Guide 9.4
Intermediate, 7 min read. Checked with Scalpel 1.12.8 on .
In short
Scalpel grades every headline from the news providers on your IBKR account by kind (FDA, trial data, deal, earnings, M&A, hype, offering and more) and strength, from None to Explosive. The strongest recent one becomes a chip such as FDA 4m: green for strong news, blue for decent, grey for weak, and red for an offering, which is never graded as good news. A keyword grader answers instantly; with an OpenAI key, an AI grade replaces it within seconds.
"Why is it moving?" is the first question on any runner, and the answer decides a lot. FDA approval on a $20M company can carry it all day; a conference attendance press release usually cannot; an offering means the company is selling to the people buying the spike. Scalpel answers it on the scan row and the desk, so you do not have to open a news window while the stock moves.
Before you start
Scalpel uses the news providers your IBKR account already has through the API, such as Dow Jones and Briefing.com. There is no separate paid news feed. When it starts watching a stock, it loads the last News history days of headlines (3 by default), then streams new ones live. The same story arriving on several Dow Jones feeds within 10 minutes is merged into one, and market columns and roundups are never graded as a catalyst.
Each headline gets a kind, shown on the chip:
| Chip | Kind |
|---|---|
| FDA | An FDA or other regulatory decision, designation or clearance |
| Trial data | Clinical trial results or data |
| Deal | A contract, order, partnership, licence or collaboration |
| Earnings | Financial results or guidance |
| M&A | A merger, acquisition, buyout or tender offer |
| Hype | A pivot into a hot theme: AI, crypto treasury, quantum, drones, nuclear, rare earths |
| Uplist | A move up to a major exchange |
| Offering | Raising money by selling stock, warrants or convertibles, or registering an offering |
| Reverse split | A reverse stock split |
| Delisting | A listing deficiency, a delisting notice, or regaining compliance |
| Analyst | An analyst rating or price target |
| Legal | Lawsuits, settlements, patents |
| Fluff PR | A press release with no real content, such as a conference appearance |
| News | Anything else |
And a strength, judged for a low-float momentum stock, not a large cap:
| Strength | Grade | Meaning |
|---|---|---|
| Explosive | 4 | Company-changing: FDA approval, a buyout, a contract bigger than the company, a hot-theme pivot |
| Strong | 3 | Important: positive trial data, a large contract, a strategic deal |
| Moderate | 2 | Real but ordinary: a meaningful contract, solid results, trial progress |
| Weak | 1 | Minor: a small order, a conference, a routine update |
| None | 0 | Nothing price-moving |
An offering is always red, whatever else the headline says, and never graded above weak: "Acme Prices $10 Million Offering Following FDA Approval" is offering news, not FDA news.
The chip shows the lead catalyst: the strongest graded headline from the last 24 hours, newest on a tie. An offering headline always leads, so it can never hide behind good news. Automatic filing notices never lead. The chip's label is the kind and the headline's age: FDA 4m, Deal 2h, Fluff PR 16h.
Hover it for the verdict and the recent headlines, each with its age, strength and kind:
| Chip colour | Card title | Card says |
|---|---|---|
| Green | Strong catalyst | Real news that can bring in buyers. The best runners have this |
| Blue | Decent catalyst | Can move it, but it needs volume to follow through |
| Grey | Weak catalyst | Little real news here. The move is mostly momentum or hype |
| Red | Offering news | The company is selling stock. Usually bad for longs: expect sellers into strength |
The keyword grader runs on every headline the moment it arrives, and stands alone without an OpenAI key. It checks rules in order, dilution first, so an offering word wins over everything else, and shrinking or ending an offering program is correctly not dilution. A concrete dollar figure on a deal or result adds half a point.
| Headline | Keyword grade |
|---|---|
| Acme Receives FDA Approval for Lead Candidate | FDA, Explosive, green |
| Acme Reports Positive Topline Phase 2 Results | Trial data, Strong, green |
| Acme Signs $40 Million Supply Agreement With Major Retailer | Deal, reads Strong, blue chip (2.5) |
| Acme Launches Bitcoin Treasury Strategy | Hype, blue chip (2.5) |
| Acme to Present at Healthcare Conference | Fluff PR, Weak, grey |
| Acme Announces Pricing of $5 Million Registered Direct Offering | Offering, red |
The AI grader, with an OpenAI key and Grade headlines on, sends the headline (and the start of the article, when IBKR supplies it within a few seconds) with the ticker, price, change and market cap. It answers four questions: the kind, the strength against the company's size (a deal worth more than the market cap is huge for a micro cap), whether it raises money, and whether it is new information rather than a recap. Its grade replaces the keyword grade when it lands, usually within a few seconds; if it times out, the keyword grade stays. Of the history loaded at start, the 8 most recent headlines from the last 24 hours per stock are AI-graded.
Click a toast to load its stock onto a desk. More in guide 10.2.
The news and the filings race each other. Dow Jones posts a line for each 8-K within about a minute of EDGAR. A company's press release usually hits the wires before the matching filing, and an offering priced in the evening usually reaches you as a headline hours before its 424B prospectus appears the next morning. Scalpel reads both (guide 9.2), so whichever arrives first sets the chip.
Paid news terminals will be ahead on some headlines. The point here is having the news, the dilution read and the borrow read on the desk where you trade, graded the same way every time.