Your Stop-Loss Is Sitting Exactly Where Everyone Else's Is
Orders cluster at round numbers, and that clustering costs US investors close to a billion dollars a year. The problem is not the stop-loss. It is where you put it.
Your Stop-Loss Has a Price Tag (And Liquidity Hunters Know Where)
Prefer to watch? This article is the written companion to the video above.
You set a stop-loss. The price drifts down, clips it, sells you out — and then reverses almost immediately.
You are not imagining it. And it is not a conspiracy.
It is structure, and once you see it you cannot unsee it.
First, the important caveat
Stop-losses are a legitimate and valuable tool. For most investors, especially long-term ones, they belong in the toolkit.
This article is not an argument against using a stop.
It is an argument about where you put it.
What a stop-loss actually is
Not a shield. Not protection. Here is the mechanism, stripped of comfort:
Your stop sits on your broker’s server, waiting. The moment the price touches your level, it converts into a market order — an instruction that says:
“Sell now. At whatever price is available.”
Read that again, because everything follows from it.
A market order is not a request for a good price. It is a supply of shares, handed to whoever happens to be buying, at whatever they are willing to pay.
And large buyers need exactly that. If a fund wants to acquire a serious position, it needs sellers — a lot of them, all at once, without moving the price against itself.
When your stop triggers, you become that seller. You did not choose the price. You did not choose the moment. You supplied the liquidity.
And here is why they do not need to see your order
Nobody is hunting your stop. Nobody can see it, and nobody cares about it.
They do not need to.
What they know is something weaker and vastly more useful: statistically, where the clusters are.
And the crowd is astonishingly consistent about where it puts them.
The round number problem, measured
This is not folklore. It has been measured, on real orders, at enormous scale.
Bhattacharya, Holden and Jacobsen examined more than 100 million US stock transactions and published the result in Management Science.
What they found:
- Systematic excess buying at prices one penny below round numbers
- Systematic excess selling at prices one penny above round numbers
- The effect is strongest at whole integers, weaker at half-dollars, weaker still below that — it scales with how “round” the number is
Human beings use round numbers as cognitive anchors. We do it without noticing. And it shows up in the tape.
The cost, in their estimate: losses approaching $1 billion a year in US stocks — transferred from the participants who cluster at round numbers to the participants who do not.
That figure is not specifically about stop-losses; it covers all orders placed at obvious levels.
But if your stops habitually sit at $100, $50, $20 — you are paying into it, a small slice at a time, forever.
If you put your stop at exactly $100, you are not being precise. You are being predictable.
The mechanism, in four steps
1. The cluster forms. Price moves into a pattern that retail traders recognise — a double bottom, a clean support line, a textbook setup. You see the signal. You enter. You place your stop just below the obvious low, exactly where the book says — and exactly where thousands of others have placed theirs.
2. The level breaks. Through ordinary volatility, or because a large participant is executing size. It does not matter which.
3. The stops fire, all at once. Every one of them converts to a market order. The book is suddenly flooded with sellers who have no choice and no price limit.
For anyone looking to buy a large position, this is close to ideal. Abundant sellers, precisely when needed.
4. The selling stops. The cluster is cleared. The buyers are filled. The pressure evaporates — and the price snaps back.
You watch the recovery from the sidelines, having sold at the worst tick of the move.
The research, because this deserves better than a chart with arrows
Brunnermeier and Pedersen, Predatory Trading, Journal of Finance (2005).
Their model is precise: when some traders are known to be forced to sell — by margin calls, or by stops — other traders can profit by selling first, pushing the price further against the distressed party, and buying back afterwards.
The consequence they identify is the important one:
Price overshooting — and a market that is least liquid exactly when liquidity is most needed.
Forced selling is not merely unfortunate. It is exploitable, and the model shows how.
Carol Osler, Federal Reserve Bank of New York, Staff Report 125.
Osler went further and looked at actual stop-loss and take-profit orders at a large FX dealing bank — real orders, not inferences from a price chart.
Her finding:
- Take-profit orders tend to reverse price trends
- Stop-loss orders tend to intensify them
Stops trigger stops. That is a self-reinforcing cascade, documented from a real order book by a Federal Reserve researcher.
This is not conspiracy. It is market microstructure, and the people who study it publish in the Journal of Finance.
What August 2024 actually showed
5 August 2024. Bitcoin falls to around $49,000. More than $1 billion of leveraged positions are liquidated.
The temptation is to call this a stop hunt. Resist it — the honest reading is more useful.
The trigger was macro and entirely public: the Bank of Japan raised rates on 31 July 2024, the yen carry trade began to unwind, and risk assets sold off across the world. The episode was significant enough that the Bank for International Settlements published a bulletin on it.
Nobody came for your stop.
But look at what forced selling did once it started. The liquidations were mechanical — margin engines, not decisions. Nobody weighing value, nobody choosing a price. And the sharpest part of the recovery came after the liquidations cleared, because the selling had never been about the asset in the first place.
The lesson is not that they hunted you. It is that forced selling is its own accelerant — and being on the forced side of it is a position you choose in advance, by how you size and where you place.
What to actually do
Three adjustments. None is a guarantee. All of them move you out of the obvious place.
1. Stop being where the crowd is
Support at $100? The crowd’s stop is at $99.50, or $99.
Place yours further out — many traders use a buffer of one to two percent, scaled to how volatile the asset actually is. A stock that routinely moves 3% in a day needs more room than one that moves 0.5%.
And avoid the tidy number.
$97.83 is better than $98.00, and the round-number research tells you exactly why.
2. Position size beats stop placement — and it is not close
This is the one that actually matters, and it is the one nobody wants to hear, because it is boring.
If you size a position so that ordinary volatility cannot hurt you, you do not need a tight stop.
And it is the tight stop, sitting in the obvious place, that gets taken out.
A common rule of thumb: risk no more than 1–2% of the portfolio on any single position.
Get that right and you have bought yourself room to breathe — room for the position to be wrong for a while and still recover. That is worth more than any amount of cleverness about levels.
The trader who gets stopped out at the bottom is almost never the trader who placed the stop badly. It is the trader who bought too much and therefore had to place it tightly.
3. Consider a time-based exit
Instead of exiting only on price, exit on the calendar:
“If this has not started working within X weeks, I reassess.”
The logic: if the thesis was right, it should have begun to show by now.
And the advantage nobody mentions: a time-based rule cannot be hunted. There is no level for anyone to push through, because there is no level.
The one line to keep
When a retail trader looks at the chart, they see:
SUPPORT BROKEN.
When a large participant looks at the same candle, they may simply see:
LIQUIDITY AVAILABLE.
Two readings of the same bar. Only one of them is about you.
Your stop-loss is not a shield. It is an order, sitting on a server, waiting for somebody to fill it.
Stop being where the crowd is. Size so that being wrong is survivable. And never confuse a market order with protection.
Educational content only — not investment advice, and not a personal recommendation. The examples are illustrative and are not evidence of market manipulation. Speak to a qualified, licensed professional before acting.
Primary sources
- 01Penny Wise, Dollar Foolish: Buy–Sell Imbalances On and Around Round Numbers — Management Science 58(2), 2012, pp. 413–431. Losses approaching $1bn per year in US stocks, from a sample of over 100 million transactions — Bhattacharya, Holden & Jacobsen / Management Science
- 02Predatory Trading — Journal of Finance, Vol. 60, No. 4 (2005), pp. 1825–1863 — Brunnermeier & Pedersen / The Journal of Finance
- 03Currency Orders and Exchange-Rate Dynamics: Explaining the Success of Technical Analysis — Staff Report No. 125 — Carol L. Osler / Federal Reserve Bank of New York
- 04The market turbulence and carry trade unwind of August 2024 — BIS Bulletin No 90 — Bank for International Settlements
Questions people actually ask
Are stop-losses a bad idea?
No. A stop-loss is a legitimate and valuable risk management tool, and for most investors it belongs in the toolkit. This is not an argument against using one. It is an argument about where you place it — because the single most common placement is also the most predictable one, and predictability is what costs you.
What actually happens when my stop-loss triggers?
It converts into a market order — an instruction to sell now, at whatever price is available. That distinction matters more than anything else in this article. A market order is not a request for a good price. It is a supply of shares, delivered to whoever happens to be buying. When your stop fires, you are not exiting a position so much as providing liquidity, and you have no say in the price at which you provide it.
Do big players hunt my specific stop-loss?
No, and it is worth being precise, because the conspiracy version of this story is both wrong and unhelpful. Nobody can see your individual order, and nobody cares about it. What large participants know is something weaker and far more useful to them: statistically, where clusters of orders are likely to sit. They do not need your stop. They need the crowd's, and the crowd is remarkably consistent about where it puts them.
Why do orders cluster at round numbers?
Because human beings use round numbers as cognitive anchors, and this shows up in the data with almost embarrassing clarity. Bhattacharya, Holden and Jacobsen examined more than 100 million US stock transactions and found systematic excess buying one penny below round numbers and excess selling one penny above — with the effect strongest at whole integers, weaker at half-dollars, weaker still below that. It is not superstition. It is a measurable, repeatable pattern in real orders.
What does that clustering actually cost?
The same study puts the losses at approaching $1 billion a year in US stocks — money transferred from participants who cluster at round numbers to participants who do not. That figure is not specifically about stop-losses; it covers all orders placed at obvious round levels. But if your stops habitually sit at $100, $50 or $20, you are contributing to it, one modest slice at a time.
What is predatory trading?
A documented model, not a theory about villains. Brunnermeier and Pedersen showed in the Journal of Finance that when some traders are known to be forced to sell — through margin calls, or stops — other traders can profit by selling first, pushing the price further against them, and buying back afterwards. The result they identify is price overshooting: the market is at its least liquid precisely when liquidity is most needed. Forced selling is not just unfortunate. It is exploitable, and the model shows exactly how.
Is there evidence stop-losses move prices?
Yes, and it comes from the Federal Reserve Bank of New York. Carol Osler examined actual stop-loss and take-profit orders at a large foreign exchange dealing bank and found something specific: take-profit orders tend to reverse price trends, while stop-loss orders tend to intensify them. Stops trigger stops. That is a self-reinforcing move, documented from real order books rather than inferred from charts.
Where should I put my stop instead?
Not at the level the textbook shows you, and not at a round number. If the obvious support is $100, the crowd's stop is at $99.50 or $99. Sitting a little further out — many traders use a buffer of one to two percent, adjusted for how volatile the asset actually is — puts you outside the cluster. And an unlovely number like $97.83 is better than a tidy $98.00, for exactly the reason the round-number research describes. These are heuristics, not guarantees.
What matters more than stop placement?
Position size, and it is not close. If you size a position so that a normal, boring, everyday fluctuation cannot hurt you, you do not need a tight stop — and it is the tight stop, sitting in the obvious place, that gets taken out. A common rule of thumb is to risk no more than one to two percent of the portfolio on any single position. Get that right and you have bought yourself room to breathe, which is worth more than any amount of cleverness about levels.
What is a time-based exit?
A rule that exits on the calendar rather than on the price: if this position has not done what I expected within a defined period, I reassess. The logic is simple — if the thesis was correct, it should have started working by now. And the practical advantage is one nobody talks about: a time-based rule cannot be hunted. There is no price level for anyone to push through, because there is no price level.
Was the August 2024 crypto crash a stop hunt?
No — and using it that way would be exactly the kind of overreach this article is arguing against. The trigger was macro: the Bank of Japan raised rates on 31 July 2024, the yen carry trade unwound, and risk assets sold off worldwide, an episode significant enough that the Bank for International Settlements published a bulletin on it. Bitcoin fell to around $49,000 and more than $1 billion of leveraged positions were liquidated. The lesson is not that someone came for your stop. It is that forced selling, once it starts, becomes its own accelerant — and the fastest recoveries tend to follow the liquidations, because the selling was mechanical rather than considered.
So what is the single takeaway?
When a retail trader sees 'support broken', a large participant may simply see 'liquidity available'. Those are two readings of the same candle, and only one of them is about you. Stop being where the crowd is, size so you can survive being wrong, and understand that your stop-loss is not a shield — it is an order, sitting on a server, waiting to be filled by somebody.