Take a focus into real games

Each practice is a focus you play with a hypothesis — not a drill. Practice 2 is your priority.

Play with a plan — not reps

Every time you play you’re doing one of two things, never “grinding reps”:

• Find new theory — coaching, watching players, thinking/talking about the game. (Needed when you’re stuck.)
• Refine theory — deliberate play + VOD review. Checking does my idea actually make sense.

Training mode is not practice. Use it only for tech / specific execution (your warmup). Reads, spacing, decisions are learned by playing with a plan.

The arc — Four Stages of Mastery

Unconscious incompetence (build awareness) → conscious incompetence (theory + field tests) → conscious competence (grind hours to translate theory → autopilot) → unconscious competence. Recognition timeline: late → slowly faster → on time → before it happens.

The session loop — scientific method

1. Hypothesis + plan. The idea you’re testing AND how you’ll play it.
2. Deliberate pilot on ONE thing. Attention is limited — spend it only on your focus, to define your flowchart. Trust autopilot for the rest.
3. Gather info — make it measurable. Watch how they react to your movement:
   • Reactive = they wait and respond → create your situation (your setup beats their reaction).
   • Preemptive = they act before you’re set, whiff-punish → change strategy to punish how they preempt.
4. Review on two fronts (ASAP): Idea → Knowledge / Read. Execution → Valuation / Execution (→ training mode) / Piloting. Context check: did my assumption lack context?
5. Adjust the hypothesis and loop. Once the idea is right, grind it to autopilot.

If the info is very impactful, play a little LESS and let it sink in.

Ramses: habits betray intentions — a dashback/airdodge betrays fear, not analysis. Name the feeling. Never accept “the game just sucks” — ask how you influenced it.

• Mon P2 (Defend Space) · Tue P1 (1-2 Punch) · Wed P3 (Advantage)
• Thu P2 again (your #1) · Fri whichever front the review flagged worst
• Sat tournament + VOD review of your worst set (backwards from the kill) · Sun free play, no focus — note what leaks

One focus, one hypothesis per session. If a lesson hit hard, play less and let it sink in.

Players to watch — Ramses’ list: Sonix, Spargo, Gluto, Light. Roy: Light (advantage/conversions), Goblin (positioning after hits), Kola (spacing). Watch a winning VOD backwards from the kill.

The framework (true every day)

• 3 Layers: Strategy (where to be) → Situation (is it good yet?) → Interaction (play it out).
• Create → Capitalize. Strategy creates the situation; the Interaction capitalizes.
• RPS has layers: L0 nair → L1 their counter → L2 your answer → L3 back to nair.
• Recognition, not problem-solving. You name what they did — the answer is pre-loaded. The goal is a better autopilot, not less autopilot.
• Initiative = you move first so they react; reacting is the only thing you can read.

Practice 1 — validates

Your own words, Ramses confirmed: “interrupt → change how I use forward movement, usually stationary; contest → mix aggression with landing fairs, dash jabs, dash tilts, dash shields.” Feedback: jab more, dtilt/side-B less; up your speed of decision-making.

Practice 2 — validates

THIS is your “hold the line.” Ramses: “wall it, less dashback.” You give up space out of fear. Can’t set tempo (vs G&W) → play around theirs (“I cannot oppress you, but if you mess up I’ll wreck you”). Empty movement doesn’t work for Roy → build around spacing for nair.

Practice 3 — validates

Advantage is a series of 50/50s run by the pressure ladder (anti-reversal → feint → direct pressure). Roy’s edge: guaranteed combos front-load reward, so lead with the kill threat → their escapes shrink → work the filter. At ledge/corner be in position, don’t commit — let them reversal into your jab/d-tilt.

In one breath — play with a hypothesis, deliberate pilot on one thing, review the two fronts, adjust, grind to autopilot. Take your space and hold it (P2), run the 1-2 punch once you have it (P1), threaten the kill and position in advantage (P3).

1. Baja blast (jab → side B): always full hop it (rage/weight vary the height). Low % = jump→side B; higher % (tumble) = quick low run-up first or it sour-spots. Cue: dash right as Roy lowers his sword after jab. Kill input = F → F → UP → F (kills way earlier than F-F-F-F, ~60–70% at ledge on lights).
2. Jab → F-Smash (“jeff smash”): needs the shutter-step — hold dash a few frames after jab, cancel into F-Smash → sweet-spots. Raw F-Smash sour-spots. (Look 55–70%.)
3. Delayed uairs: in every uair string, delay the follow-up uair — that’s the difference between true and a drop.

Daily touch-up (5 min): D-throw → Nair → Fair · D-throw → Jab → SH rising Uair → Uairs · Jab → Nair → Bair · + one kill confirm (rotate: baja blast, jab→bair, Nair 1→bair, Landing Uair→bair). In-match, one thought: what percent → the confirm is pre-decided.

“VOD review is comparing reality to expectations. If you don’t have expectations, it becomes hard to structure.”— Ramses

Your old review had reality (what happened) but no expectation tied to what you were practicing — so it turned into hunting random errors. The fix is one added step: before you play, write the day’s hypothesis. That hypothesis IS your expectation. Now the review has something to check against.

Scientific method, exactly: observation → hypothesis → test → revisit. No hypothesis = going by feel.

The hypothesis to set, what to pause on, what a gap looks like that day.

• Watch a winning Roy VOD backwards. Start at the stock/kill and rewind. How did Light/Kola set it up? What pattern did they notice, and when did they first notice it?
• Improvement = making better decisions — a decision is (1) how many variables you weigh + (2) how you value them. When you review a decision, ask: did I miss a variable, or value one wrong?

Every practice day ends the same way: check the day’s hypothesis, classify each surprise, route it to training mode or a new hypothesis. That’s the loop — no more random mistakes.

• A hit is the result of the setups beforehand — not a lucky guess. Fix the setup, not the last button.
• Create → Capitalize. Move to get a position (not to hit), then play the RPS. Movement that exists for the hit comes out sloppy.
• Acting second is top-level. React to the info on the screen — “look at the screen.” Replace “what do I want to do?” with “what does this situation ask me to do?”
• 🗡 Initiative + reaction, merged. Take initiative to a good position → they feel pressured → they commit → you react. That’s why position makes you faster. (Your “playing fast = being in position.”)
• Recognition, not problem-solving. The first to recognize a situation and its likely results, wins.
• Play instinctual — the goal is a better autopilot. “The master practices one swing 1000 times, then forgets it all.”
• Orthodox vs unorthodox: optimal play wins the 100-battle average; unorthodox counters the specific person in front of you. Relying only on “optimal” makes winning luck.

• Burst range = space where you throw a hitbox they can’t react to (dash attack, rising aerials). To beat it you must NOT press a button — wait/shield/jump/dashback.
• Contested ground RPS: occupy it (walk/dash in — rewarding, risky) or control it (park a hitbox without committing). Occupy < Control < Waiting < Occupy.
• 🗡 Undershoot vs overshoot: undershoot = space between you (claims space, defensive — your “hold the line”). Overshoot = space behind them (counters retreating) — rarely correct.
• Feints are the most missing tool. A clear dash stimulus makes them react so you can punish. Don’t tie a strategy to one movement.
• 🗡 Accuracy > reward. Structure neutral around what’s most likely to hit, not what follows the hit. (Kola d-tilt’d MKLeo at ~30 instead of fishing a kill.)
• 🗡 The move AFTER their strongest tool. Roy Nair / Cloud Bair / ROB Nair are strong rising but unsafe — stay flexible, move out of range, confirm when they jump poorly.
• Punish “safe” tools through pressure. A safe bair on shield gives them frame advantage — make them pick again; at disadvantage they use unsafe low-startup tools → wait on those and punish.
• Winning neutral via stage control — push them to the corner instead of risking a hard punish. Less reward, far less risk.

• Advantage is a series of 50/50s. Reversal (high risk) vs safe option (low risk). Cover the low-risk options until they overcommit, then punish. Top players hold advantage forever this way.
• Three priorities, in order: cover their return to neutral → bait a reaction → only then set up a hit. Don’t convert advantage straight into a hit.
• Make them feel helpless — take away the option they’re comfortable with.
• Identify the moment your advantage ends. Overextension (reversal after reversal) is the #1 mid-level mistake.

• 🗡 Disarm their swing FIRST — parry it (→ punish / shield-drop jump nair), or space below them. (Dashback fsmash on a nair-from-ledge can end a stock.)
• The landing RPS: attack / air-dodge / double-jump. Cover two to force the double jump, then punish the fast-fall airdodge.
• Fall-speed sync: stay in your burst range but outside theirs, fall together — they keep guessing while you stay safe.

• High recovery = usually reactable → your go-to, cover first; even if they go low they’re ledge-trapped.
• Low recovery = hard, low reward → the read; commit (match height, read DJ/up-B).
• 🗡 Mid recovery (Fox illusion, Palu warp, Chrom directional airdodge) = wildcard; 2-frame or commit.
• 🗡 Roy tools: roll to ledge → d-air; 2-frame d-air; d-air the ledge hang. Don’t ledge-trump a ton.

• The gamestate is a slider. Climb back toward neutral one low-risk step at a time. Don’t reversal-gamble unless they overcommit.
• Limited air resources (one double jump, one airdodge). Spend them on low-risk situations — double jump when they’ll actually swing and miss.
• Fastfall is control — the one mechanic giving you both space and time control in the air.
• Air game = REACTION, not autopilot. Spacing bad → disengage; they shield → empty land; they jump → consider airdodge.
• Empty landings are your best mixup — remove the parry, keep all options open.

• Play to WIN, not to avoid losing. Playing “not to lose” sacrifices control.
• Frustration is fine — channel it. Anger is emotional investment. Let a loss gnaw, then turn it into work.
• Mushin / be present. Don’t dwell on past grievances or future kills mid-match.
• Win consistently, not once. Set goals you control (“UpB OoS every landing aerial”), not “win.”
• Tournament energy: reserve mental energy for new info / reads. Big breakfast, small snacks, warm up ~15 min.

• Improvement = better decisions = more variables weighed + better valuation. Add variables early (big jumps); refine valuation later.
• The 5 Aspects → where to practice: Execution → training mode · Knowledge → research vods · Yomi → play vs others · Valuation → research + experiment · Piloting → work on yourself.
• Have a friendlies plan — a specific goal per session. Focus lasts ~5 min at a time; guided 2 hours beats aimless 4.
• Plateauing = no new theory to apply. Be the student before the master.

• P1 (1-2 Punch) ← create→capitalize, burst range, contested-ground RPS, feints, accuracy > reward.
• P2 (Defend Your Space) ← undershooting, small-adjustment defense, punish safe tools, the move-after-their-strongest-tool.
• P3 (Advantage) ← the 50/50 slider, the 3 priorities, disarm the swing.
• Not yet a practice: disadvantage/air-game, edgeguard chunking, juggling — reference for now; add one only if VOD review says it’s costing you stocks.