The name is doing most of the damage. If dark pools had been called non-displayed crossing networks, which is much closer to what they are, nobody would be drawing red circles on volume charts and talking about hidden hands. But "dark pool" sounds like a place where the real market happens while the rest of us trade a decoy, and a whole cottage industry of dashboards and social accounts has grown up around that feeling. The actual mechanics are more boring than the branding, and once you see them, most of the scary charts stop being scary. A few become genuinely useful, which is the part worth getting to.
A dark pool is a registered trading venue. In the US these are called Alternative Trading Systems, they file with the SEC, FINRA examines them, and most are run by large brokers or independent operators. The single feature that makes them dark is that they do not publish quotes before a trade happens. On a lit exchange your resting order sits in a public book where anyone can see the price and the size. In a dark pool the order rests invisibly, and when a matching order arrives the trade executes. What stays hidden is the order before it fills. The trade itself prints to the consolidated tape seconds after execution, marked as off-exchange, where anyone with a data feed can see it. The secrecy is narrower than people assume. It covers what a participant intends to do, and everything they have already done is public.
Why institutions trade in the dark
Think about a fund that needs to buy a position worth several days of a stock's average volume. If it displays that interest on a lit exchange, the price starts moving away before more than a fraction of the order fills. Faster traders fade it or step ahead of it, other institutions adjust, and the fund ends up paying meaningfully more per share than where the stock traded when it started. The industry term for this is market impact, and for large orders it is usually the biggest cost of trading, bigger than commissions and fees put together.
So institutions slice big orders into small pieces and route many of those pieces into venues that do not leak. The value of a dark pool to a pension fund is the value of not announcing what you are doing while you are still doing it. There is nothing sinister in that by itself. At institutional size, information leakage is expensive enough that entire venues exist to reduce it, and the funds using them are mostly managing retirement money, which is a much less cinematic story than the one the name suggests.
Dark pools borrow their prices from the lit market
Here is the piece that quietly kills most of the conspiracy framing. Dark pools reference prices that were discovered somewhere else. The typical dark execution happens at the midpoint of the national best bid and offer, which is computed from the quotes on the lit exchanges. If a stock is bid at 50.00 and offered at 50.02 in the public markets, the dark pool crosses buyers and sellers at 50.01. US rules also prevent a venue from executing at prices worse than the best displayed quotes, so a dark print cannot happen at some secret disadvantaged level. The lit market sets the reference and the dark venue matches inside it.
This dependence is also the serious criticism of dark trading, and it is a better one than anything in the conspiracy threads. If enough volume leaves the lit exchanges, displayed quotes get thinner and less informative, and the midpoint everyone pegs to becomes a worse number. Regulators have worried about this out loud for years, and Europe went as far as capping dark trading in individual stocks. I do not know where the tipping point is, and I am fairly sure nobody else does either. What matters for a retail trader is simpler. Dark pools free ride on lit price discovery rather than replacing it, and the prices you trade at are still made in the open.
What dark volume share can and cannot tell you
Now the data everyone screenshots. In the US, FINRA publishes per-stock volume for each ATS on a delay of a couple of weeks. Separately, the consolidated tape shows in near real time how much of a stock's volume printed off-exchange. Those two things get conflated constantly, and the conflation is where most of the bad takes come from. Off-exchange volume includes dark pools, but in many stocks the bigger share of it is wholesalers internalizing retail orders. When you buy shares through a zero commission broker, your order typically never touches an exchange. A wholesaler fills it from its own inventory and reports the trade off-exchange.
That detail flips the usual story. The stocks with the highest off-exchange share are often the ones with the most retail activity, because retail flow is the flow that gets internalized. A meme stock showing a huge off-exchange percentage is mostly evidence of heavy retail participation, and the popular reading, that institutions are quietly accumulating in the dark, gets it roughly backwards. I have watched people buy a beaten down name specifically because its dark share spiked, when the spike was largely retail panic selling being internalized by wholesalers. The chart was real and the interpretation was upside down.
The other structural problem is direction. Every print has a buyer and a seller, in the dark exactly as on an exchange. The raw data contains dark pool volume and nothing that deserves to be called dark pool buying. On lit markets you can at least infer the aggressor from whether a trade hit the bid or lifted the offer. Midpoint prints, by construction, give you neither. Services selling dark pool sentiment are layering inference on top of prints that are close to directionless, and some are just relabeling FINRA short volume data, which mostly reflects market makers selling short in the ordinary course of filling customer buy orders. Treat any claim of real-time dark pool positioning with suspicion.
How I actually read the numbers
None of this makes the data useless. It makes it a context signal rather than a trigger, and it behaves fine if you treat it that way. My working rules:
- Compare a stock's dark or off-exchange share to its own trailing baseline, never to a universal threshold. A liquid large cap and a thin small cap have naturally different mixes, and one unusual day is noise. Sustained deviation over a week or more is where I start paying attention.
- Separate ATS volume from total off-exchange volume whenever the data lets you. Elevated ATS share built from block-size prints looks institutional. Elevated off-exchange share built from thousands of tiny prints looks like retail internalization, which is a different story entirely.
- Weight individual large midpoint prints over aggregate percentages. A handful of big blocks crossing at the midpoint on an otherwise quiet tape suggests someone is working size. The aggregate share number can move for a dozen unrelated reasons.
- Never assign direction from volume alone. Pair the dark data with price behavior. Rising dark share while price holds flat near a low is at least consistent with patient accumulation. The same dark share while price bleeds lower is equally consistent with patient distribution, and the volume number cannot pick between them.
The honest summary is that dark volume tells you where business is getting done and roughly what kind of participant is doing it, while staying silent on which way they lean. That partial picture is still worth having. Knowing that someone is working institutional size in a name, or that a move is mostly retail churn being internalized, changes how much respect I give the price action. We fold off-exchange share into the market scorecards at Blockcircle for exactly this reason, as one context input among several, and it earns its place without ever being the reason for a trade. Used as a question generator rather than an answer machine, dark pool data turns out to be quietly practical, and you can drop the conspiracy without losing the parts that were interesting.