Followup to my training video. I mentioned implicit learning in the video and recent blog posts. The most common way to implicitly learn is to play lots of hands. Over time you develop a feel for frequencies and betting patterns.But I believe there are some issues with this approach, which makes it painstaking and very long.

Recent event bias. Our minds are biased to remember recent events. Running good or bad, or running into the top/bottom of people's ranges can drastically skew reality. This is where math is very useful. Hand combos and equity is objective. That theoretical basis gives a foundation for understanding the correct plays. Of course we often give weights to hand combos, and recent event bias effects those weights.

Confirmation bias. We see things that confirm our existing beliefs and emotions. When running bad you might make ridiculous folds, or even calldown when beat to confirm that you're just running bad/it was a cooler situation. When getting run over (perhaps opponent is also running hot), you can bluff or calldown in terrible spots since you want confirmation that they can't always have it. But in reality most players are extremely exploitable, don't bluff with a balanced frequency and there's a reason you keep running strong hands when playing badly.

Fragmented learning. What I mean is that we learn slowly and nonlinearly. We learn in little chunks. This is the nature of poker learning. Learning involves periods of rapid growth and periods of plateauing. How this relates to implicit learning is that you can learn something (say a particular line/spot) but it doesn't occur often enough, so there is no re-enforcement. This is why people often repeat mistakes. They learn, but without re-enforcement they make the same mistake (particularly in more rare spots).

Hindsight bias. Can call this the Daliman syndrome. It's much easier to be correct after seeing results. But even in 2009 people are results oriented when it comes to poker. This is related to confirmation bias and recent event bias. But again, a fundamental understanding of theory, frequencies, hand combos, equity, etc can minimize this.

Finally, how can we improve on implicit learning? One way which is very easy with today's tools is the massive hand review.

Frequency analysis exercise.

• Pick a chunk of 30-50k hands. 10k is probably too little and 100k is too much, but somewhere in between works.
• Using HEM (can probably do in PT3 too), filter the hands by a very specific scenerio. For example all hands when you are IP, the flop was checked HU, and villain bets the turn. This is a common spot where people don't really understand what to do.
• Analyze the hands to get a feel for people's ranges, frequencies, and how you played the hand.
• Think about ways to improve your play in that spot. If you're losing in that spot what can you do differently? Tighten up with calldowns? Bluff raise more? Take snabs because people have weak ranges there?
• The point is to look at large chunks of similar situations. You can avoid some of the pitfalls mentioned above by looking at repeated situations.
• Note: This is also a great way to simply fix leaks. Look at specific spots you have trouble with, figure out typical opponent ranges, and determine appropriate adjustments.