strategy · s7 · prediction markets
adaptive venue arbitrage
on binary outcome markets.
S7 trades binary outcome markets on Polymarket (Polygon CLOB, USDC-settled, EIP-712 signed) and Kalshi (CFTC-regulated US event exchange, USD-settled, RSA-PSS signed). Five engines harvest different edges: latency, market-making rebates, cross-venue divergence, intra-market dislocations, and news-sentiment fair value. Markets settle on resolution, not on continuous price exit — so cycles are weekly and the gate is inventory-aware.
cycle RX01 · produced 2026-05-15 12:00 UTC · 16 configs · 5 engines · 2 venues
⚠ synthetic data · live trading not authorized
S7 RX01 runs on a synthetic market provider and a synthetic sentiment generator. All param fingerprints carry the | SYNTHETIC_SENTIMENT_RX01 suffix. No config is promotable to live trading from this cycle. Engines E2 (MM_REBATE) and E4 (INTRA_ARB) are SD-06 FATAL by design on synthetic data — no order acks exist to validate fills.
A single risk kernel over three payoff types (event perps / defined-risk options / binary events), built gate-first: assumes no edge until one survives an adversarial test. As documented, no configuration has yet cleared that test — it ships nothing on an unconfirmed signal.
source · S7 Abstract, §10, §12The §10 result is the headline risk: no confirmed edge; ETH/XRP sub-threshold, BTC/SOL overfit, native instrument too young to validate. Nothing deploys until a config clears the full gate.
source · S7 Abstract, §10, §12Calibration criteria are gates, not outcomes; no realized-performance claim is made.
engines & regime · methodology
A single payoff-agnostic risk kernel spanning three payoff types (event perps, defined-risk options, binary events), built gate-first — it assumes no edge until a configuration survives an adversarial, multiple-testing-corrected gate. Not an engine roster.
Signal → confidence gate → hybrid-AI gate → payoff-agnostic sizing under a per-trade clamp → entry, with a soft entry-block and a hard-flatten. The kernel is gate-first: as documented in §10, no configuration has yet cleared the multiple-testing-corrected gate, so the system ships nothing on an unconfirmed signal.
Calibration criteria are applied as gates, not outcomes — a configuration either clears the gate or it does not ship.
source · S7 §3, §6, §10The architecture's integrity does not manufacture an edge. Section 10 is the central limitation: as of this writing S7 has no confirmed directional, carry, or relative-value edge that clears the program's multiple-testing-corrected gate, and the whitepaper claims none.
source · S7 §12- ✓available
Whitepaper
Q7_PRISM_Requirements_S7_PredictionMarkets_v1.0.pdf in OneDrive 15 NEW S7
- ○pending
Standing requirements
Standing requirements doc not yet authored
- ◐partial
Live heatmap
RX01 stub fixture — real heatmap when Calibration emits
- ○pending
Calibration matrix
Calibration_Matrix_S7.xlsx not yet built (per standing rule #138)
- ○pending
Change / defect / repair log
Not yet built — S7 first cycle pending
five engines.
full methodology →Latency Arbitrage
Acts on Polymarket CLOB updates faster than slower participants
Market-Making Rebate Harvest
Quotes both sides of binary outcome books to capture maker rebates
Cross-Venue Divergence
Captures price divergence between Polymarket and Kalshi on linked markets
Intra-Market YES/NO Dislocation
Arbitrages YES + NO prices when their sum diverges from 1.00
News-Sentiment Fair Value
Bets on event markets where news sentiment diverges from implied probability
sixteen configs · last cycle
heterogeneous · not a grid| id | market family | venue | tf | engines | trades | inv·loss | pnl | status | |
|---|---|---|---|---|---|---|---|---|---|
| C01 | BTC up/down | Polymarket | 5M | E1 E2 E4 | 312 | 1.8% | +$428.40 | passing | detail → |
| C02 | BTC up/down | Polymarket | 15M | E1 E2 E4 | 198 | 5.2% | −$42.10 | criticalSD-13 | detail → |
| C03 | BTC up/down | Polymarket | 60M | E2 E4 | 0 | — | — | dropped | detail → |
| C04 | BTC up/down | Polymarket | 1D | E2 E5 | 47 | 6.1% | −$18.85 | criticalSD-13 | detail → |
| C05 | ETH up/down | Polymarket | 5M | E1 E2 E4 | 287 | 1.9% | +$389.20 | passing | detail → |
| C06 | ETH up/down | Polymarket | 15M | E1 E2 E4 | 174 | 5.8% | −$31.55 | criticalSD-13 | detail → |
| C07 | ETH up/down | Polymarket | 60M | E2 E4 | 0 | — | — | dropped | detail → |
| C08 | ETH up/down | Polymarket | 1D | E2 E5 | 39 | 6.7% | −$22.40 | criticalSD-13 | detail → |
| C09 | SOL up/down | Polymarket | 5M | E1 E2 | 156 | 5.4% | −$28.60 | criticalSD-13 | detail → |
| C10 | SOL up/down | Polymarket | 15M | E1 E2 | 112 | 5.9% | −$24.85 | criticalSD-13 | detail → |
| C11 | FOMC rate decision | Kalshi | EVENT | E2 E5 | 0 | — | — | held | detail → |
| C12 | CPI surprise | Kalshi | EVENT | E2 E5 | 0 | — | — | held | detail → |
| C13 | NFP print | Kalshi | EVENT | E2 E5 | 0 | — | — | held | detail → |
| C14 | US election | Polymarket + Kalshi | EVENT | E3 E5 | 0 | — | — | held | detail → |
| C15 | Geopolitics | Polymarket | EVENT | E5 | 22 | 7.2% | −$11.30 | criticalSD-13 | detail → |
| C16 | NFL game | Polymarket + Kalshi | EVENT | E3 | 0 | — | — | held | detail → |
pass criteria · weekly gate
Different gate shape from S1-S6 — adds inventory-resolution loss and signature-reject ceilings unique to prediction markets.