Roughly half of US equity volume does not happen on an exchange. The venues where it does happen are collectively called dark pools in the popular telling, which merges several genuinely different things and has produced a small industry of indicators that measure something other than what they claim.
This article is for informational and educational purposes only. It is not financial advice, investment advice, or a recommendation to buy, sell, or hold any security, cryptocurrency, or financial product. Always verify data with official sources before making financial decisions.
Short answer: where trades happen off-exchange
Short answer: Off-exchange trading covers any execution that does not occur on a lit exchange. It splits into alternative trading systems — the venues properly called dark pools — and internalisation by wholesale market makers who fill retail orders from their own inventory. FINRA publishes weekly volume data for ATS venues with a two-week lag. Neither the data nor the venues reveal direction, and the widely quoted “dark pool index” products are not derived from the FINRA data at all.
ATS vs internaliser vs single-dealer platform
- Alternative trading systems. Registered venues matching buyers and sellers without displaying quotes. Used primarily by institutions working large orders they do not want visible. This is what “dark pool” properly refers to.
- Wholesale internalisers. Market makers that buy retail order flow from brokers and fill it against their own book. Most retail equity orders in the US execute this way. This is the largest component of off-exchange volume and it is not a dark pool.
- Single-dealer platforms. A bank trading against clients on its own system.
Conflating these produces the most common error in off-exchange analysis: rising off-exchange share gets read as institutional accumulation when it frequently reflects an increase in retail activity being internalised.
The FINRA weekly file and its two-week lag
FINRA publishes ATS transparency data weekly, broken out by venue and by security, covering shares and trade counts. Tier 1 securities — NMS stocks in the S&P 500 and Russell 1000 — are published two weeks after the trading week; less liquid names take four.
That lag is the binding constraint. By the time you can see that a venue traded unusual volume in a stock, the information is a fortnight old and any move has happened. The data is well suited to studying market structure and useless for anything resembling a trading signal.
Off-exchange share as a proxy for retail participation
Because internalisation dominates off-exchange volume, the aggregate off-exchange share tracks retail activity reasonably well. It rose sharply during periods of heavy retail engagement and falls when retail steps back.
Used that way — as a participation gauge rather than an institutional-flow gauge — it is genuinely informative. Used the other way, it is backwards.
The dark pool index myth
Several vendors publish a “dark pool index” or short-volume ratio presented as showing whether off-exchange activity is buying or selling. The construction does not support that interpretation.
Every trade has a buyer and a seller. Venue data records that a transaction occurred, not who initiated it. Some products infer direction from trade prices relative to the prevailing quote — a technique with known and substantial error rates in fragmented markets. Others use the FINRA short volume file, which reports shares sold short but includes market maker hedging, meaning a high short-volume ratio often reflects internalisers hedging retail buying rather than bearish positioning.
The specific irony is worth stating: a high short-volume reading frequently indicates retail buying, because the market maker taking the other side of those buys records a short sale.
Why a large print does not tell you the direction
Institutions use dark venues precisely to avoid revealing intent. A large block crossing tells you size transacted, not whether an accumulator found a seller or a liquidator found a buyer.
Block prints also frequently represent the completion of an order worked over hours or days, or one leg of a multi-venue strategy. Reading a single large off-exchange print as fresh institutional conviction assumes away the reason the venue was chosen.
What this article does not conclude
Off-exchange data describes where trading occurred. It does not reveal institutional positioning, does not indicate direction, and is published with a lag that precludes tactical use.
The FINRA ATS transparency data is free and published on FINRA’s website with full methodology. Commercial products layered on top add inference the underlying data does not support.