Protecting Positions: Stop-Loss, Take-Profit and Trailing Stops
Prediction markets do not give you a native stop order. Polymarket's order book will happily let you place a limit or market order, but nothing in it watches your position after the fill. If a YES contract you bought at 60¢ slides to 20¢ while you are asleep, the book does not care. Predite fills that gap with protective exits: rules attached to a position that monitor the price and fire a real closing order when your trigger is hit.
This guide covers how they work, which kind to use when, and — most importantly — the failure modes you need to understand before you trust one with real money.
The three trigger kinds
Stop-loss closes the position when the price falls to or below a level you set. It caps how much a single bad thesis can cost you.
Take-profit closes when the price rises to or above your level. This matters more in prediction markets than in equities: a contract that runs from 40¢ to 85¢ has captured most of the move it will ever make, because the ceiling is $1. Taking profit at 85¢ and redeploying beats holding for the last 15¢ while carrying full resolution risk.
Trailing stop follows the price up and closes when it retraces by a set amount from the peak. Use it when you think a market has further to run but you are unwilling to give back a win that has already materialized.
How an exit actually executes
This is the part that separates a protection that works from one that only looks like it works.
When a trigger is hit on a live position, Predite sends a real SELL order to the CLOB, routed through the same guarded executor as any other live trade — so your master switch, live-trading flags, and kill switch all still apply. The order is sent as fill-or-kill: it either executes immediately or it does not execute at all.
That choice is deliberate. A resting sell order that sits unfilled in the book would let the interface claim your position was "closed" while you still hold every share. Fill-or-kill removes that lie. Either the position is genuinely reduced, or the protection reports that it is not.
If the sell does not fill, the protection stays armed. It does not mark itself as triggered, and it tries again on the next cycle. The system will never tell you a position was protected when it was not.
Protections are evaluated on a recurring cycle, not tick by tick. This is the single most important thing to internalize: a stop-loss is not a guarantee of price, it is an instruction to attempt an exit once a level is breached.
One position, one exit
If you attach both a stop-loss and a take-profit to the same position, they behave as a one-cancels-other pair. When one fires, the sibling protections on that same position are canceled automatically.
Without that, the second protection would keep monitoring a position that no longer exists — and eventually try to sell shares you already sold. Predite cancels the siblings for you, so you can safely bracket a position with a floor and a ceiling.
Paper and bot positions
Protective exits work in paper mode too, and bot strategies carry their own exit conditions — stop-loss, take-profit, trailing, breakeven, and pre-resolution exits. Paper exits simulate the close and record the result; live exits go to the book. The logic you learn in paper transfers directly, which is exactly why you should test your exit levels there first.
Gap risk: the thing that actually hurts
In a thin prediction market, price does not glide from 60¢ to 40¢ touching every level in between. News drops, the book empties, and the next trade prints at 35¢. Your 50¢ stop does not get you out at 50¢ — it gets you out at whatever the book will pay when the order arrives.
You cannot eliminate this. You can only manage it:
- •Size for the gap, not for the stop. Ask "what if this resolves against me tomorrow?" before "where is my stop?" A stop is a second line of defense; position sizing is the first. See Risk Management and Bankroll.
- •Prefer liquid markets. A tight book absorbs your exit. A thin one turns a 10-point stop into a 25-point realized loss.
- •Do not set stops inside the noise. If a market routinely swings 8 points on no news, a 5-point stop is a guaranteed exit at the worst moment — you will be stopped out and then watch it recover.
Choosing levels
A workable starting framework:
- •Set the stop where your thesis is wrong, not where your wallet hurts. If you bought at 60¢ because you believed the true probability was 75%, a drift to 55¢ is noise; a fall to 35¢ means the market knows something you do not. Put the stop past the noise, at the level that says "I was wrong."
- •Set take-profit where the remaining upside stops paying you for the risk. At 90¢ you are risking 90 to make 10 — usually a bad trade to keep holding.
- •Use a trailing stop only on positions with genuine momentum, and give it room. A tight trailing stop in a choppy market is just an expensive way to exit early.
A practical routine
- Open the position with a size you would accept losing entirely.
- Attach a stop-loss at your thesis-invalidation level.
- Attach a take-profit where the risk/reward inverts.
- Let the one-cancels-other pairing handle the rest.
- Review after resolution: was the stop hit by noise, or by information? Log it in your journal.
That last step is what compounds. Over a few dozen trades your journal will tell you whether your stops are too tight, too loose, or well placed — which is knowledge no default setting can give you.
Good to know
- •Protections apply to positions you hold; they are not entry orders.
- •A protection that cannot fill stays armed rather than silently failing.
- •Nothing here removes resolution risk: a market can resolve against you between two monitoring cycles. Sizing remains your real protection.