24h NBA Game Structure Trader

> This is a template.

> The default signal detects structural inconsistencies across correlated NBA game markets -- remix it with additional sports, league-specific heuristics, or live odds feeds.

> The skill handles all the plumbing (market discovery, game grouping, cross-market checks, trade execution, safeguards). Your agent provides the alpha.

Strategy Overview

Polymarket lists multiple correlated markets for each NBA game:

• Moneyline: "Clippers vs. Pacers" = 79.7%
• 1H Moneyline: "Clippers vs. Pacers: 1H Moneyline" = 50.5%
• Full-game O/U: "Clippers vs. Pacers: O/U 235.5" = 56.7%, O/U 236.5 = 55.2%, O/U 237.5 = 52%
• 1H O/U: "Clippers vs. Pacers: 1H O/U 114.5" = 50.5%
• Spreads: "Spread: Clippers (-8.5)" = 41%, "1H Spread: Clippers (-5.5)" = 50.5%

Retail trades each market as an isolated bet. But together, these markets form a structural web that must be internally consistent.

This skill reconstructs the full game structure and finds where it is mathematically broken.

The Edge: NBA Game Structure Arbitrage

Inconsistency Type 1: Moneyline vs 1H Moneyline

If the full-game moneyline says a team is an 80% favorite, the 1H moneyline cannot be a coin-flip. The expected 1H advantage is dampened but must exist:

If ML_full > 70%, then ML_1h > 50% + (ML_full - 50%) * dampening

When the 1H moneyline diverges too far from what the full-game moneyline implies, the 1H market is mispriced.

Inconsistency Type 2: O/U Monotonicity

Within the same game, the probability of going OVER must decrease as the line increases:

P(O/U 235.5 OVER) >= P(O/U 236.5 OVER) >= P(O/U 237.5 OVER)

If a higher line is priced above a lower line, the curve is broken -- pure structural arbitrage.

Inconsistency Type 3: 1H O/U vs Full-Game O/U

At the same line value, the full-game total always exceeds the 1H total, so:

P(Full O/U X OVER) >= P(1H O/U X OVER)

If a 1H O/U market is priced higher than the equivalent full-game O/U market, the relationship is violated.

Inconsistency Type 4: Spread vs Moneyline Direction

The spread favorite (negative line) must be the moneyline favorite. If the spread says Clippers (-8.5) but the moneyline says Pacers are favored, the markets contradict each other.

Why This Works

1. Retail trades in silos -- most users view each market independently and do not cross-reference the full game structure
2. No market maker enforcement -- unlike sportsbooks, there is no central entity maintaining consistency across related markets
3. Mathematical, not opinion -- the violations are provable inconsistencies in the implied game model
4. NBA-specific density -- NBA games generate 5-15+ correlated markets per game, creating more surface area for inconsistencies
5. Rapid line movement -- NBA spreads and totals move on injury news, creating temporary cross-market divergences

Signal Logic

1. Discover all NBA-related markets via keyword and team name search
2. Parse each question: extract teams, market type (moneyline/spread/O/U), scope (full/1H), line value
3. Group ALL markets for the same game into a GameGroup
4. For each game with 2+ markets:
• Check moneyline vs 1H moneyline consistency
• Check O/U monotonicity within full-game and 1H lines
• Check 1H O/U vs full-game O/U at matching lines
• Check spread direction vs moneyline direction
5. Rank inconsistencies by magnitude
6. Trade only inconsistencies that also pass threshold gates (YES_THRESHOLD / NO_THRESHOLD)
7. Size by conviction, not flat amount

Safety & Execution Mode

The skill defaults to paper trading (venue="sim"). Real trades only with --live flag.

| Scenario | Mode | Financial risk |

|---|---|---|

| python trader.py | Paper (sim) | None |

| Cron / automaton | Paper (sim) | None |

| python trader.py --live | Live (polymarket) | Real USDC |

autostart: false and cron: null mean nothing runs automatically until configured in Simmer UI.

Required Credentials

| Variable | Required | Notes |

|---|---|---|

| SIMMER_API_KEY | Yes | Trading authority. Treat as a high-value credential. |

Tunables (Risk Parameters)

All declared as tunables in clawhub.json and adjustable from the Simmer UI.

| Variable | Default | Purpose |

|---|---|---|

| SIMMER_MAX_POSITION | 40 | Max USDC per trade at full conviction |

| SIMMER_MIN_TRADE | 5 | Floor for any trade |

| SIMMER_MIN_VOLUME | 5000 | Min market volume filter (USD) |

| SIMMER_MAX_SPREAD | 0.08 | Max bid-ask spread |

| SIMMER_MIN_DAYS | 0 | Min days until resolution (0 = allow same-day) |

| SIMMER_MAX_POSITIONS | 8 | Max concurrent open positions |

| SIMMER_YES_THRESHOLD | 0.38 | Buy YES only if market probability <= this |

| SIMMER_NO_THRESHOLD | 0.62 | Sell NO only if market probability >= this |

| SIMMER_MIN_INCONSISTENCY | 0.05 | Min structural inconsistency magnitude to trigger a trade |

Edge Thesis

Traditional sportsbooks have professional line-setters who enforce consistency across all markets for the same game. Polymarket has no such mechanism -- each market (moneyline, spread, O/U, 1H variants) is priced by its own order book with its own liquidity pool. This creates systematic micro-inconsistencies in the implied game model, especially when:

• Injury news moves the moneyline but not the 1H moneyline
• O/U lines are added at new values without re-pricing existing ones
• Large directional flow on spreads does not propagate to correlated markets
• 1H markets diverge from full-game markets during low-liquidity hours

This skill treats the full game structure as a consistency web and trades the repair.

Dependency

simmer-sdk by Simmer Markets (SpartanLabsXyz)

• PyPI: https://pypi.org/project/simmer-sdk/
• GitHub: https://github.com/SpartanLabsXyz/simmer-sdk