What Are Dark Pools?
A dark pool is a private trading venue where orders are not displayed before they execute. Trades still print to the public tape afterwards, so the information arrives late rather than never, and the delay is the entire reason large institutions use one.
How it works
A dark pool is a trading venue that does not display its orders. Buyers and sellers submit interest privately, the venue matches what it can, and the resulting trades are reported to the consolidated tape like any other trade.
The word “dark” refers to the order, not the transaction. Nothing is concealed permanently. What is withheld is the intention — the fact that someone wants to trade a large quantity — and it is withheld only until the trade is done.
These venues are regulated entities. In the United States they are registered alternative trading systems, they file with the Securities and Exchange Commission (SEC), and their volumes are published. The name is far more dramatic than the arrangement.
The market-impact problem
Start with the problem. A pension fund wants to sell two million shares. The visible order book holds perhaps thirty thousand near the current price. Displaying the full order tells every other participant that a very large seller exists.
What happens next is not a conspiracy; it is arithmetic. Buyers withdraw or lower their bids because they now expect to buy cheaper. The price falls before most of the order has traded, and the seller receives a worse average for the shares they had not yet sold.
That is market impact, and it is the cost a dark venue exists to reduce. Trading without announcing lets a large order find the other side without first repricing the market against itself.
Most dark venues price off the public market rather than setting their own price. A common design matches at the midpoint of the national best bid and offer, which means both sides do better than they would have crossing the public spread.
On the site’s shared history the public spread measures 2% of a typical bar’s range. A midpoint execution splits that, which is a real saving repeated across a very large order.
In practice: reading a print
The volume you see on a chart is consolidated after the fact. It includes off-exchange prints, so the total is right — but the timing is not. A block negotiated at 11:02 may print at 11:04, and the candle it lands in is not the candle during which the decision was made.
That lateness is the single most important property for a retail trader, and it is the one most often glossed over by services selling dark-pool data. Whatever you are looking at has already happened, and the participants who caused it have already finished.
There is a second, larger problem with reading prints. A trade has a buyer and a seller. The tape records that a quantity changed hands at a price. It does not record which side initiated, which side was passive, or whether either was hedging something unrelated.
Direction on a dark print is therefore inferred, not observed. Services that label prints as bullish or bearish are applying a heuristic — usually whether the price was nearer the bid or the ask — and that heuristic is a guess with a known error rate, not a reading.
The share of activity happening off-exchange is also fairly stable. It moves with market conditions, but it does not swing dramatically week to week, which limits how much any single day’s figure can tell you.
When it fails
Most off-exchange volume is unremarkable. It is retail flow internalised by wholesalers and ordinary institutional business, not a signal about anything. Treating routine plumbing as an event generates a stream of alerts with no content.
The failure that costs money is acting on an unknowable direction. A large print appears, a service labels it, and a trade is placed on the label. The print was real and the label was invented, which is the worst possible combination — a true fact wrapped around a false one.
A second failure is a sizing illusion. Prints look enormous compared with a retail order, and that size is emotionally persuasive. It is also routine at institutional scale, where a block that dwarfs your account is an ordinary rebalancing trade.
A third is attributing intent to a fund that has none. A great deal of institutional trading is mechanical: an index fund tracking a reconstitution, a pension fund rebalancing to a policy weight, a manager meeting redemptions. Those trades happen on a calendar, not on a view, and reading conviction into them attributes an opinion to a spreadsheet.
And the venues themselves are not uniform. Some match only at the midpoint and admit no high-frequency participants; others are broker-operated and internalise their own customer flow. The word covers arrangements different enough that a claim about “dark pools” as a category is usually too broad to be either true or false.
The reasonable use of this information is negative rather than positive. Knowing that a large share of volume happens away from the visible book explains why depth looks thin relative to the turnover a symbol reports. That is a useful correction to an assumption. It is not a trade.
The original data
Only 2 of the 24,971 videos measured for this site cover dark pools — at a median of 94,085 views, the second-highest median in the entire market-mechanics group. Enormous interest, almost no supply, and most of what exists sells inference as observation.
The defensible summary is short. Dark pools reduce market impact for large orders, they report to the same tape everyone else sees, their prints are late, and their direction is unknown. Everything beyond that is interpretation, and it should be labelled as such.
Related
The order book is the visible market these venues are defined against. Market makers are the firms that internalise much of this flow. And payment for order flow is how a retail order ends up executing off an exchange without anyone choosing that.
Every dark-pool alert service I have looked at sells the same thing: a print, after the fact, with a direction attached that nobody actually knows. The print is real. The direction is a guess dressed up as data, and I would rather be told that plainly.
— Michael Whitman
This page is educational, not financial advice. Test every idea on your own charts before risking money.