Games as Representative Practice

Representative practice means the information in training matches the information in the fight. If the live problem includes a resisting partner, time pressure, and a score that can flip, the training activity should too.

Isolated drilling can teach a shape. It rarely teaches when that shape is available. Games keep perception and action coupled: the passer reads frames; the retainer reads pressure; both adapt mid-exchange.

Use games when you want decision-making. Use a short isolated rep only to feel a mechanic an athlete cannot yet perceive inside the game — then put them straight back into the representative problem.

How it fitsNaming the canonThe positional games library is the training substrate of the constraints-led approach. Each game is a designed problem — a starting position, a goal for each side, a constraint, a reset rule — that a grappler solves against a resisting partner. Read this way, the library is the part of the site you take to the mat.

Every game in the library is written as a set of task constraints:

A starting position — where both players begin, which fixes the problem.

A win condition for each side — the goal that defines success, and the lever a coach adjusts to make the game ask for more or less.

A partner asymmetry — symmetric, top- or bottom-advantaged, or role-rotating, which sets who is under pressure.

A round length and reset rule — how long the problem runs and what restarts it.

The invariants it pressures — the mechanical truths the game is built to surface.

These are the constraints from the approach, made concrete. The Leg Entanglement Full System game, for example, removes the scaffolding of the earlier entanglement games and rotates roles, which forces a grappler to manage entries, transitions, and finishes continuously and exposes a one-entry game fast.

Representative learning design holds that practice transfers to competition when it carries the same information and decisions the real task does. The games are built to that standard. They start from live, contested positions; the partner resists; the grappler reads what the position affords and acts. A grappler running the back-retention game is solving the same perceptual problem they will face in a round, in a narrower frame.

The win conditions are where the design lives. A goal of “hold a confirmed dominant entanglement for thirty continuous seconds” teaches that positional quality is itself worth something, alongside the finish. A goal phrased as a pin past the knees in a time limit trains the passer to value speed and the guard player to value the position-before-submission order of recovery. Change the win condition and the game teaches something else.

A win condition should name the outcome a player has to produce, not the technique they should use to produce it. “Force the tap by compressing the carotids” and “complete the rear naked choke” point at the same finish, but they train differently. The first leaves every strangle the back affords in play — the rear naked choke, a short choke, a switch to a different strangle — and lets the player take the one the position gives them. The second names a single solution and quietly removes the rest from view.

So the games here phrase their win conditions as functions: force the tap by compressing the carotids, force the tap with rotational load on the knee line, isolate an arm and extend the elbow line, pin past the knee line, take the back. Those are invariants — the mechanical outcomes that have to hold — and they are the right language for a constraint, because they fix what must happen and leave how to the player. The position a game starts from already shapes which class of solution is available, so the win condition does not need to prescribe the move on top of it. Naming a technique there would narrow the affordances the game exists to open.

The library is filterable by hub, ability, and asymmetry. A practical loop: