Dislocation board
| Stable | Binance vs USDT | Kraken vs USD | Gate.io vs USDT | On-chain ref mid | Best net |
|---|---|---|---|---|---|
| Waiting for quotes | |||||
Flagged cycles
Waiting for quotes
Cost model
Replay, last 30 days
Waiting for candles
| Stable | Venue | Hours | Above cost | Cycles | Deployed | Avg hourly vol | Best net | Sim PnL |
|---|
Method and limits
The cycle. A discount cycle buys a stable below par at the venue ask and redeems it for $1.00 at the issuer. A premium cycle mints at $1.00 and sells above par at the venue bid. Gross bps is the distance from par; net subtracts taker fee per CEX leg, the issuer fee, funding for the settlement lag, and gas. USDT-quoted venues cost a second leg because the proceeds still have to become dollars, and the conversion uses the live Kraken USDT/USD book.
What is real. Every quote on the board is the current best bid and ask as returned by each exchange's public API, refreshed every 30 seconds through a server-side proxy. The visible size is top-of-book only. On-chain mids come from DeFiLlama and are a reference, not a fill.
What is simulated. The replay assumes the hourly close was hittable at the stated half-spread, caps size by participation and by the per-cycle cap, allocates one pool of capital chronologically across all pairs, and locks each cycle's capital for the lag. It does not model queue position, redemption-window limits, issuer capacity, bank hours, or a venue refusing to sell below par once a large buyer shows up. A persistent discount that survives many well-capitalised market makers is usually priced redemption friction, not free money.
What this is not. A paper monitor. No capital, no custody, no orders. Path is a protocol-intelligence layer; this surface reads the market, it does not trade it.