Two books to prove it in, one vault to trade it for real. The same rule ends all three.
Every prop firm has a drawdown limit, and everybody reads it the same way: lose more than the number and you are out. The word that costs people their season is trailing.
The limit is measured from your high-water mark. Make a new high and the level you may not close below rises with it — and it never comes back down. An account up 12% and back to flat has not broken even. It is most of the way to the floor.
This is not an illustration. The arithmetic is the same function the engine marks accounts with, a breach is terminal there as it is here, and the limits are the ones the API is serving right now.
Tokenised equities against Chainlink reference prices. The deep, slow book: the test is whether you have an edge at all, not whether you can survive a thin market. Feeds run 24/5, and a stale quote rejects the order rather than filling it at a price nobody is trading.
Fresh pools.trade listings, priced straight off the reserves. Tighter on every limit, because here the book itself is the test: fills move the price you are filling at. Universe floors keep the track off pools too shallow to be a fair one.
What the two above are auditions for. An EIP-1167 clone holding LP capital for thirty days, which you can trade along registry-fixed routes and settle. There is no function on it that moves the principal to an address you choose — not for you, not for us.
Finish in the top band and the protocol deploys a sub-vault with your address written into it. For thirty days you can open and close positions inside it. That is the whole of what the key does.
Realised profit splits 80 / 15 / 5 — you, the LPs, the protocol — and it settles on-chain when the term ends. The principal is never yours to move, which is the only reason LPs are willing to be on the other side of it.