Your Stop-Loss Is a Signal: How Algorithms Feed on the Levels You Think Are "Logical"

You know the move because it has happened to you. You did the work. You found a clean support level — a round number, a prior swing low, the line the whole chart seems to respect. You put your stop just beneath it, where a break would “prove you wrong.” Price drifts down, kisses the level, and then, in a single ugly candle, knifes through your stop, fills you at the worst possible tick, and reverses — climbing back above the level as if the whole thing were staged for your benefit.

You were right about direction. You still lost. And you are left with the queasy sense that the market reached down, took your money specifically, and moved on.

That feeling is usually dismissed as paranoia or sour grapes. It shouldn’t be. The mechanism behind it is one of the best-documented phenomena in market microstructure, and understanding it changes how you place every stop for the rest of your trading life.

This is not a story about a cabal watching your screen. It is a story about clustering — about the fact that the level you found “logical” is the exact level tens of thousands of other traders found logical too, and that your identical stops, stacked in the same place, form a pool of liquidity that the rest of the market can see, price in, and consume.

Key takeaways

• The pattern is real and every trader has seen it: price pushes just past an obvious support, forces a wave of selling, then snaps straight back above the level.
• The academic explanation is order clustering. Osler showed stop-loss and take-profit orders pile up at round numbers — the exact levels traders call support and resistance — and that stops create self-reinforcing cascades.
• Brunnermeier & Pedersen showed the strategic version: when a trader is forced to liquidate, others sell into it to overshoot the level, then buy back — price overshoots, then reverts.
• The fix isn’t to abandon stops. It’s to stop placing them where everyone else does — beyond the crowded level and outside the market’s normal noise.

The anatomy of a stop run

Start with the picture, because once you see the shape you cannot unsee it. Below is the sequence in its purest form: price grinds into a level where stops are resting, sweeps through them in a burst of forced selling that overshoots, and then reverses hard once that selling is exhausted.

Why the “logical” level is the trap

The uncomfortable core of this is that good technical analysis is what builds the pool. The whole point of a support level is that it is obvious — visible to everyone, which is why everyone acts on it. But that shared obviousness is exactly what concentrates orders. The clearer the level, the denser the stops beneath it, and the more attractive it becomes as a target for anyone who benefits from triggering them.

This is not speculation. In a landmark 2003 Journal of Finance paper, Carol Osler examined the actual order book of a large foreign-exchange dealing bank and found that stop-loss and take-profit orders are strongly clustered at round numbers — the same round numbers traders use as support and resistance. That single fact explains two classic technical-analysis predictions at once. Take-profit orders cluster just ahead of round numbers and push price back, which is why trends reverse at support and resistance. Stop-loss orders cluster just beyond them and accelerate price, which is why trends gain momentum once a level breaks.

Support and resistance “work,” in other words, not because of magic lines but because of where people pile their orders. The level is real. The stops beneath it are real. The reason it gets swept is that “obvious to you” and “obvious to everyone” are the same sentence.

From clustering to cascade: why the overshoot happens

Clustering explains where the fuel is. Osler’s follow-up work explains the fire. In “Stop-Loss Orders and Price Cascades in Currency Markets,” she showed that stop-loss orders generate positive-feedback trading: a stop-sell triggers, that selling pushes price lower, which triggers the next stop-sell, and so on — a self-reinforcing cascade. Crucially, she found the market’s response to stop-loss orders was larger and longer-lasting than its response to take-profit orders, and that price trends are unusually rapid precisely when rates reach levels where stops are known to cluster. The effects were significant over hours but not days — which is the statistical fingerprint of exactly the move you keep getting caught in: a violent break that overshoots, and then mean-reverts once the cascade burns out.

That is the whole shape, sourced. The stops don’t just sit there; when the first ones go, they knock over the next ones, price gaps past the level further than any news would justify, and then — with the forced sellers flushed and resting buy interest around the round number still intact — it snaps back. You didn’t misread support. You correctly identified a pool of liquidity and then parked your own order inside it.

Who is on the other side — and is it even legal?

Here is where the honesty matters, because the internet’s “smart money hunts your stops” story is half-right and half-myth. There is a spectrum of what’s actually happening, and only part of it is anybody deliberately coming for you.

The strategic middle tier is the one worth taking seriously, and it has its own landmark paper. In “Predatory Trading” (Journal of Finance, 2005), Markus Brunnermeier and Lasse Pedersen model traders who induce and exploit another investor’s need to liquidate: when someone is forced to sell, predators sell too, driving price past the level so the distressed seller gets a worse fill, and then buy the asset back at the artificially low price. Their phrase for it is the one every stop-loss trader should internalize: the market “is illiquid when liquidity is most needed.” Price overshoots, then reverts — the same shape as the stop run, produced deliberately. The authors point to the trading against LTCM’s known positions in 1998 as the archetype: when the whole market knows you have to sell, your forced order becomes everyone else’s opportunity.

Now add machine learning to that picture. Modern quant and high-frequency systems don’t need to know your individual stop; they infer resting liquidity statistically from the order book and trade history, and can act on those clustered zones in microseconds. Detecting and trading around obvious stop pools is, for a well-resourced algo, close to free money — and mostly legal order anticipation. The genuinely illegal version, spoofing and layering (flashing fake orders to manufacture a break, then cancelling), is real too, and regulators including the SEC and FINRA have increased their scrutiny of it. But you do not need to invoke crime to explain your stopped-out trade. Emergent clustering plus legal order anticipation is more than enough.

The wrong lesson, and the right one

At this point a lot of traders draw precisely the wrong conclusion: if my stops keep getting hit, I’ll stop using stops. This is the single most dangerous idea in this entire essay, and it needs to be said plainly. Removing your stop to avoid being hunted trades a small, survivable, recurring cost for the possibility of an unlimited, account-ending one. The stop-run costs you a fraction; a naked position on the wrong side of a real move costs you everything. Never solve a stop-placement problem by removing the stop.

Keep the stop. Move it. The entire remedy is placement, not abolition. If the crowd’s stops rest just below 100.00 and the swing low at 99.85, then those two prices are the worst places for yours. The safer stop sits beyond the pool and beyond the market’s normal noise — and is paired with a smaller position so the wider distance still risks the same small, fixed amount.

Concretely, three principles fall out of the research. First, place the stop where your idea is actually wrong, not where it’s convenient — the invalidation price, which is usually a little beyond the obvious level, not right at it. Second, put it outside normal volatility: a stop set a multiple of Average True Range away from entry is far harder for routine noise (or a shallow sweep) to reach than one pinned to a round number. Third, and most importantly, let sizing absorb the wider stop. A stop that’s twice as far away is fine if your position is half as large; your dollar risk is identical, but your stop is now sitting outside the pool instead of inside it.

The trader who complains about being hunted is almost always the trader who set a tight stop at the obvious level so they could put on a big position. That’s not a market-structure problem. That’s a sizing decision that made you predictable.

From belief to behavior: is the market hunting you, or are you volunteering?

The good news is that “I keep getting stopped and then it reverses” is not a feeling — it’s a measurement. Your trade history knows exactly how often it happens and what it’s costing, and the pattern tells you which fix you need.

Resources and further reading

• Order clustering & technical analysis: Osler, C. L. (2003), “Currency Orders and Exchange Rate Dynamics: An Explanation for the Predictive Success of Technical Analysis,” Journal of Finance 58(5): 1791–1819 — stops and take-profits cluster at round numbers used as support/resistance.
• Cascades: Osler, C. L. (2005), “Stop-Loss Orders and Price Cascades in Currency Markets,” Journal of International Money and Finance 24(2): 219–241 — stops create positive-feedback cascades, larger and longer-lasting than take-profit effects, significant over hours.
• Predatory dynamics: Brunnermeier, M. & Pedersen, L. (2005), “Predatory Trading,” Journal of Finance 60(4): 1825–1863 — inducing and exploiting forced liquidation; “the market is illiquid when liquidity is most needed.”
• Market structure: broker and exchange education on liquidity sweeps and stop hunts, and SEC / FINRA materials on order anticipation versus prohibited spoofing and layering.
• Placement practice: primers on Average True Range (ATR) stops and volatility-based position sizing, for placing stops outside normal noise.

Frequently asked questions

Essay in Gecko’s trading psychology series. The microstructure findings are drawn from Osler (2003, 2005) and Brunnermeier & Pedersen (2005), with market-structure context from broker education and SEC/FINRA materials, as of July 2026. Descriptions of order anticipation, spoofing, and layering are general and educational, not legal advice. Nothing here alleges specific wrongdoing by any firm. Gecko is an educational and informational tool. Nothing here is financial, investment, or trading advice, or a recommendation for or against any strategy or stop-placement method. Trading carries substantial risk of loss.