Stop Hunting

What the claim says

The idea goes like this. Many traders put their stop losses at obvious places: just under a recent low, just over a recent high, at a round number. Those orders pile up. Someone with the power to move price, the story says a dealer or a large trader, pushes it to that level for a moment, the stops trigger, and price returns. The stopped-out traders took the loss and the one who pushed it took the other side.

The claim is old and widely repeated. It is also impossible to settle from a single account. This page does not accuse any firm of anything, and nothing here says it happens or does not happen at any particular place. It sets out what is known about how stops work, what can look like hunting without being it, and what you can and cannot prove.

What is real: the mechanics

A stop loss is an instruction to close a trade when price reaches a level. It does not promise a fill at that price. Some well-understood mechanics move a stop's trigger and fill without anyone aiming at it:

• The spread. Long trades are closed by selling at the bid. Short trades are closed by buying at the ask. Charts often show the bid or a mid price only. When the spread widens, as it often does around news, at the daily rollover or in thin hours, the ask can reach a short's stop while the chart line is still below it. See spread.
• Slippage. Once a stop triggers it usually becomes a market order. In a fast move the fill can be well beyond the level. See slippage.
• Thin liquidity. When few orders are resting near the price, a modest order can move it further than usual. Stops themselves are orders: a cluster triggering can push price toward the next cluster, with no plan behind it.
• Different price feeds. Different venues and brokers quote slightly different prices. A spike that appears on one feed may not appear on another, and the stop is judged on the feed of the venue it sits with.
• Wicks around obvious levels. Many traders place stops at the same obvious spots, so a normal wick through one will trigger many of them at once. That can look targeted when it is only crowded.

A worked example: the spread alone

This is a made-up short trade on a made-up pair quoted to five decimals, where 1 pip is 0.0001. A short is closed by a buy stop, and a buy fills at the ask, which is the mid price plus half the spread. The stop sits at 1.10500. The highest the mid price reaches is 1.10460, which is 4 pips below the stop. The chart line never touches the stop. Whether the trade is stopped depends on the spread at that moment.

Calculated by us on 29 Oct 2026, exact for these made-up inputs. The highest ask is 1.10460 plus half the spread. At a spread of 1 pip the ask peaks at 1.10465 and the trade survives. At 8 pips the ask reaches 1.10500, exactly the stop, and the trade is closed. At 12 pips it passes it. Nothing about the mid price changed between those rows. Only the spread did.

The trader sees a chart that never reached the stop, a trade that was closed, and a price that then falls away. That looks like a hunt. It is also exactly what a wide spread at a quiet moment produces. The two cannot be told apart from the chart alone.

How it hits bots

• Tight stops. Scalping bots and breakout bots that use stops a few pips away are the most exposed to the spread effect, because the spread is a large share of the stop distance.
• Obvious levels. A bot that puts its stop at the previous day's low, a round number or a fixed distance joins the same crowd as everyone else. A normal wick can trigger it.
• Backtests. Most backtests assume a constant spread, so they never trigger a stop on the spread alone. Live trading does. This is one reason live results come in worse than tested ones. See backtesting.
• Rollover and news. Spreads often widen around the daily rollover and scheduled news. A bot left running through those moments with tight stops will see the effect more.

What cannot be proven

You cannot prove hunting from your own account. One stopped-out trade followed by a reversal is expected: if a stop sits next to a recent extreme, price often does retrace after touching it, because that is what an extreme is. To say a venue targeted stops you would need to show that moves to stop levels happen more than chance and ordinary liquidity would give, across many accounts, with price records from several independent sources. A single trader's statement and one chart cannot do that.

The reverse is also true. You cannot prove it never happens. Rules about how a venue may trade against its clients differ by country and by venue, and they change, so check the rules and the disclosures of the one you use. Nothing here is a claim about any named firm.

How to check your own fills

• Log the time, the trigger level, the fill price and the spread at each stop-out. Over dozens of stops, see how many happened when the spread was well above its usual size.
• Compare your fill with a price record from a second source for the same second. A gap that shows only on one feed is worth knowing about.
• Compare your fills in busy hours with fills in quiet hours. Wide-spread stop-outs cluster in thin hours.
• Test the bot on a demo account and a small live account side by side. If the live one is stopped far more often on the same signals, the difference is costs or fills. See demo account.
• Build the spread into the stop distance. A stop that must survive a spread of 8 pips cannot be 4 pips away.