Friday, February 5, 2016


(This document is likely to expand over time)

Before you read this, go ahead and do this exercise.
If you can't hold yourself accountable to do a simple writing exercise in order to better identify your problems then you will probably continue to marginally suck forever.

- - -

What is Practice For?

For the purpose of this discussion, let’s assume that success in melee is based on two things:
1) the decisions that you make and
2) your ability to execute those decisions.
Both are critically important in that if one suffers then the other is useless. Now, because of melee’s incredible pace, many of the decisions that you make will be habitual and your execution must be automatic. You simply do not have the time to think double-check your thinking or to obsess over inputs. These assumptions make the following model perfect for us.
When practicing, our goal is to identify a problem and systematically work from unconscious incompetence (wow, I didn’t know that my fastfalls are always later than they could be) through conscious incompetence (ah, I missed the fastfall timing again) and conscious competence (ok, this time I’m going to fast fall before hitlag) into unconscious competence (*unconsciously fastfalls at the correct time every time*).

So in order to correct a problem you have to
1) identify a problem (oh, my fastfalls are later than hbox’s are, why is that?)
2) understand the problem (you can shff at frame 19, additionally you cannot input a fastfall during hitlag. Understanding the problem is usually the bulk of the problem.)
3) pragmatically fix the problem (sit in frame counter and practice ff timings in common scenarios)

It is not overly complex or difficult, so there is no reason to treat it as such. It's once again just a simple matter of solving problems efficiently and effectively.

Practice is synonymous with teaching yourself good habits.
Individual practice is for teaching your hands effective execution habits.
Practice with opponents is for teaching your mind effective strategy habits.

Remember, as you play and practice you are literally carving paths into your brain in order that your brain can follow those paths with ease later. The more often you go over these paths the more profound they are and the longer they last (the brain is an organic material and frequently lets things fall away that aren't used regularly). This makes practicing extremely powerful and thus extremely important. However, practice should be explicitly designed to solve problems. Generally speaking, if you are not solving a problem then it is not good practice. In fact, practicing without a plan in mind is the best possible way to carve bad habits into your brain that will come back to haunt you later. Being mindful of what you’re focusing on when can help you keep tabs on your decision-making so that when you face a strong opponent you are well prepared. Most of your battle happens well before you sit down.


Faltering techskill during tournament matches is more often than not unrelated to nerves or stress but instead because of a change in the in-game context. Freestyling around the stage by yourself has a completely different set of demands than having accurate inputs in any given moment in a tournament match. The small changes that come with a moving target or with acting out of hitstun etc can be significant. Maybe your thumb isn’t used to starting from the angle that you’re suddenly forced to start from. It’s unsurprising that you would drop that input. Your goal is to minimize these instances.

Diminishing returns

I lost the literature that I had on this but you will learn a skill much faster practicing 30 minutes twice a day than you will practicing 1 hour once a day. Even 15 minutes twice a day is more than sufficient for rapid improvement in the kinds of simple tasks that techskill involves.
It is also good to note that a change in setting, etc contributes in the same manner.


When you are practicing something, you want to design a methodology that is simple, challenging, effective, and gives you instant and accurate feedback.
At this moment this is usually writing down the set of corresponding numbers in Frame Counter and drilling it for a predetermined amount of time (6 minutes works well for me for simple things (shielddrops) and 10 minutes for complex things (like techchases))*.

You should constantly tweak your routine in order to best prepare for your gameplan (remember, individual practice is just to teach your hands the exact good habits that you need. As soon as you can perform up to your standard it is no longer necessary or effective). I recommend keeping a list (as I always do lol) of tasks with times attached, running through them, and reviewing according to your current need.

- - -

Improving Strategy/Playing to Learn

So far I’ve written mostly about individual practice in order to prepare execution. It falls into the Preparation phase of Improved Drastic Improvement. The following information falls into the Analysis phase. Because it makes a nice dichotomy I'm going to put it here.

It is important to distinguish between two types of learning. The first, detailed above, is execution-focused. It is defined by Playing to Win and internalizing what you already know to the point of unconscious competence. It raises your floor.

The second type is expansion-focused. It is defined by Playing to Learn and investing in losses and experimentation in order to push the boundaries of your strengths. You try to construct a deeper, better gameplan while you organize and build on your knowledge of how to act in different MUs. It raises your ceiling.

These two models are both very important but they have different goals and different processes. This is why I separate them into different phases and time-periods. Confusing the two can get you into a hopeless conceptual mess.

Playing to Learn during Analysis is necessarily less clear-cut than Playing to Win during Preparation or Performance, but that does not mean that it has to be less deliberate.
When analyzing strategy and pushing your ceiling, define the problem you mean to solve for. Dissect footage for patterns and check this against your knowledge of frame data and hitboxes.

No comments:

Post a Comment