The stop hunt: what the evidence can and cannot show
Posted: Fri Oct 30, 2026 8:59 am
The stop hunt: what the evidence can and cannot show
Every trader has a story of a stop being hit by a spike that reversed instantly. The conclusion is often that someone targeted it. Sometimes true. Often just the market.
Stops cluster around obvious levels, so price attractive to liquidity seekers tends to visit those levels. Large institutional orders need counterparties, and stop clusters are a source of them. This happens in all markets and does not require a conspiracy.
A broker hunting a specific client's stop is a different claim, and a much harder one to prove. It would need the broker to see the stop, move its own quote, and trigger it. Some cases of that have been documented in disputes. Most spikes on stops are ordinary market behaviour.
A way to separate the cases: compare the spike on the broker's feed with a second independent feed. If the second feed shows the same spike, the market did it. If only the first shows it, the broker's price did something odd.
Another check is whether the spike happens only on the trader's stop level or on every nearby level. A pattern that follows one account is suspicious. A pattern that follows round numbers and prior highs is the market.
Placement matters too. Stops sitting exactly on a round number or a prior swing high are far more likely to be hit than ones a few pips beyond.
None of this makes it comfortable. A stop-out that reverses immediately hurts, whatever the cause. But knowing which kind it was changes the response: switch brokers or adjust placement.
A practical remedy that works regardless of cause is stop placement relative to structure and spread. Beyond a prior swing by a spread's width plus a margin for slippage, the stop avoids the crowded zone where most hits occur. It costs a little more per losing trade, and it reduces the number of stop-outs that reverse immediately. Comparing both placements on the last fifty stopped-out trades shows whether the extra distance would have survived.
How many stop-outs in the last month could be checked against a second feed, and what did the check show?
Every trader has a story of a stop being hit by a spike that reversed instantly. The conclusion is often that someone targeted it. Sometimes true. Often just the market.
Stops cluster around obvious levels, so price attractive to liquidity seekers tends to visit those levels. Large institutional orders need counterparties, and stop clusters are a source of them. This happens in all markets and does not require a conspiracy.
A broker hunting a specific client's stop is a different claim, and a much harder one to prove. It would need the broker to see the stop, move its own quote, and trigger it. Some cases of that have been documented in disputes. Most spikes on stops are ordinary market behaviour.
A way to separate the cases: compare the spike on the broker's feed with a second independent feed. If the second feed shows the same spike, the market did it. If only the first shows it, the broker's price did something odd.
Another check is whether the spike happens only on the trader's stop level or on every nearby level. A pattern that follows one account is suspicious. A pattern that follows round numbers and prior highs is the market.
Placement matters too. Stops sitting exactly on a round number or a prior swing high are far more likely to be hit than ones a few pips beyond.
None of this makes it comfortable. A stop-out that reverses immediately hurts, whatever the cause. But knowing which kind it was changes the response: switch brokers or adjust placement.
A practical remedy that works regardless of cause is stop placement relative to structure and spread. Beyond a prior swing by a spread's width plus a margin for slippage, the stop avoids the crowded zone where most hits occur. It costs a little more per losing trade, and it reduces the number of stop-outs that reverse immediately. Comparing both placements on the last fifty stopped-out trades shows whether the extra distance would have survived.
How many stop-outs in the last month could be checked against a second feed, and what did the check show?