Institutional Trading: How Big Orders Move
Institutional trading is the buying and selling done by funds, pensions, banks and other large managers, usually in orders far bigger than the market can absorb at once. Most of those orders are cut into small pieces, so the typical fill looks like any other trade on the tape.
A lot of trading education talks about what institutions are doing on a chart. This page is about what can actually be observed: how a large order gets executed, what the public data shows about the trades, and what the funds must disclose and when.
How it works
An institution rarely buys all at once. A manager deciding to add a position hands the order to a trading desk or a broker. The desk’s job is to fill it at a good average price without announcing the size to the rest of the market.
The standard tool is an execution algorithm. It cuts the parent order into many child orders and sends them over hours or days, often timed against the day’s volume. A benchmark such as VWAP is common: the desk is measured on whether its average price beat the day’s volume-weighted average.
The pieces go to many places. Some go to public exchanges, some to alternative trading systems (ATSs, the registered venues often called dark pools), and some to broker desks that fill them directly. Each piece looks like an ordinary small trade when it prints.
Occasionally a whole block does trade at once. When a natural buyer and seller of similar size meet, a single large print can happen, usually on a venue built for it. The data below shows how rare that is.
Where big orders trade
FINRA publishes each ATS’s weekly volume. Under Rule 6110(c), weekly totals for Tier 1 NMS stocks are published “no earlier than two weeks following the end of the ATS Trading Information week,” and for other NMS stocks no earlier than four weeks after. The rule also requires monthly block-trading statistics, published at least a month after the month ends.
The totals for one recent week tell the story. In the week starting 31 Aug 2026, 33 ATSs reported 5,922,875,028 shares of Tier 1 stocks in 88,903,039 trades. Divide one by the other and the average ATS trade was 66.6 shares.
Most of those venues run on small fills. 22 of the 33 averaged under 100 shares a trade. The largest by volume, Intelligent Cross, traded 896,021,172 shares in 12,334,499 trades, an average of 72.6.
A few venues are built for size. Liquidnet’s Negotiation ATS averaged 22,715.5 shares over 349 trades, and Instinet BlockCross 2,851.7 over 6,537. Dealerweb averaged 103,384.7 over just 317 trades.
What gets disclosed, and when
Holdings are disclosed quarterly, on a delay. The SEC’s Form 13F FAQ says managers that “exercise investment discretion over $100 million or more in Section 13(f) securities must file Form 13F.” The report is due “within 45 days after the end of the calendar quarter.”
It shows long positions only. The same FAQ tells filers not to include short positions, and not to net them against longs. A 13F can therefore show a fund holding a stock it is hedged against.
And there is no single size that makes a trade a block. The SEC’s Rule 10b-18, which covers company share buybacks, defines a block as a trade of $200,000 or more, or at least 5,000 shares worth at least $50,000, or a size measured against the stock’s trading volume. Other rules and venues use their own thresholds.
A worked example
Take a hypothetical fund buying 500,000 shares. At the week’s average ATS trade of 66.6 shares, that is 500,000 divided by 66.6, about 7,508 separate fills. Nobody watching the tape sees a 500,000-share buyer; they see thousands of small prints mixed in with everyone else’s.
On a block venue the arithmetic changes. At BlockCross’s average of 2,851.7 shares, the same order is about 175 fills. At Liquidnet Negotiation’s 22,715.5, it is about 22. That is why block venues exist, and the data shows how little of the week’s volume they actually handle.
Then the block test. Under Rule 10b-18, 5,000 shares at $12 is $60,000, which meets the “5,000 shares and at least $50,000” test. The same 5,000 shares at $9 is $45,000, which does not, and it is also under $200,000, so it is not a block under that rule.
Finally the disclosure lag. For the quarter ending 30 Jun 2026, the 45-day deadline falls on 14 Aug 2026. A 13F read on that date describes holdings that are already six weeks old.
The original data
The data: FINRA’s weekly ATS summary by firm for NMS Tier 1 stocks, week starting 31 Aug 2026, first published 21 Sep 2026, downloaded on 25 Sep 2026 and published here as a CSV of FINRA ATS volume by venue. Average shares per trade is shares divided by trades, computed for each venue.
Across all 33 venues the average was 66.6 shares a trade, on a notional value of about $702.5 billion. The three venues averaging over 1,000 shares, Dealerweb, BlockCross and Liquidnet Negotiation, handled 59,341,968 shares in 7,203 trades between them.
How concentrated the venues are
Four venues carried 45.1% of the week’s ATS shares: Intelligent Cross, UBS ATS, Level ATS and Sigma X2. The three block-sized venues carried 1.0%. The other 26 carried the remaining 53.9%.
So the picture of institutions trading in giant hidden blocks is mostly wrong for this data. The volume is there, but it moves through a handful of venues in fills about the size of a retail order.
This site’s video corpus reflects the other picture. Of 24,971 videos, 171 have “institutional” or “institution” in the title, at a median of 6,066 views across 142 channels.
74 of those pair it with chart and order-flow vocabulary such as supply and demand, order blocks, liquidity or smart money. One is about dark pools, and none mentions Form 13F or block trades.
When it fails
The first failure is reading a chart pattern as an institution’s order. The fills of a large order are spread across dozens of venues and days. A single candle, zone or level on one chart cannot show where a fund bought, and the smart money concepts page covers which parts of that vocabulary are testable.
The second is treating a 13F as a live signal. It is a snapshot of long positions at quarter end, filed up to 45 days later. The fund may have sold since, and any short hedge is missing.
A third is assuming one large print means one large buyer. A block has a seller on the other side of exactly the same size, and the print alone does not say which side started it.
A fourth is using weekly ATS data for timing. Tier 1 figures arrive at least two weeks after the week ends, and the week in this page was published on 21 Sep, 17 days after its last trading day. It describes where volume went, not where price goes next.
And a fifth is copying the execution style without the reason. Splitting an order makes sense for 500,000 shares. For a few hundred shares, it only adds slippage and time.
Related
Dark pools explains why large orders trade without being displayed first. Smart money concepts sets out the chart-based vocabulary that borrows the word institutional, and what in it can be checked. And VWAP is the benchmark many execution algorithms are measured against.
Treat anything described as institutional footprints on a chart with care. The real footprints are in FINRA’s weekly files and in 13F filings, both public and both late, and neither tells you what a fund will do next.
— Michael Whitman
This page is educational, not financial advice. Test every idea on your own charts before risking money.