This site is a personal research terminal. It tracks several hundred names across the desk's watchlists, publishes the desk's research, and keeps a daily board of setups worth acting on — then grades every call and decision after the fact. This page explains how to get around and how to read what you see.

Getting around

• Sidebar — the daily pages first, tools after. Today and Tape sit as single rows at the top; below them, Reading (what the desk published), Book (what's called, held, and graded), then Dig deeper (the research tools — trend map, scans, tickers, watchlists, sources, filings, archive — for when you want to go further than the daily read), and Labs (prototypes and archives — collapsed; click the Labs heading to expand). This Guide lives behind the (?) icon at the bottom of the sidebar, next to the theme toggle.
• Search / jump box (⌘K) — the fastest way anywhere:
  • Type a ticker (NVDA or $nvda) to open that ticker's page.
  • Type a page name or alias (tape, calls, score) to jump to it.
  • Prefixes jump to a specific entity: w: watchlist, p: campaign, u: source, e: event id.
  • Anything else runs a full search.
• Ticker chips — hover any ticker chip on any page for a quick price chart; click it to open the ticker's page.

The Today page

Today reads top to bottom: market weather first, then the board, then your book and the feed. The board is the page's point — everything else on the site feeds it or checks it.

Below the board, Campaigns shows active and monitoring counts and the earliest dated review among live campaigns. Open a count to browse that status, or open the review to read its case. Past dates remain visible as overdue; event conditions can prompt an earlier review. Campaigns also has a permanent link under Reading in the sidebar.

Hovering any ticker opens its preview card; when a name has an earnings print scheduled inside the next two weeks the card says so ("earnings tomorrow", "earnings Wed 7/29") — red when it's today or tomorrow.

The right rail carries Upcoming Earnings — the next seven days of prints, one row per day. Each row shows the mega-cap and focus-list names printing that day plus a count of the day's full slate ("+41 more"), so a heavy week is visible at a glance. It sits above Next Catalysts, which stays reserved for big structural events (IPOs, deals, listings) — single routine prints never appear there.

When data is held

If part of today's data could not be pulled or failed its checks, the Market Snapshot header shows one line beginning Data held. It names the affected area in plain words ("Price pulls (5): …", "Earnings calendar: …") and the reason the desk recorded, and says when today's data run ended with an error. Numbers in a held area stay on their last good value instead of being shown as current. No line means nothing is held.

Today's Board

The board runs one thesis: durable secular uptrends, bought after the breakout confirms, held for six to forty-eight months. A buy-side row is a timing signal against that thesis — it means the entry window for a multi-month position is opening, not that the stock is cheap or good for every style of trading. Two horizons stack on each row: the setup word is the timing horizon in days ("coming into play" and "a fresh setup" mean wait and watch the trigger; "at its trigger", "in its buy zone", and "breaking out" mean the act window is open), and the tactic tag is the holding horizon (hover it — trend-hold is 6–48 months). A holding-pattern row is not a rating: it means the desk has no signal on the name today.

Two columns:

• Buy side — entries, strongest first: held winners extending, fresh breakouts, names coming into play, then desk-call triggers and buy-zone hits. The top five show; "More entries (N)" opens the remaining entries and counts only the rows inside the fold.
• Sell side — what's weak: coverage names breaking down on the tape. If there are more than five, "More sell-side exits (N)" opens the rest.

The board never reads your book. Sell or trim advisories on positions you actually hold are position management, not analysis, so they render in The Book card in the rail ("Exits firing · N"), separate from the board and never ranked against its entries. The card shows up to five; the rest are on Positions.

Every name on the focus list is always on the board. A focus name with no setup appears inside "More entries" as a holding-pattern row — its "why now" states there is no setup and gives the current tape state (trend, week move, distance from the 52-week high). When one of these names develops a real setup it moves up the board and gets a "just flipped" badge; holding-pattern rows are never filtered by the desk read.

Each row shows: the ticker · badges · a trend badge · a plain-language "why now" · the level or trigger price on the right · a tactic tag. The plan badge is the readiness claim: research means a tactic or archetype is still missing, waiting means no settled observation proves an entry, armed means the declared trigger remains ahead, entry fired means it crossed on the settled session, and already ran means the entry has passed without a current-session cross. The tactic tag only names the play and holding horizon; it does not make an entry actionable by itself. Hover the plan badge for its entry reason, risk policy, exit policy, as-of date, and plan ID. Exact shares and account state never appear on the board.

Badges on board rows

• the turn fired — was coming into play, now breaking out. The pop is the signal to watch for the flag, not the fill.
• just flipped — the row's setup changed since yesterday (hover for from → to).
• day N — how long the row has been posted; it turns amber past ~10 days, meaning the setup has been sitting unresolved.
• already ran — price is well through the trigger; chasing here buys the gap, not the setup.
• desk ✓ / desk caution — the desk read's verdict (below).

The desk read

Every board build is reviewed by a model before you see it: it verifies each row's numbers against the tape, then judges fit. Endorsed rows get desk ✓, real-but-read-the-note rows get desk caution (hover for why), and junk names in range get filtered into the fold at the bottom — "Filtered by the desk read" — with the reason, so a wrong veto is visible and gradeable. A "desk-read ✓" stamp next to the board date means the review matches this exact build.

Logging what you did

The site is read-only everywhere. Nothing on the site takes taps or records an action. When you act on or deliberately pass on a name, tell the desk in chat — it lands in the Decision Log. The Decision Log grades whatever gets logged, including passes.

Departures

A "Departed:" line under the board lists rows that recently left — what each was, and when it was last seen.

Trend badges

Ticker lists across the site (the board, watchlist rosters, positions) carry a three-state trend badge instead of just today's green/red:

• coming into play — basing after a decline; the tape is turning.
• in trend — the uptrend is working.
• retreating — rolling over or rotating out.

Hover the badge for the precise underlying state and reason.

Market

• Tape — the current tape: movers, sector strength, where signal events cluster. Signal cards show a short preview; Read signal opens the complete interpretation. The Winners & Losers board switches between daily and weekly returns. Its desk-read label compares each move with flags recorded before that return window opened: Expected means a standing directional flag agreed with the move, Contrary pointed the other way, Watched was on a neutral scan in the preceding week, and Blindside had neither. These labels grade the desk's preparation; they do not explain what caused the move. Held returns and geared/synthetic-income wrappers are omitted; the latter keep their mechanism-aware scan rather than posing as company moves.

Reading

A “reference unavailable” note in an event or research article means its local citation could not be found. Available citations remain clickable.

• Feed — the surprise stream. The top of the list is the unexpected stuff: curated picks, source tweets, the desk's own buys and sells, market signals, and world events — newest first. Use the underline tabs to show All, Briefs, Sources, Research, or Signals. The routine desk output (scan runs, articles, profiles) sits below it, behind the "show more" fold, so the surprising reads never queue behind the mechanical ones. The newest Market Conditions card stays pinned at the top of the stream; a fresh one is built after each trading day, and the replaced cards fall back into the stream in date order — each is a frozen snapshot of its own day, so the trail of cards reads as the climate history. Importance (heuristic) sorts by the type and context of a feed item; it is not an accuracy score. Hide low signal filters lower-ranked items, while Needs review shows raw captures and explicitly unverified items. Links moving through the desk groups source URLs with their dated receipts and links to the feed items that cite them.
• Library — everything the desk has written, newest first, in one table: briefs, company profiles, deep dives, investigations, conviction docs, and articles. Sector research is the shelf for vertical/industry reports; filter by kind with the underline tabs above the table, or search by title or ticker. A brief row opens that date's brief as a single page: the day's narrative (with a table of contents), then the desk's data modules — the market conditions snapshot for that date (frozen, so it stays true on old issues), the top of today's board (the strongest entries and anything weak, on the current issue only), focus list, outliers and movers, the week's catalysts, what moved, the perspective and scan dashboards, desk calls, the alarm board, and the paper report card. Live modules read the latest data and state their own as-of dates. Previous brief and Next brief move between available publication dates, skipping dates with no brief.
• Campaigns — investment claims with their own evidence, notebook, ideas and reviews. Open Campaigns from the main navigation. Filter by status, theme or review state; compare two campaigns to read their claims and horizons together. Themes group related campaigns without combining their claims. Opposing views can remain separate, including views on the same ticker over different windows. The notebook shows rough material as unverified; it does not automatically appear on the Feed. Ask the research agent to create or update a campaign. The website is a reader: filtering and comparing do not change research. A dated desk review can span campaigns and preserves its original call references. Legacy scope badges mean the earlier perspective has not yet been reviewed for this model. An overdue review does not mean the campaign has closed.

Book

The two pages are one sequence. Calls is what the desk said would happen and Positions is what is held right now. How each choice you logged turned out lives in the Decision Log, under Labs — it grades decisions, not holdings, so it sits with the self-reflection tools rather than the book.

• Calls — open desk calls with their trigger conditions and live status. Nothing here is a position yet: these are claims waiting on a condition. Each call also carries the same book-blind plan badge used by Today's Board; only entry fired is an actionable entry proposal, while research, waiting, armed, already ran, unsupported, and plan unavailable remain visibly non-actionable. Hover the badge for the shared plan ID, risk policy, and exit/invalidation policy; exact shares remain outside this page. Lifecycle statuses come from today's tape: APPROACHING, IN ZONE, RAN (the zone played out — re-anchor if you still want in), BROKE ZONE, HOLDING / NEAR STOP / TRIGGERED for stops. Alarm bells show how many independent signals agree on a call. The page opens with Stalking — the highest-conviction names the desk is not in (plus names it passed on), each pointing at the call or target below that would get it in. The homepage Stalking card is a preview of this section.
• Personas — the paper books by cadence, with each book's playbook introduction, results, and dated decisions. Select a name to read its full playbook and decision details.
• Positions — what is currently held, across every book the desk tracks: open positions and since-cost P&L, the exit advisories each position is triggering (SELL, TRIM, WATCH, with the rule that fired), the advisor votes on them, and how the paper strategies are performing. Every row names the book it belongs to, because one ticker can be held in several books at different entry prices — two rows for the same ticker showing different P&L are two books, not a contradiction. Recent paper actions show the recorded execution price and rationale. Opening a strategy's name shows that strategy's own page. Desk-wide entry zones, stops, and target conditions live on Calls; they are not paper positions unless a recorded action opened one.

Dig deeper

The research tools, for going further than the daily read. Every ticker chip across the site already links into them.

• Trend Map — every tracked name in one breadth view, bucketed constructive / transitional / damaged.
• Scans — the run history for every sector and theme scan. Use the category tabs to isolate one history; the selected category stays in the page URL so the filtered view can be bookmarked or shared. Open a run for its full read with price tables and entry zones.
• Tickers — the index of every tracked name; each ticker page has charts, events, research coverage, and watchlist membership.
• Watchlists — the tracked universe grouped by theme; a roster shows each member's tape state and recent events. The ETF Holdings view shows the board's latest holdings date; each fund card shows the snapshot age and marks holdings past the source's expected update window as stale.
• Sources — the accounts and feeds the desk follows, with a track record per source: what they ping, what routed into research.
• Filings — what disclosed money is doing: 13F convergence, congressional and insider disclosures.
• Events — the full event record, day by day: every class including the mechanical items the Feed folds away. Each date is an accordion row — the newest day opens by default; click a date to open it. An Earnings strip above the record carries yesterday's prints, today's slate, and the next five days of scheduled prints, showing only the mega-cap and focus-list names (each row still counts its full slate). A recent day's count ("98 reporting today") is a link — it opens that day inside the record and scrolls to it, so any name, tracked or not, is findable there. "Full schedule" expands the strip into every future date — upcoming earnings, dividends, and logged catalysts, one line per day, soonest first; each joins the record when its day arrives. Recorded earnings and dividend dates also note themselves on their day's row ("+N scheduled dates") and clicking that reveals them. Filtering by kind, direction, or text opens every day with a match (the strip steps aside while a filter is active). The strip's Past prints fold expands into the season's earnings history, newest day first: every recorded print with its beat/miss verdict where graded (verdicts compare reported EPS against the analyst consensus and begin 2026-07-28; earlier prints predate the grader), the desk's one-line read of the night, and a "Closest-tracked / All" filter. A print with no write-up yet says so — the row itself is the record that it happened. Per-kind deep archives (scans, deep dives, candidates, and the rest) open from its foot.

Labs

Concepts is the desk’s working library of market rules and frameworks. Each entry shows its source, confidence, action conditions, and related tools. Proposed tools are labeled Not yet available.

Prototypes, self-reflection tools, and archives: Advisory (suggested actions from alerts × your stated intent), Loop (the autopilot's report card), Decision Log (your decision journal, graded — every buy, sell, pass, and watch scored against the S&P at 1 week / 1 month / 3 months; an account's P&L can't grade a pass, this can, and a skipped winner counts against you same as a bad buy. The desk also grades its own calls there with no logging involved: every board row, desk-read verdict, and exit advisory is scored against the S&P over 7 to 180 days, beside random-entry and momentum baselines — your journal grades you, that table grades the desk), and Candidates (setups that cleared a mechanical screen but haven't earned a desk call), and The Read (a prototype of a shorter front page: four blocks — world, rotation, act on, money — each capped, each line opening to its detail; a block that has no sound data says so instead of showing an older value). Frozen copies of retired pages still resolve at their -old URLs but are not listed.