Dark
Pool
Where the shares actually changed hands. Institutions trade size away from the lit exchanges so they do not move the tape, but the prints still get reported.
Off-exchange block prints reported to FINRA across the 25 most liquid names we track, rolled up into a net accumulation or distribution lean per ticker. Read the limits section before you use this one. There are real constraints on what the data can tell you, and most of the internet gets them backwards.
THE 60-SECOND VERSION
Why size hides, and why it still shows up.
If a fund needs four million shares and sends that order to the open market, the price runs away before it is filled. So it trades off-exchange instead, quietly, in blocks. Those trades still have to be reported to FINRA, which means the footprint survives even though the execution was hidden.
CYAN · ACCUMULATION
More notional printed above the prior close
The lean suggests a desk is building a position. On a multi-month view that biases the name higher, because somebody large wants more of it than they currently hold.
A REASON TO LOOK, NOT TO BUY
ORANGE · DISTRIBUTION
More notional printed below the prior close
A desk is feeding stock back out. This makes a name go sideways or stall, not fall off a cliff. A patient seller caps the upside, they do not create a crash.
NOT A SHORT SIGNAL. SERIOUSLY.
Notice the colors are cyan and orange, not red and green. That is deliberate. Every other price signal on the platform uses red and green, and the lean here is not a price signal. It is a positioning read on a months-long horizon. The colour choice exists specifically to stop your brain filing it under "buy" and "sell."
READ THIS FIRST
Five honest limits on this data.
Dark pool data is the most over-hyped feed in retail trading. Here is exactly what it does and does not know, so you can use it properly instead of the way finance social media uses it.
FINRA publishes no aggressor side
Nothing in the feed says "this was a buy." There is no venue and no initiating side in the data. The lean you see is inferred by comparing each print's price against the prior session's close. Above it leans accumulation, below it leans distribution, and a print within about five basis points of that close counts as neutral and is dropped from the net notional entirely, not forced to a side. That is a reasonable heuristic. It is not the same thing as knowing who initiated.
The tape is not institutions only
The off-exchange feed carries every off-exchange trade, including ordinary retail order flow internalized by a market maker, not just negotiated institutional crosses. Most retail-sized trades never clear the block floor, but the feed itself makes no distinction between the two kinds of print.
Distribution is not a short signal
This is the one people get backwards, so it is worth being blunt. Both words describe what a desk is doing with size over months, not what price does tomorrow. A large seller working an order patiently is what makes a name stall, and while they are working it they are also the reason it does not gap down. Shorting into distribution means shorting a name that is being held up by somebody who does not want it to move.
Big does not mean directional
A large cross can be a hedge, an index rebalance, a portfolio transition, or one fund handing stock to another because a mandate changed. Size tells you something happened. It does not tell you what happens next, and it very often has no opinion about price at all.
The tape is row-capped
When a session exceeds the cap the oldest prints are dropped, and the net lean for that window is computed from what survived. The page tells you when this has happened. Believe it, and treat the lean as partial for that session rather than arguing with the number.
THE ALPHA
Levels, patience and a second opinion.
Used within its limits this page is genuinely useful. Three specific things it gives you, none of which are entries.
Repeated blocks around one price mark where real size was willing to transact. That makes a much better stop reference than a trendline you drew yourself.
A cyan lean on the shares plus bullish options flow on the same name is the actual signal. Neither one alone is worth much. Together they are hard to dismiss.
A $40M block in a mid-cap is a completely different event from the same block in SPY. Block share of volume is what stops you being impressed by big numbers on huge names.
ANNOTATED · THE REAL SCREEN
Block % before Gross. Always.
Gross tells you the dollar size. Block % tells you whether that size actually mattered to the stock, and the ten-session column tells you whether it is a trend or a one-off. Only those last two are useful on their own.
| Ticker | Last | Prints 3 | Gross | Net lean 1 | Biggest | Block % 2 | 10 sess |
|---|---|---|---|---|---|---|---|
| MU | $104.80 | 34 | $412M | ACCUMULATION+$182M | $61M | 18.4% | rising |
| HOOD | $38.60 | 41 | $505M | DISTRIBUTION-$240M | $88M | 21.1% | declining |
| SPY | $571.40 | 308 | $2.20B | ACCUMULATION+$690M | $210M | 5.4% | flat |
| BAC | $44.10 | 17 | $96M | FLAT+$4M | $12M | 4.2% | flat |
Illustrative rows built from the real field set. Not live quotes, not recommendations. Click any row to expand it: the drawer adds Avg print, Concentration, Lean ratio, Block shares and Sector, plus that ticker's recent prints. Row cap hit: oldest prints dropped for this window, net lean computed from what survived.
SPY is the row to ignore. SPY has a $690M accumulation lean on $2.20B gross, by far the biggest number on the screen. Block % is 5.4% and the ten-session trend is flat. That is just Tuesday. SPY trades enormous off-exchange volume every single day and almost none of it expresses a view. MU, at $412M gross and 18.4% block share with a clearly rising ten-session trend, is the row describing something unusual. Block % and the trend column are what separate them, which is why you read those before the dollar figure.
Net lean, an inference not a report
Notional that printed above the prior close minus notional that printed below it, rolled up per ticker for the selected window. Cyan leans accumulation, orange leans distribution. Remember there is no aggressor side in the source data, so this is a heuristic doing its best, not a fact.
LOOK FOR → a lean that persists over weeks
Block %, the only scale that matters
Block shares measured against the name's own consolidated share volume, not notional. It is a same-session number and only shows on the Today window; every other window reads as a dash. This is the column that stops you being impressed by a headline dollar number on a mega-cap.
LOOK FOR → 10%+ block share on the Today window
Prints and Biggest, size versus spread
Prints is how many blocks made up the total, Biggest is the largest single one. A few enormous crosses and three hundred small ones mean different things. Expand the row and the drawer gives you Avg print, Concentration, Lean ratio and the ticker's recent prints by time, size and side. The drawer does not show a price, so export the prints to CSV when you actually want a level to defend.
LOOK FOR → high concentration, then open the tape
10 sess, the trend that makes a lean real
Ten trading sessions of history, not ten calendar days, independent of whatever window you have selected above. This is the most important column on the board, because a one-day lean means almost nothing and a lean that has been building for two weeks means a great deal.
LOOK FOR → a trend, not a spike
WORKED EXAMPLE · ILLUSTRATIVE
Using the prints as a level, not a signal.
Note carefully what starts this trade: options flow, not the dark pool page. The blocks only ever do two jobs here, confirming the lean and giving you a stop. Real ticker, illustrative fills.
The options tape flags MU first
Repeat bullish call sweeps on MU over several sessions, spotted on Unusual Activity. That is the reason you are looking at this name at all. Now you come here to ask a single question: is anybody accumulating the actual shares?
Cyan lean, and it has held for two weeks
MU shows accumulation at 18.4% block share with a rising ten-session trend. Expand the row and the drawer lists repeat prints with notional climbing print over print, an accumulation pattern rather than one big cross. Export the tape to CSV and the fills cluster between $103 and $105. Two independent sources now agree: options flow says bullish, the share prints say somebody is building.
One session of lean would not have been enough. Two weeks of the same lean is the bar.
The block cluster becomes the stop
MU at $106.20. Buy 3 MU $110 calls, 60 DTE at $5.20, which is $1,560 at risk. Long expiry because accumulation plays out over months, not days. The stop is written down as a close below $102, just under where the size transacted. If price cuts through the level real money defended, your read was wrong.
Out when the lean flips
MU reaches $116 and the dark pool lean turns orange as the same desk starts feeding stock back out. That is your cue, and it is the most useful thing this page will ever do for you. Contracts mark $10.40. Sell all three.
Illustrative. The dark pool data contributed two things to this trade: a second opinion, and a stop level. It did not generate the idea and it should not.
MU closes $101 in week three and the cyan lean is still showing. Your stop hits and you are down $920. What actually happened is one of the limits above: the "accumulation" was an index rebalance, or a fund taking on a position from another fund, and it never had a view on price at all. The inference was reasonable and it was still wrong. This is exactly why the lean is context and the stop is a rule.
THE FILTER
Green flags, red flags.
✓Worth using when…
✕Ignore it when…
HOW PEOPLE LOSE MONEY HERE
Three mistakes, in order of cost.
Shorting distribution
It reads like a sell signal and it is not one. A desk feeding stock out over weeks makes a name stall, and their working order is part of what keeps a bid under it. Traders short into that, watch the name grind sideways for a month, and pay theta the whole way.
Treating an inference as a report
People say "the dark pool bought 4 million shares." Nobody knows that. FINRA gives no venue and no aggressor side. The lean is derived from print price against the prior close, which is a decent guess and nothing more. Certainty here is always somebody's invention.
Mismatching the timeframe
This is months-long positioning data. Buying a Friday expiry off it is like reading a weather forecast for next season and packing for tomorrow. Even when the read turns out right, the option expires long before the thesis resolves.
PAIRS WITH
Neither lean is an entry. Both are context.
Find the setup somewhere else, then bring it here. That order matters more on this page than on any other tool in the platform.
Pick three names and log the lean every Friday.
Cyan or orange, and the percentage of ADV on the Today window. Do it for two months and then look back at what the stocks did. You will learn more about the real predictive value of this data from your own log than from anybody's thread about it.
Every ticker, print, lean, fill and P&L on this page is illustrative and built from the real structure of the tool. Off-exchange prints are reported to FINRA with a delay and carry no venue or aggressor side, so accumulation and distribution leans are inferred, never reported. Nothing here is a recommendation, a live quote, or a promise of a result. Options carry substantial risk of loss, including total loss of premium paid.