AI Skill Report Card

Managing FPL Squad

A-84·Aug 22, 2026·Source: Web
14 / 15

Given the current squad, budget, and upcoming gameweek, produce:

  1. Transfer recommendation (in/out, cost, points hit if any)
  2. Starting XI + bench order (by formation)
  3. Captain + vice-captain pick
  4. Confidence reasoning backed by numbers (xG, xA, fixture difficulty, minutes risk)

Example output format:

GW15 Recommendation
Transfer: OUT Player A (£5.5m) → IN Player B (£5.8m) [-0.3m, no hit]
Formation: 3-4-3
Captain: Haaland (vs bottom-table team, home, xG 2.1, 90% start probability)
Vice: Salah (backup if Haaland rotates)
Bench order: GK2, Def3, Mid5, Fwd2
Recommendation
Add a concrete worked example with real numeric xPts calculations for 2-3 players to show the formula in action, not just narrative reasoning
15 / 15

Progress checklist for each gameweek decision:

  • Step 1: Pull fixture difficulty for next 1, 3, and 6 gameweeks per team
  • Step 2: Check player status (injury, suspension, rotation risk, minutes trend)
  • Step 3: Gather form signals (last 4-5 games: goals, assists, xG, xA, bonus points, ICT index)
  • Step 4: Convert everything into expected points (xPts) per player
  • Step 5: Check budget, free transfers, and chip availability
  • Step 6: Build optimal 15-man squad under constraints (£100m, max 3 per club, valid formation)
  • Step 7: Select starting XI + bench by highest xPts and start probability
  • Step 8: Fixture clash check — verify no opposing-team pairings that cancel out (e.g., don't captain a striker while also relying on the opposing keeper/defense for clean sheet points that gameweek)
  • Step 9: Pick captain/vice by highest floor+ceiling combination, not just highest average
  • Step 10: Log the decision and result to refine the model next week

Step-by-step detail

1. Fixture analysis

  • Rate each fixture 1-5 (FDR) using opponent's defensive/attacking strength, not just table position
  • Weight home/away — home advantage worth ~0.2-0.3 goals expected
  • Look for fixture swings: teams entering a run of 3+ easy games (double/triple up on their assets)

2. Player & team condition

  • Read team news, press conferences, injury tables (physio reports), and 2-3 trusted FPL blogs/statistical sites per gameweek
  • Track minutes played trend over last 3 games — a "returning from injury" player is a risk even if talented
  • Note managerial rotation patterns (cup games, European fixtures midweek increase rotation risk)

3. Quantify into probability

  • Convert qualitative signals into: start probability (%), expected minutes, xG90, xA90, clean sheet probability
  • Expected points formula (simplified): xPts = (start_prob) × [(xG90 × goal_pts) + (xA90 × assist_pts) + (CS_prob × CS_pts) + bonus_avg + appearance_pts]
  • Adjust for fixture difficulty multiplier (0.8x for hard away, 1.2x for easy home)

4. Clash/correlation check

  • Never have your captain's opposing defense/keeper also in your squad expecting a clean sheet same week
  • Avoid over-reliance on one fixture (if you own 5 players from one match, a red card or blowout tanks your week)
  • Diversify captaincy risk if two nailed options play at the same time (can't watch both — pick by xPts ceiling)

5. Squad construction rules

  • £100m budget, 2 GK, 5 DEF, 5 MID, 3 FWD, max 3 players per real club
  • Balance "enablers" (cheap nailed starters, £4.0-4.5m) with 2-3 premium assets (£10m+)
  • Keep 0.5-1.0m in the bank as flexibility unless a clear upgrade exists
  • Bench should still have playing players (bench boost potential + injury cover)

6. Chip timing

  • Wildcard: use before fixture swings or after squad value/injuries force major restructure (typically GW4-8 and GW20-30)
  • Bench Boost: use when bench players have double gameweeks or easy fixtures
  • Triple Captain: use on premium player's double gameweek or best single fixture of season
  • Free Hit: use for blank gameweeks or one-off fixture exploitation
Recommendation
Include an example of a bad/suboptimal decision (e.g., a wrong captain pick with a losing outcome) to contrast with the good examples, since all three examples currently show correct reasoning only
15 / 20

Example 1: Input: "Haaland vs bottom-of-table team at home, Salah away at a top-6 defense — who do I captain?" Output: "Captain Haaland. Home fixture vs weak defense (FDR 2) gives higher floor and ceiling — expected 2.1 xG vs a defense conceding 2.3 goals/game average. Salah's fixture is FDR 4 away, historically scores 40% fewer points in these matchups. Vice-captain Salah as safety net in case of Haaland rotation news."

Example 2: Input: "I have 2 free transfers and £1.2m in the bank. My midfielder is injured (hamstring, 2-3 week estimate)." Output: "Use 1 transfer now: sell injured midfielder, buy in-form replacement in similar price range with good next-3 fixtures. Bank the 2nd free transfer for next week rather than making a speculative second move — avoids point hits later and keeps flexibility if news changes (e.g., a rotation risk emerges elsewhere)."

Example 3: Input: "Should I captain my striker even though the opposing keeper is also in my squad?" Output: "No — this is a direct clash. If your striker scores, fine, but if the match ends 0-0 or your striker is blanked while the keeper you own concedes, you gain nothing from either. Re-evaluate: either bench the opposing keeper this week or pick a different captain without this conflict."

Recommendation
Provide a quick-reference table or template for the FDR rating scale and price tiers instead of embedding all constraints in prose, to speed up scanning during live gameweek decisions
  • Prioritize start probability over raw talent — a bench player who plays 90 minutes beats a star who plays 20
  • Use expected points over 3-6 gameweeks, not just next week, when deciding transfers (avoid short-termism)
  • Track price rises/falls — transfer a day early if a player is about to rise, saves budget long-term
  • Differentiate from the "template" (highly-owned players) only when data supports it — for mini-leagues, sometimes matching the template is the safer path to not falling behind
  • Re-verify team news right up to the deadline (final training session news moves lineups)
  • Don't chase last week's highest scorer blindly — check if the underlying fixture/form is repeatable or a one-off
  • Don't take points hits (-4) unless the gain in xPts over the following gameweeks clearly exceeds the cost
  • Don't ignore fixture clashes between your own captain and clean-sheet assets — this quietly costs points every season
  • Don't overreact to a single bad or good gameweek — use rolling multi-game averages, not one data point
  • Don't forget the 3-per-club rule when double/triple-fixture teams tempt over-investment
  • Don't hold onto sentimental picks (players you like) once the data no longer supports them
0
Grade A-AI Skill Framework
Scorecard
Criteria Breakdown
Quick Start
14/15
Workflow
15/15
Examples
15/20
Completeness
18/20
Format
15/15
Conciseness
13/15