Stop hunting — is it your broker or market mechanics?

A stop can trigger before price reverses, but that observation alone does not establish why it happened. Examine the executable quote, spread, order type and timestamps before attributing the move to a broker or a liquidity-driven stop run.

What stop hunting actually is

A stop run describes price crossing levels at which stops are assumed to activate. An ordinary stop can initiate market execution under its product’s trigger rules; it does not guarantee a fill at the trigger price. Sell stops closing longs can add to existing selling. If available bids are depleted, subsequent fills can occur lower and activate further stops. The mechanism can amplify a move without proving deliberate manipulation or a reversal afterward.

Carol L. Osler’s July 2002 New York Fed Staff Report 150 presents historical evidence that currency stop-loss orders can contribute to self-reinforcing moves and sometimes activate in waves. It concerns particular data and documented clusters, not a current map of all traders’ stops. Round numbers and visible highs or lows may suggest hypotheses to test, but no “half the market” proportion, universal cluster or price magnet is established by the chart.

How available bids affect a possible cascade

Activated market sells normally consume available bids; another participant can take the other side, but that does not mean the orders add resting liquidity or reveal an institution accumulating a position. In a hypothetical finite book, bids are 20,000 units at 1.08450, 30,000 at 1.08440 and 50,000 at 1.08420. There are no replenishments or withdrawals. Stop A sells 30,000: it consumes the first level and 10,000 of the second, leaving the best bid at 1.08440.

Assume that new best bid triggers stop B for 40,000 units. B consumes the remaining 20,000 at 1.08440 and 20,000 at 1.08420. The last level retains 30,000 units, and the best bid is three pips below the initial quote when a pip is 0.0001. Total selling is 70,000, including 40,000 triggered after A. This artificial sequence illustrates one feedback mechanism, not the full OTC market or anyone’s actual orders.

OTC FX has no single public book of every broker’s and bank’s orders. One platform’s depth may be based on provider quotes, or even calculated levels if volume is not supplied. Other offers can be added or withdrawn during a real execution. The structure of the OTC market explains that scope. BIS’s January 2017 summary of the 7 October 2016 sterling event points to several contributing factors, including time of day and options-related hedging flows. It does not identify a universal stop-hunt cause or prove the intention behind your wick.

Is it your broker hunting your stop

A provider’s legal counterparty role, pricing and external hedging are separate questions. Principal exposure can create conflicts; some or all risk may be hedged. A market-maker label does not mean every client loss is its net profit. A suspicious quote or execution warrants evidence and an explanation under the actual policy, rather than a claim about the frequency of manipulation. See whether your broker is hunting your stop-loss for the investigation process.

Compare the executable side of the quote and timestamps with independent references. A shared move can support a market explanation; a discrepancy warrants investigation but does not by itself prove artificial pricing. Related reading: ECN versus Market Maker broker.

Execution models can create conflicts of interest. Supervision and best-execution duties do not eliminate every possible abuse, while an ECN label does not prove the absence of a conflict. Examine the provider’s policy and actual order records.

What a stop-run wick looks like

Use an invented EUR/USD range from 1.0850 to 1.0900 and assume some sell stops near 1.0845. These are modelling assumptions, not an observed distribution, everyone’s chart or evidence that retail participants all hold the same orders.

In one hypothetical branch, price reaches 1.0843, assumed stops activate and price later returns to 1.0880, inside the range whose midpoint is 1.0875. In another branch it continues to 1.0800. At activation, the later branch is unknown. A return after a break does not prove an earlier manipulation, and the fact of a stop run does not itself give a buy signal.

Chart levels are hypotheses, not a public stop map

• Round levels. Prices such as 1.1000 and 1.2500 are visible references that can be included in a clustering hypothesis; their simplicity does not prove the current size of stop orders.
• Recent highs and lows. Define the source, session and timeframe before comparing yesterday’s or last week’s extremes. Different feeds need not show identical levels.
• Range edges. A selected boundary is a technical reference, not proof of protective orders held there or which direction follows a break.
• Other technical levels. Retracements, pivots and support/resistance zones are candidate references. A price chart does not reveal every participant’s resting orders or intentions.

Visible chart levels can suggest possible order concentration, but the chart alone does not reveal every participant’s stops. Placement should follow the strategy’s invalidation level and risk limit rather than a guarantee that less obvious prices cannot be reached.

A wider stop requires a different risk calculation

Choose the invalidation level and planned loss under a tested strategy before entry. Moving a stop away from a round number or multiplying ATR does not guarantee that price cannot reach it. Compare changes across the full sample, including losses and missed opportunities.

1. Avoid treating an offset as protection. Moving 1.0900 to 1.0885 changes the stop distance by 15 pips, not merely a few. No chosen offset is demonstrated to avoid an order cluster or improve a strategy.
2. Reassess the plan before entry. A wider distance changes the trade’s invalidation and sizing assumptions. Do not widen protection on an open position merely because a loss is approaching.
3. Use volatility as context. ATR describes recent average ranges, not a boundary that price cannot cross. An assumed 1.5 × ATR stop can still trigger and does not prove that it lies outside an order cluster.
4. Account for spread and slippage. Your stop fills at the market price, so in moments of volatility execution can land a few pips further out. I describe how that works in the piece on price slippage, and the speed of fills themselves in the article on order execution time.
5. Recalculate size. With a hypothetical USD 100 budget and USD 10 pip value per lot, a 20-pip distance allows 0.50 lot before costs; 40 pips allows 0.25 lot. Costs and an execution allowance reduce the available budget. A gap can exceed planned loss, so this does not calculate every realised loss in advance or prove fewer failed trades.

If you trade on very tight spreads and short timeframes, sensitivity to these wicks rises — in the article on the spread for scalping I show why the choice of broker matters especially then. For the broader market-structure background, Forex Mechanics covers choosing a broker in depth.

What to do before you blame the broker again

After a stopped-out trade, compare order records with independent quotes at matched timestamps and on the correct bid or ask side. Check the trigger, spread, slippage and strategy’s invalidation level. A wider stop requires a smaller position for the same planned risk; an ECN label alone does not resolve every execution conflict.