Skip to content
SVX

Cross-venue arb

testnet · paper

The cross-venue arbitrage strategy — buy mispriced Polymarket binaries when Predict's SVI surface disagrees by > threshold. Positions are naked binaries under hard per-trade clips; the Hyperliquid delta hedge is disabled post-audit (HL now supplies the realized-vol feed). Independent from the standalone HL IV-RV divergence strategy.

Operator: 0x73f4…d53c3 · paper signals on testnet
Fork the repo to run your own

On testnet the same loop runs in paper mode against live signals — useful for watching strategy mechanics without the mainnet wallets in flight. For real-money PnL flip the network toggle to mainnet.

  • Edge captured chart below is the strategy's report card — entry edge identified vs realised return. Positive slope = the math finds real edge.
  • Mid-life exits (closed trade rows where poly_settlement_outcome = early_exit) sold the spread back before UMA settled — caught compression instead of waiting hours.
  • The HL hedge (now disabled) stripped directional BTC exposure: short when we bought Yes, long when we bought No. The 2026-07 audit found it sized at the wrong expiry and it was shut off — historical rows still show poly leg + hedge leg + funding.
Combined PnL
0.00
0 settled
PnL 24h
0.00
paper-mode 24h
Win rate
0%
0/0
Avg edge captured
—
Predict − Polymarket at exec
In flight
0
paper open count

Edge captured — math validation

For each closed trade: entry edge (Predict − Polymarket probability gap at execution) vs realized return on cost. The yellow line is the least-squares fit — tilting up to the right means deeper edges identified deliver larger realized returns.

No mainnet fills closed yet.
The bot stamps an edgeAtExec on every fill — this chart compares it to realised return once a trade closes (mid-life exit or UMA settlement).

Closed positions

0 settled trades

No closed poly-arb trades yet.
CONNECTINGtestnet…