Track record

Our result covers the 6 real locks that have ended: what the hedge paid or closed for and what the terms collected from the holder, less what the fills cost and what the terms paid the holder.

Across 1,537 settled events the same positions returned $123,255.95 with the lock and $128,090.35 without, marked at the official result. The lock cost $4,834.40, 3.8% of the outcome without the lock; 3.0% was quoted up front. Quoted is the expectation at the market's odds; realized is what this sample did.

On 676 of those 1,537 (44%) the pick lost and the floor paid. Those are the positions that would otherwise have ended at nothing.

Our side netted $642.26 on them against $642.25 of take booked. The hedge pays the floor, the slice above the cap pays the hedge; nothing we earn depends on who won, and nothing we earn is a trader's loss.

Of 2,168 events the engine priced, 259 (12%) were refused rather than offered: too rich to offer 209, credit too small to offer 47, no book could fund it 3. A refusal is written down like an offer, so the record is the whole board.

214 positions were left before the game at live prices and came out at $14,834.56 against $15,601.93 without the lock; we carried no price risk on the exit.

On 1537 settled events our P&L is $642.26 and the take booked is $642.25: a gap of $0.01 (hedge units round up, never down). The hedge layer carried no outcome risk.

In the last 30 days locks put 69,221 contracts on public books across 1078 positions: $34,333.00 of flow against $161,709.61 of stake locked in. Since 2026-09-12 UTC: 168,096 contracts across 1935 positions, $79,684.68 of flow. Simulated here, like the positions; live, this is order flow onto the books that price the event, placed the second the user taps.

On the 931 settled, our P&L is $433.01 against $433.00 of take booked; the positions were credited $3,374.19 at lock in. 153 left before the game at live prices; see the early closes below.

775 of 1537 positions carry the venue's rules as they read when they opened; the rest were opened before we kept them. At settlement we read them again: 775 unchanged.

The lock paid the floor on 676 of 1537 settled events (44%). Across all of them it cost $4,834.40, 3.8% of what the same positions paid without the lock; 3.0% was quoted up front. Quoted is the expectation at the market's odds; realized is what this sample did.

214 positions left before the lock. Each exit was a new trade at the live books that second: the position's contracts closed at the bids, our other side closed at its bids, and the locked terms settled at what both fetched. They came out at $14,834.56 against $15,601.93 had they never been locked in (-$767.37). Crossing both books cost $1,173.76 against the mid (3.56% of stake). Our result on the way out is the take booked, $96.69: the position pays live prices on both legs, so we carry no price risk on the exit either. 12 other scheduled exits found no depth on a leg and ran to settlement instead.

63 tickets reached the last leg with a priced lock, on average $1.19 credited at lock in on a ticket worth $48.83 at fair. On the 61 settled, locking paid $2,909.38 against $3,024.25 for cashing out and $2,773.00 for holding; the lock beat the cash-out on 38, and holding beat both on 38.

On 61 settled locked tickets our P&L is $12.34 against $12.34 of take booked: the identity holds, we carried no outcome risk on the last pick.

509 of 575 legs carry the venue's rules as they read when they opened; the rest were opened before we kept them. At settlement we read them again: 323 read so far, 323 unchanged. 186 not read again yet: the leg has not settled, or its ticket ended before it did.

The lock is cheapest on NFL (2.44% of expected value quoted) and richest on NHL (4.23%); the record-wide average is weighted by how many of each the scanner found, not by what a user would choose.

This is a SHA-256 over every byte of the ledger up to the last line of that day, chained to the day before, so changing or removing any earlier line changes it and every later one; to check it, save it today and compare it later, or recompute it from a copy of the ledger.

It shows that the history has not been changed since each fingerprint was first published (this one on 2026-11-14, the first on 2026-11-11, so anything older is fixed only as of that day); it does not show that a line was true when it was written; and the full check needs the raw ledger, which carries private details, so it can be run by us and by anyone we give a copy to.