How many prints, not how many shares
A million shares in forty prints is an institution working an order. A million shares in four thousand prints is a crowd. Same volume, opposite meaning, and only one of them is visible on the tape.

42 scans · 17 raw feeds · full US equity universe
Retail scanners run on the consolidated tape — last price, last size, aggregate volume. Takion takes seventeen raw exchange feeds direct, so its screener filters on things the tape doesn't carry: net bid and ask pressure at the quote, NYSE and Nasdaq auction imbalance, borrow cost, tick repetition, distance to the halt band. This library turns that into 42 named scans.
This is 39-OrderFlow at 09:36. BidNet and AskNet are net order-flow pressure read at the quote — not derived from candles, not inferred from price action. No screener running on the consolidated tape can produce this column, because the tape does not contain it.
Not a dashboard. Not a card layout. Windows of live numbers, each one filtering the entire listed universe and re-ranking on every print. This is one workspace, mid-morning.
Takion's servers sit in the Equinix NY4 facility beside the exchange matching engines and ingest these feeds directly. Everything the screener can filter on comes from this list — which is why the field set has no equivalent at retail pricing.
Depth-of-book feeds from six venues are what make bid and ask net pressure computable. NYSE Imbalances is what makes the closing-auction scans possible. Neither exists on the consolidated tape.
Every mainstream retail screener derives its filters from the same consolidated feed. The differences between them are interface and alerting. The difference here is the input.
| Filter field | Takion | Consolidated-tape screeners |
|---|---|---|
| Price, volume, % change | yes | Yes — this is where the overlap ends |
| Relative volume | yes | Yes, computed from tape volume |
| Net bid / net ask pressure | yes | not available — requires depth of book |
| NYSE closing-auction imbalance | yes | not available — separate exchange feed |
| Nasdaq imbalance, near / match price | yes | not available |
| Tick count and tick repeat count | yes | not available — needs every print, not a summary |
| Borrow status and borrow cost | yes | usually a separate locate product |
| Distance to LULD halt band | yes | rarely surfaced |
| Live exchange trading state | yes | rarely surfaced |
| VWAP distance, bid and ask sides separately | yes | Single signed VWAP distance at best |
| Rolling intraday window on any filter | yes | Fixed intervals, usually |
The consolidated tape gives a screener aggregated volume. It does not tell you how that volume arrived. Takion sees every print, so the screener can filter on the shape of the tape rather than its total.
A million shares in forty prints is an institution working an order. A million shares in four thousand prints is a crowd. Same volume, opposite meaning, and only one of them is visible on the tape.
The same price printing over and over means a resting order is absorbing everything hitting it. That level breaks or it reverses — and either way you knew about it before the candle closed.
Volume tells you what already happened. Rate of volume tells you what is happening right now. Sorting a new-high scan by rate surfaces the move that is starting rather than the one that is finished.
Net pressure at the quote separates buyers lifting the offer from sellers hitting the bid. Price can be flat while one side is being quietly taken out.
A gap on twice its average daily volume before the open is a different animal from a gap on nothing. This is the field that separates a real episodic pivot from a quiet drift.
Web screeners poll on a timer. This one updates when the print happens. There is no window in which your scan is showing you the previous minute.
The scan, the montage and the order router read the same book. With a screener from one vendor and a broker from another, they never quite agree.
Every field above can run on a rolling intraday window instead of the session. That is what makes velocity, tick surge and short-horizon order-flow scans possible at all.
The filter panel exposes roughly two hundred fields on one screen. No wizard, no onboarding carousel, no AI assistant guessing at your intent. It assumes you know what you are looking for and gets out of the way.
A Bloomberg terminal is dense, unfriendly to a newcomer, and completely unapologetic about it. It is designed for someone who uses it eight hours a day and values a second saved over a pixel polished. Takion's sorter is built on the same premise.
Forty-two scans fit on one screen because the windows carry no chrome, no padding, no illustrations. You see the whole market's behaviour in one glance instead of clicking between tabs.
There is no scoring model deciding which alerts you deserve. You write the conditions, the engine applies them to every symbol, and the ranking is whatever column you clicked. What you see is the data, not an opinion about the data.
Each scan names a specific setup and ships with its filters, columns and sort already configured. Ten of them are only possible on this data.
The scans are not a separate window you alt-tab to. They share a layout with Level 2, charting, order entry and the position blotter, and they push symbols out to external charting through the link menu.
Direct exchange feeds require subscriber agreements with each venue, and your professional or non-professional status determines what you pay. Both are covered before anything gets installed.