chart and divide it into two parts that helps further simplify the reading process:

The left side of the chart is where we do basic "typecasting" of the villains we face. This includes gross stereotyping (which I wrote about here), followed by pegging the villain on a "PATL," or Passive-Aggressive vs. Tight-Loose chart, followed by determining the villain's level of thought (i.e., how deeply they think about a poker hand, and how deceptive they are in nature). Said another way, this is a basic categorization of the opponents we face, and it will actually get us a long way toward assigning an eventual hand range to the opp. We aren't actually trying to put our opponent on a very specific range of hands at this stage, but instead are mostly thinking in broader terms about the types of hands he or she plays, as well as a general assessment of the style of play they employ when playing poker.

This first typecasting step is crucial for two reasons: a) it helps us determine basic strategies to employ immediately against the opponent (such as whether we should try stealing their blinds or not if it folds to us on the button, or whether continuation betting will be effective or not on a disconnected board); and b) it helps us perform the second step to hand reading, where we will really start narrowing down the range of possible hands....

...which brings us to the second, right-hand side of the chart, which is known as "range reduction." This is the time in which we put our opponent on a specific hand range and line, and then narrow it as the hand progresses and we learn more. We start with the typecasting information we've gathered on our opponent, and then we factor in specific things such as our board texture, the opponent's statistics, combinatorics and dead cards, bet sizing, villain second order stats, tells, meta-game factors like tilt and wins, leveling, and so on...

...and by the time we're at the river we can often narrow the possible hands our opponent holds to just a few cards. We can, in effect, predict that proverbial AQ and amaze our friends and family.

And make a lot of money, too. :-)

To repeat, hand reading is a reductive process; we start wide, and then take away from, or reduce, the range of cards our opponent could have as we add in information we collect. Our method is to reduce and simplify, reduce and simplify, reduce and....

In my next post on this subject, I'll discuss a bit more about the typecasting process.

All-in for now...

No comments:

Post a Comment