Proof Run · Hedge Book
Day one of the live hedge book — on real Bitcoin.
2026-06-08 session · settled at the close
Sidepit is a flywheel: Alpha attracts liquidity, and liquidity attracts Alpha. A market maker that quotes tight, deep and reliable pulls traders in; that flow lets the maker capture spread — and, once it can lay the risk off, scale without taking a directional bet. Today we put the missing half of that loop live: the hedge book.
Okay fills, every time
No flash-crash fills. A deep auction that bounds the worst case.
Tape vs tape. Sidepit's worst fill was 28 bps; Hyperliquid's worst wick was 188 bps — a 6.7× smaller tail, tighter at every percentile. And Sidepit takers pay no fees (Hyperliquid charges takers 4.5 bps). Cheaper to cross, every time.
Everyone trades with Alpha
Sidepit has no fees and no house rake — it's a fair, zero-sum market. On day one, Jay's prop desk traded against Alpha and did well; Alpha was the OTC counterparty on the other side; the breakout bot was roughly flat. Marked-to-market at the close:
| Account | Result (sats) | Position |
|---|---|---|
| JayProp — counterparty | +119,500 | +15 |
| Alpha — the desk | −112,500 | −21 |
| Taker — breakout bot | −7,000 | +6 |
| Total | 0 | — |
On Sidepit alone it nets to exactly zero — JayProp's win came straight across the book, no fees skimmed. But Alpha doesn't keep that −$71, because it hedged.
Alpha hedged — and came out ahead
Alpha laid its directional risk onto Hyperliquid. The hedge book's settled 24 hours ran +$144 PnL on $190k of volume, 3.1% max drawdown. So the maker's actual day:
| Alpha's day | P&L |
|---|---|
| Sidepit (the −21 it took on) | −112,500 sats (≈ −$71) |
| Hedge book — Hyperliquid | +$144 |
| Alpha, net | ≈ +$73 |
So the real day-one scoreboard: JayProp +119,500 sats, Alpha net ≈ +$73 (with the hedge), Taker −7,000 sats. The market maker and its biggest counterparty both made money. Alpha took the other side of all the flow, hedged it, stayed risk-neutral — and still came out ahead. That's the engine that lets it scale.
We shipped, and we fixed it live
Day one wasn't sterile — and that's the point. The exchange ran out of disk and went offline for 99 minutes mid-session; we caught a crossing bug in the hedge bot the same morning. Both were found and fixed the same day, the book recovered, and the back half ran exactly as designed — passive, never chasing. Breaking and fixing live is the muscle that matters before scale.
The minute-by-minute panel — Sidepit vs Hyperliquid, with the outage marked: