Fs Worldcup Knockout

Trade World Cup 2026 knockout-stage player fantasy-score markets on propSPACE (functionspace.dev). Each market is a scalar distribution over a player's aggregate fantasy points for a match. Builds a Gaussian belief from xG/xA/passing priors and trades when it diverges from consensus. Use when a user wants to trade FS WC knockout competition markets with an evidence-backed distribution strategy.

WC Knockout — FunctionSpace

By FunctionSpace — built with Simmer's agent trading infrastructure. Play at propSPACE. Play-money only. Real USDC prizes administered by FunctionSpace.

Trade World Cup 2026 knockout-stage player fantasy-score markets. Your $1,000 play-money competes for a real prize pool (≥$5,000 USDC) matched by FunctionSpace. Round of 32 kicks off June 28, 2026.

🧪 Dry-run by default. Pass --live to place propSPACE play-money positions. This is a mutating competition-engine action, but it is not Simmer $SIM, Polymarket, Kalshi, or real-money trading.

How It Works

propSPACE markets are continuous scalar belief markets — the outcome is a player's aggregate fantasy score for the match, not a binary yes/no. You trade a probability distribution over the score range. The engine uses a scoring-rule mechanism: positions pay out more when the consensus shifts toward your belief.

Competition format: each round you pick 1 FWD + 1 MID + 1 DEF. Those 3 selections are locked until round settlement. The skill ranks all open markets per position by edge (your belief mean vs consensus mean) and buys the best player in each slot.

Each run:

1. Authenticates with propSPACE (passwordless — auto-creates account on first run)
2. Discovers open WC knockout player markets (~100 per round)
3. For each market: matches the player, estimates their expected fantasy score, builds a Gaussian belief
4. Groups markets by position (FWD / MID / DEF from market metadata)
5. Picks the highest-edge player per position (best 1 FWD + 1 MID + 1 DEF)
6. Buys those 3 play-money market positions when --live is passed

Strategy

Expected fantasy score = f(xG, xA, passes, FIFA scoring weights) × sentiment adjustment.

The belief is a Gaussian centered at the expected score with spread derived from goal-scoring variance (Poisson: spread ≈ sqrt(xG) × points_per_goal). Expert sentiment from WC 2026 fantasy articles shifts the mean ±15% and tightens the spread when consensus among experts is strong.

Best pick = highest |our_belief_mean - consensus_mean| within the position slot.

⚠️ Points formula is a placeholder. POINTS_PER_GOAL, POINTS_PER_ASSIST, POINTS_PER_PASS in main.py use FPL-style defaults until FunctionSpace confirms their FIFA scoring weights. Update the four POINTS_* constants before going live.

⚠️ Engine URL placeholder. FS_BASE_URL defaults to the dev engine. Set it to the competition engine URL from FunctionSpace before going live.

Setup

1. Set your username and engine URL

   bash

   export FS_USERNAME=simmer_agent_1
   export FS_BASE_URL=<competition engine URL from FunctionSpace>
   

2. Enrich sentiment (optional but improves pick quality)

   bash

   export BRAVE_API_KEY=your_key
   python3 scripts/enrich_from_web.py --round "round of 32"
   

3. Dry run — see which 3 players the skill would pick

   bash

   python3 main.py
   

4. Browse markets to see what's live with position labels

   bash

   python3 main.py --list-markets
   

5. Inspect a specific market

   bash

   python3 main.py --inspect 261
   

6. Go live — executes 1 FWD + 1 MID + 1 DEF propSPACE play-money trade

   bash

   python3 main.py --live
   

Quick Commands

bash

python3 main.py                    # dry run: pick best FWD/MID/DEF
python3 main.py --live             # execute 3 picks
python3 main.py --list-markets     # list open markets with position labels
python3 main.py --market 261       # target one market
python3 main.py --inspect 261      # show consensus vs belief
python3 main.py --all-markets      # trade all markets above edge threshold

Configuration

Variable	Default	Description
FS_USERNAME	(required)	Agent username
FS_BASE_URL	https://fs-engine-api-mech-v0-4.onrender.com	FunctionSpace competition engine URL
FS_PASSWORD	simmer-wc-bot	Account password; set a stronger value for production bots
FS_ACCESS_CODE	(none)	Competition access code
FS_MAX_COLLATERAL	333	Max play-money per trade (× 3 picks ≈ 999 total)
FS_MIN_EDGE	0.05	Min edge for --all-markets mode only
FS_MARKET_FILTER	(none)	Market title keyword filter

Player Data

The current production path uses each market's FunctionSpace metadata.expectedPts / line as the base expected fantasy score and then optionally applies sentiment enrichment. If data/player_data.json exists, it can add player aliases and sentiment fields; if it is absent, markets still run using the FS line as-is.

Before Going Live — Required from FunctionSpace

1. Scoring/variance confirmation — the base expected score now comes from FunctionSpace market metadata (expectedPts / line), while position-aware spread uses the constants in main.py. Confirm those spread assumptions against the official scoring model.
2. Competition engine URL — update FS_BASE_URL if FunctionSpace moves from the mech-v0-4 engine.
3. Market metadata schema — confirm the field name for player position in market metadata (checked as position, positionType, player_position in that order). Correct it in _market_position() if needed.
4. Access code — if signup requires a partner code, set FS_ACCESS_CODE.
5. Market range — lower_bound / upper_bound for a typical player score. Confirms the Gaussian fits the market range at runtime via --inspect <id>.