Your first hour in the terminal.

1. 01

   Command bar

   Across the top. Which company, which case, saved or not, when the research was read, when the quote was taken, and the headline value with the method behind it. It does not scroll away.

2. 02

   Panel strip

   Thesis, Valuation, Pipeline, Financials, Catalysts and Compare, plus the Watchlist home. The keys 1 to 6 switch between the six.

3. 03

   Rail

   Down the left. Your lists and saved cases, each with its price. It fills in once you sign in.

4. 04

   Workspace

   The panel you are reading. It is the only part of the screen that scrolls, and it remembers where you were on each panel.

5. 05

   Evidence pane

   Opens on the right when you click a drug, a valuation line, a disease or a source, so the evidence sits beside the number it supports. Escape closes it.

6. Two clocks

   The command bar keeps the research date and the quote time apart. Research is dated to the day the dossier was read. The quote carries its own time and says so when it is stale. A fresh price never makes old research look current.

1. 01

   Click the search box in the command bar, or press the slash key. Type a ticker or part of a company name. FDA approval records turn up in the same list under a drug's brand name, and open at their own page.

2. 02

   Pick the company. It opens on Thesis: the price history, a plain-English account of what the company is and the thesis for it, and the bull and bear cases from the research.

3. 03

   Look at the date under the company name. The research was read on that day. If nobody has opened the company before, the terminal researches it while you wait, about a minute, and shows which parts are done. It keeps going if you close the tab.

4. 04

   Scroll on for recent coverage and the list of documents the facts came from, each with a link to the original.

1. 01

   Open Pipeline. One row per programme and indication: its phase, its next catalyst, then the peak sales, probability of success and launch year the model uses for it, and the dollars per share that line contributes.

2. 02

   Click a programme name. The evidence pane opens beside the table: what the drug does, how its trials went and where it stands against rivals, in plain English. Claims link to the pages they came from.

3. 03

   Click the dollar figure at the end of a row to jump to that line's inputs on the Valuation panel. Click the indication to read the disease beside the case.

4. 04

   A tick beside a name means the programme was matched to the sponsor's own records. "unresolved" means the research described it but could not place it, so it is listed and not valued. Where a programme has an FDA approval record, the row links to it.

5. 05

   On Financials, each figure names its filing: the concept, the period it covers, the form, the accession number and the day it was retrieved. Click a row to see them, then "Inspect source beside case" to open the document itself.

Runway is worked out the same way, in the open: cash divided by the operating cash burn per quarter, with both inputs and the periods they cover printed next to the answer.

1. 01

   Open Valuation. The headline names its method, scenario and share basis on one line. The bridge under it is the same result in rows, one per programme, then corporate costs and net cash, and the rows add up to the headline.

2. 02

   Scroll to Drivers and open a programme. Its inputs are there: peak sales, probability of success, launch year, years to peak, years at peak, decline rate, revenue share and pre-launch spend. The research's reasoning for them sits underneath.

3. 03

   Type a new number. The headline recalculates as you type, in your browser, with nothing sent to a server. The field is marked YOU and gets a Reset beside it.

4. 04

   Bear, Base and Bull are three assumption sets from the research. Your edits sit on top of whichever one is selected.

5. 05

   Undo, or ⌘Z, walks edits back one at a time. The label beside the button says what it would undo.

6. 06

   Go back to Thesis. A card called Your changes states what your edits did to the headline: the model run twice, before and after. It is arithmetic, not a rewritten paragraph.

1. 01

   Open Catalysts. Each event has a type, a date with its precision, and a significance. When enough of them carry exact dates there is a timeline.

2. 02

   Under an event, the indications it would change. Where the research tied the event to a programme, they are listed. Where it only named the drug in passing, the terminal suggests the link and waits for you to confirm it. Nothing is assumed.

3. 03

   Pick success, failure or delay. Success asks for your post-event probability of approval; failure sets it to zero; delay moves the launch year out by one. The exact edit is written out, with the value now, after, and the change.

4. 04

   It is a preview. Nothing changes until you click Apply to my case, and applying is a single undoable edit however many indications it covers.

Event scenarios are never added up into a total value at risk. Two events can overlap, and a sum of their deltas would count the same drug twice.

1. 01

   Without an account, your edits are kept on this browser and the command bar says so: Unsaved, local only. The Case menu exports the draft as JSON or CSV at any time.

2. 02

   Sign in and the badge offers to keep the draft. Name the case and save it, or press ⌘S. A case is private. Saving does not share your assumptions with anyone.

3. 03

   A saved case reopens to the same number next month and on another device, because the research it used is kept with it. It appears in the rail and on the Watchlist panel.

4. 04

   When newer research exists for a saved case, Thesis lists what changed, fact by fact with its source, and asks. Accepting makes a new revision. The one you saved stays exactly as it was.

5. 05

   A saved case exports as JSON or CSV, every input and result included. If a trial or subscription ends, saved cases stay readable and exportable.

1. 01

   Open Compare and add up to four companies from the search box in the panel.

2. 02

   Read across: stage, price with its quote time, cash, runway, lead programmes, the next dated catalyst, and which case each column is showing.

3. 03

   A field the research does not have stays on the table as unavailable. A blank never quietly reads as zero.

4. 04

   Each column keeps its own research date, method and share basis, so a per-share value in one column is not being normalised against another.

1. 01

   Open the Moderna sample on Valuation. The headline reads Operating DCF, and the bridge has three rows: forecast operating cash flows, terminal value and net cash.

2. 02

   The forecast is the editor you meet first. WACC, terminal growth and each year's cash flow are fields. Change any of them and the headline follows.

3. 03

   A sensitivity grid shows the same headline across nearby WACC and growth rates, so you can see how much of the number rests on the two inputs that are hardest to know.

4. 04

   The drug-by-drug rNPV is still there as a cross-check, with its own drivers. Editing those moves the cross-check, and the panel says so. A WACC at or below terminal growth is rejected, and no number is shown for it.

• "Not valued" on a pipeline row

  The research describes the programme but the model has no line for it. It stays on the list so you can see what was left out of the sum.

• A line excluded from the headline

  A programme missing an input the model needs, a launch year say, is dropped from the sum and named on the headline, with a link to set the missing value.

• A stale quote

  The price is marked stale, with its time. Where there is no price at all, no gap between model and price is shown either.

• A missing financial fact

  It says which kind of missing: not disclosed by the company, or not collected by the terminal. Neither is rendered as zero.

• An invalid draft

  If an edit makes the model impossible, the last valid headline stays on screen, labelled as such, and the field carries the reason.

• A company that needs an account

  Companies outside the 7 samples need a trial. The panel says so and offers a sample instead of a blank pane.

Research

A dated copy of everything read about a company: its facts, their sources and the model's baseline assumptions. It does not change once made. A refresh makes a new copy beside it.

Case

Your named, private version of a company's model: a scenario plus your edits, pinned to the research it was built on.

Draft

Edits that are not saved yet. They are kept on your browser and offered back to you when you return.

Revision

One saved state of a case. Saving again, or accepting newer research, makes the next one. Earlier revisions stay.

Scenario

Bear, base or bull. Each is a full set of assumptions from the research, not a confidence interval around the base.

Line

One programme in one indication. It is the unit the rNPV values, so a drug in three indications is three lines.

Driver

An input that moves a line: peak sales, probability of success, launch year and the rest.

Bridge

The headline broken into the rows that add up to it.

Catalyst

A dated event that could change a line: a readout, an FDA decision, an earnings date, a conference.

Evidence pane

The panel on the right that shows a drug, a line, a disease or a source document beside the model.