Post-match testing

Observation	Useful interpretation	Do not conclude
High possession, few shots	Sterile build-up or insufficient penetration	Possession never matters
Many shots, few on target	Weak shot quality/finishing; SOS may inflate volume	Formation worked because shot count was high
Few shots conceded, several goals	Finishing/GK variance or high-quality chances	Defensive system definitely failed
Low possession, many chances	Direct/counter plan may be functioning	Low possession means tactical defeat
Strong possession, shots, SOT; narrow loss	Finishing variance likely meaningful	Abandon the tactic immediately
Weak possession and shots; repeated losses	Systemic mismatch more plausible	One slider is definitely the cause
Statistically dominated win	Efficient or fortunate result	Tactic is now proven
Red card before the swing	Match is contaminated evidence	Formation alone caused the result

OSM does not publish expected goals. Shots and shots on target are proxies, not complete chance-quality measures.

1. Red cards, camps, and other major distortions
2. Squad/context mismatch
3. Shots-on-target pattern
4. Shot pattern
5. Possession/territory
6. Final score

The score determines competition points but is noisy evidence of tactical causality.

Hold constant as much as possible:

• same starting XI;
• same correct positions;
• similar fitness and morale;
• same opponent strength bracket;
• same opponent formation/game plan;
• same venue context;
• same marking/offside unless tested;
• similar referee for tackling tests;
• no camp difference;
• one major variable changed at a time.

Matched matches	Interpretation
1–3	Anecdote
4–9	Early directional signal
10–19	Useful screening evidence
20–30	Moderate practical evidence
30–50	Stronger decision evidence
50+	Stable personal library if no major update invalidates it

Points per match = (3 × wins + draws) / matches
Goal difference per match = (goals for − goals against) / matches
Shot difference = shots for − shots against
SOT difference = shots on target for − shots on target against
Possession difference = possession for − possession against

Compute results with and without red-card/camp-distorted matches.

Record:

Category	Fields
Context	Date, competition, home/away, objective, referee
Strength	XI averages, line averages, key absences, fitness/morale
Opponent	Exact formation, game plan, marking, offside, known camp
Your tactic	Formation, plan, all line instructions, P/S/T, marking, offside, tackling
Bonuses	Correct positions, nationality, Pitch, camp, Secret Training
Output	Score, possession, shots, SOT, corners, cards, player ratings
Interpretation	Distortion flag, one hypothesis, confidence change

Do not keep scaling by adding wider Markdown tables. Use linked, append-only records:

Table	Primary contents
Saves / clubs	Save ID, club, competition, objective, season phase, active/archive status
Matches	Observation ID, date, type, opponent, venue, result, statistics, distortion flags
Tactics	Match ID, exact formation/plan, line instructions, P/S/T, marking, offside, tackling, source status
Lineups	Match ID, player, exact position, starter/bench, specialist roles
Squad snapshots	Timestamped rating, age, value, fitness, morale, availability, line averages
Transfers/listings	Append-only purchase, list, reprice, sale, cycle exposure, realized profit, slot-days
Training	Player, timestamp, progression, event, facility context, current-season role

Required safeguards:

• Use stable IDs and timestamps.
• Never overwrite an earlier listing price or squad snapshot.
• Store planned, confirmed used, and inferred as different states.
• Preserve source filenames/checksums for every load-bearing observation.
• Merge duplicates; do not count the same match twice because both PDF and screenshot exist.
• Keep archived states out of current recommendations unless explicitly comparing history.
• Generate aggregate reports only after validating required fields and context similarity.

Evidence	Maximum confidence
One result	25/100
Fewer than 5 matched matches	40/100
Fewer than 10	55/100
10–19	70/100
20–30 plus external support	85/100
Large, current, consistent sample	90–95/100
Absolute certainty	Never use

---