When to close a position

The last line is the one people get wrong, so here it is slowly.

A refusal is not a verdict on you. It means one thing: this whole position is bigger than what the pool can take right now without pushing backing below the threshold. The contract does not close part of it and refuse the rest. It refuses all of it, and you receive nothing.

Split the position into parts and close them one at a time. Each part takes what is allowed at that moment. The allowance is recomputed after every close, so several small closes get more out than one big one. The limit itself does not move — splitting is simply the only way to collect what the limit already permits.

Numbers, to make it concrete. Say the reserve holds $10 million and 100 million tokens are in circulation: backing is $0.100 a token. The record is $0.120 and the protection level is the default, so the threshold sits at $0.0458. The distance between backing and the threshold — the headroom — is $0.0542 a token, about $5.4 million in all. One closing may use 38.2 per cent of that, roughly $2.1 million.

Now you hold a position of 40 million tokens bought at $0.01. Closing it whole would issue 40 million tokens against only $0.4 million of collateral — far past the allowance. So it is refused, whole, and nothing comes out. Split it into eight parts of 5 million and close them in turn: each takes what the share allows, and the allowance is worked out afresh each time.

And whatever the gate says, the position itself is never at risk. Redeem it and the collateral comes back in full, in the currency you brought, at any moment. The gate rations one conversion; it never rations the exit.

Why the two numbers below are so far apart is worth a moment.

Closing a costly position lifts the floor for everyone holding tokens — and you are one of them the moment you close. So part of what you hand over comes straight back to you. How much comes back depends on one thing: how large you are against everyone else.

A small position gets almost nothing back. Say you hold one per cent of circulation and your entry price is fifty times the floor. You lift the floor for a hundred people and you are one of them, so you keep about one hundredth of the lift. To come out ahead you need the market at roughly thirty-four times the floor — a long way up.

A large position gets almost all of it back. Same entry price, but now the position is twenty times the whole of circulation. You are lifting the floor mostly for yourself, and the floor rises nearly to where you paid. Five per cent above it is enough.

Same move, same entry price, and the answer differs by a factor of thirty. That is why there is no single number here and why the page gives you two sliders instead of one.

Splitting does not raise the limit. But it is far from useless: a whole position that does not fit is refused and issues nothing, while the same position split takes whatever the limit allows. Both statements are true, and people usually remember only the first.

The level is voted by holders: 61.8 per cent at the loosest rung, 23.6 at the strictest. Between them the level changes about two and a half times, and the limit above it about five.

What you lose is not the exit but its certainty. “I always pass” becomes “I pass if there is room and I was in time”. Redemption of the position stays open throughout, so nothing here threatens your principal.

And the contract is indifferent to being left. Redeeming a position returns the collateral; redeeming tokens returns a proportional share of the reserve and leaves the fee behind for those who stay. Neither can be paused, taxed into uselessness or deferred by any role or any vote. An exit is not something the protocol resists — it is something it is built to absorb.