Start with the context.

High impact surfaces hard company events plus major inflation, labor, Fed and policy catalysts. Notable catches analyst actions, launches, restructuring, growth data and fast-moving market context; everything else stays Context. Freshness, source quality and corroboration can move a story up. For articles, the green / red direction chip is read only from the headline wording. For a published economic print it stays Unclear, because hotter, cooler, stronger or weaker data is factual evidence — not a universal bullish or bearish verdict.

Active tape means a related ticker or broad index is moving materially or trading unusually heavy at the same time; it does not prove the headline caused that move. Use it as a verification queue: open the source, confirm the event is new, then check whether price and volume agree. Carried-forward context is labeled, unconfirmed stories are labeled, and exact duplicate links are collapsed without removing distinct coverage of the same event.

A narrative tells you where to look, not when to enter. Start with the trade posture and its first invalidation check, then confirm the individual ticker in the Grade tab before risking capital. Sector and story cards keep lifecycle, fundamentals-vs-hype, bull/base/bear scenarios, industry grades, and source evidence under expandable detail so the decision state stays visible first.

Share ideas, not option contracts — the Top Picks tab times leveraged trades; this page runs one buy-the-dip playbook over the same ~138-name universe, built on three separate questions answered independently (never blended into one number). 1 · Is it a good business? A hard quality gate: consistently profitable (positive net margin or free cash flow), a manageable debt load (more cash than debt, or debt/equity ≤ 2x), net margins holding vs a year ago, and revenue still growing on a trailing-twelve-month view. Names that fail are never shown, however far they’ve fallen — that’s how value traps get in. 2 · Is it beaten down right now? Five reads of “cheap vs its own recent self”: RSI(14) under 35, 4%+ below the 50-day average, 15%+ off the 52-week high, stretched −2σ against its 20-day mean (≈ the lower Bollinger band), and lagging SPY by 4+ points over ten sessions (company-specific selling, not a market-wide selloff). Each read is z-scored across the quality-passed universe and averaged into the card’s dip score, so the page surfaces the most unloved names relative to each other rather than leaning on fragile fixed thresholds; a name needs at least two reads fired to list at all. 3 · Is it down because something actually broke? Yellow trap flags — a fresh earnings print inside the drop, heavy-volume selling, a long red streak, analysts cutting estimates, a bearish news tone, or a binary event just ahead. Flags never block a candidate; they ride the card so the final call stays with you. The execution read then separates Start small (clean dip plus a positive close and improving RSI), Wait for turn (clean but still deteriorating), and Research first (one or more trap flags). Every card names the current entry or confirmation trigger, the nearest structural level that forces a thesis review, the first mean-reversion objective, and the reference payoff between them. The review level is not an automatic stop — long-horizon owners must re-underwrite the business there. Every card also carries an expandable investment thesis checklist — the full owner’s due-diligence list (management & moat, financial health & cash flow, unit economics, valuation & growth, macro sensitivity, risks & scenarios) with each question answered from the tracked data where possible and honestly labeled unsure (a heuristic or proxy read) or unanswered (not visible in our data) where it isn’t. Fully deterministic, recomputed at 11:00 and 15:30 ET, and honest — a tape with no quality name on sale shows nothing. Not financial advice.

This tab is a long-only quality-recovery screen for peer washouts — not a sector-ETF allocation model, the Market Tape offense/defense gauge, or the Heatmap breadth alert. Start with the Quality Recovery Shortlist, then use the detailed board to answer five questions in order: is the business intact, why was it beaten down, what evidence supports recovery, is valuation reasonable, and has price confirmed an entry with acceptable payoff? Qualified recovery means the profile’s required evidence clears; Verify first remains non-actionable until incomplete or uncertain business and forward evidence is resolved; and Rejected profiles stay in the near-miss research section rather than the candidate board. Each card shows the underlying profitability, balance-sheet, revenue, margin and coverage evidence when supplied, separates a sourced positive recovery driver from a statistical mean-reversion case, labels missing valuation rather than assuming a lower price is cheap, and keeps thesis risks visible before the trade plan. The rotation model then tests whether a strong company fell with a broadly damaged peer group instead of because its own business broke. For each survivor it freezes the pre-drop trend mean and robust standard deviation at the episode peak, requires the trough close to reach at least −1.5σ, and measures how much of that dislocation has reverted without allowing the mean to drift down toward price. Washed out is extreme but unproven, first thrust waits for a controlled pullback, confirmed requires group and stock follow-through with runway left, and late means the edge is mostly spent. The plan uses the earlier of the frozen mean and structural resistance as its first target and invalidates below the actual trough; no score, z-score, or live quote can upgrade a Verify-first or Rejected recovery profile into an actionable trade. The accountability ledger enrolls only the first baked ready signal confirmed by a recent in-zone regular-session quote, then freezes entry, stop and target. Live quotes can update price, payoff and sizing between bakes, but not the baked business evidence or statistical classification. σ is context, not a probability or a promise that price must revert. Not financial advice.

Start with Enter now versus Wait, then read the card’s underlying entry, invalidation and first target before looking at the leveraged ticker. The risk planner converts your account-level loss cap and the card’s estimated ETF stop width into a maximum share count once the live quote arrives. Levels are deterministic and live on the underlying; the ETF percentages are only a same-day leverage translation, so gaps, daily resets, spreads and tracking error can make the actual exit worse. The screen maps the grade engine’s trend / flow / fundamentals / narrative read onto verified listed products, requires breadth for sector trades, strips the options-only IV-cost pillar, and never invents a missing vehicle. Reset drag, carry, earnings risk, tape alignment and the simulated path remain on every card because a good direction with a bad vehicle or hold period is still a bad trade. Short-horizon trading tools only; not financial advice.

Deeper green and red mark larger session moves. In Relative volume mode, saturation shows current cumulative volume versus the fraction of a normal 20-day session expected by that clock time, using the same U-shaped intraday pace curve as the Volume desk. That keeps the morning comparable with the close instead of labeling every incomplete session quiet. Pre-market keeps the last completed-session read rather than treating prior-session volume as live. Hue still shows direction. Group by sector or industry, press Enter after searching to center the first match, and use the zoom controls or wheel/pinch to inspect the small-cap tail.

The Sector breadth streak flags a group only after at least 70% of its tracked names have closed in the same direction for two or more consecutive sessions. That proves participation is broad and persistent; it does not measure ETF inflows/outflows, relative strength versus SPY, or the Owner Sector Rotation desk’s quality-washout rebound setup. Use a green streak as a leadership candidate and a red streak as a group-risk flag, then confirm with relative performance and volume. Not financial advice.

A month-at-a-time view of every dated market event, opening on the current month — use ‹ / › to step between months (or Today to jump back), and tap any day to see its full details below the grid. It tracks: confirmed earnings dates (with AM/PM session tagging) for every curated ticker, cross-checked against a rolling 21-day Nasdaq sweep; ticker-specific catalysts (FDA dates, contract decisions, product launches, court rulings, investor days — extracted from recent news); the complete official BLS and BEA release schedules; the Federal Reserve Board calendar (speeches, testimony, minutes, conferences, and statistical releases); the Kansas City Fed’s Jackson Hole symposium; detailed market-moving reports with Actual / Previous / Consensus values; and upcoming FOMC decisions. Official source labels open the publisher’s calendar. Effective Fed Funds, FedWatch probabilities, the official vote map, and the full rate path live together in Bonds & USD. Ticker chips are clickable.

The evidence ladder is Rumor → Reported talks → Definitive agreement → Regulatory / vote pending → Expected close. These are public-source labels, not hard completion probabilities. For cash deals, consideration per share is the stated offer; stock and mixed consideration can move with the buyer’s shares. Premium to the unaffected price is shown only when a cited source states it. Estimated equity value is consideration per share multiplied by current implied shares outstanding—an estimate, not enterprise value. Rumors keep undisclosed price, premium, spread, and close fields blank. A wide deal spread can reflect regulatory, financing, timing, or break risk—never assume it is free money.

Start with the Event posture, which translates the current season into exposure risk: how many tracked names report next, how many already carry a double-digit pre-print move, whether positive reaction breadth is holding, and whether realized gaps are exceeding what options priced. The Event desk pairs each report session with its decision deadline and the straddle-implied move when a real final-week snapshot exists; it is a gap-risk screen, not a directional call. A crowded run-up can raise sell-the-news risk; a heavy selloff can mean either washed-out positioning or expectations that are still deteriorating. Neither label is an automatic fade. Use the report session to decide whether a position carries overnight event risk, then define the maximum loss before holding through it.

Below that, each season groups prints announced in one calendar quarter (Jan–Mar reports cover fiscal Q4, Apr–Jun cover Q1, and so on). The scoreboard separates EPS results, guidance, pre-print drift, the first regular-session reaction, the options market’s expected move and one-week follow-through. The implied move is the straddle-implied ±% snapshotted in the final week before a print; it estimates magnitude, not direction. Guidance and implied-move coverage accumulate going forward, so older seasons can be incomplete.

Start with the decision queue: it prioritizes explicit outlook changes, misses, management caution and disputed Q&A instead of asking you to scan every transcript equally. Open a company to separate what changed from what must be watched next, then use the research disclosures for the reported numbers, full guidance table, management wording, analyst questions, operating detail and source transcript. A constructive call is not an entry signal and a cautious call is not automatically a short — confirm the Grade view, price reaction and valuation before acting. Every brief is AI-generated from the full transcript and can contain errors.

When a company reports earnings, its same-sector peers move too — this matrix measures that read-through across the whole tracked universe, grouped into sector complexes (semis & memory, software, banks, payments, consumer, healthcare, space, …), each with its sector ETF. For every driver→follower pair inside a group it estimates the event-window beta (how much of the driver's print-day move the follower echoes, Newey-West significance, shrunk toward the pooled sector beta on small samples), the direction hit rate, and whether the follower's options already price the echo (its ATM implied move vs its realized average). Pairs must clear fixed statistical gates (R², significance after a false-discovery correction run across every sector's pairs, ≥60% direction consistency) to count as qualified. Upcoming driver events show both engines' expected follower moves — sector-routed (via the ETF) and direct-pair — with the running forward accuracy of each. This is a correlation map, not a trade signal: nothing here is a recommendation to buy or sell anything.

Statistical screens quants actually run, over the same data the rest of the site already collects. Regime conditioning — every build first classifies the tape on four axes (volatility: VIX + term structure + SPY realized vol; trend vs range: SPY efficiency ratio + higher-highs/lower-lows; risk-on/off: the Market Analysis tape; earnings-heavy vs quiet: share of the universe reporting inside ~2 weeks). The single-name sigma screen keeps a fixed 3σ bar in every regime; VRP “rich” needs a bigger z when high vol makes fat premium normal, pair-spread “stretched” widens in high vol and tightens in calm tape, and term-structure inversions are down-weighted through earnings-heavy stretches. Rows are never hidden by regime — only badged and re-ordered — and the strip at the top shows the bars in force. Aggregate ideas — a confluence table cross-referencing four independent flow screens the site already runs (the session’s largest unusual-options prints, the top intraday volume / S–R-break flags, fresh ≤3-session price streaks, and 5-day rising / surging IV): a name showing on two-plus screens ships as a candidate, three-plus is badged qualified, and a lean is reported only when the directional screens agree. Sigma deviations — names at or past their 3σ Bollinger band (20-day price z-score) or printing a 3σ daily move vs their own trailing volatility, with the option market’s expected move (1σ/2σ = S × IV × √(days/365)) beside the realized one. Vol risk premium — each name’s ATM ~30-day implied vol minus its 30-day realized vol, z-scored against the name’s own derived history: persistently positive is the premium option sellers harvest; an extreme z flags premium unusually rich (or cheap) vs that name’s norm. Pairs — within-industry pairs whose daily returns correlate ≥0.60, watched on two spreads: the hedged price spread lnA − β·lnB, with β from a one-year Engle-Granger regression whose residual ADF test (vs the MacKinnon 5% bar) supplies a cointegrated badge — read on both a 60-day and a ~1-year horizon (some pairs only mean-revert on one), with a rolling hedge-ratio drift check, a corr-stability-across-lookbacks badge, and a factor match grade (SPY-beta / size / momentum gaps + a liquidity floor) — and the implied-vol spread vs its 120-day norm (relative options mispricing between peers). Vol surface — term-structure slope (~90d vs ~30d ATM; inverted = near-term stress, badged when the name’s own print is inside ~5 weeks — event vol loading the front is expected) and 25Δ put−call skew per name; their z-scores activate automatically once enough surface history accumulates. Dispersion — an implied-correlation proxy from SPY’s IV vs the cap-weighted basket of tracked large-caps: high = index options rich relative to single names. Post-earnings drift — names inside two weeks of a print, their reaction and drift so far, against their own historical beat/miss drift tendency. Everything is deterministic — fixed formulas, documented windows and threshold tables, no AI and no cross-sectional curve-fitting.

Not feasible with current data (listed honestly rather than faked): M&A / spin-off screens (no deal data source), index add/delete prediction (no committee or flow data), gamma-scalping simulation (needs intraday delta-hedging data), Johansen / multi-year cointegration (the price history carries ~1 year of bars, so the Engle-Granger test runs on a single 1-year window and is labeled as such; Johansen adds nothing for 2-asset pairs), stock-borrow availability / borrow-fee screens (no borrow data source), and alt-data earnings nowcasts (no satellite / card-spend / web-traffic feeds). Already covered elsewhere: earnings read-through pair betas live in Event spillover, season-wide earnings stats in Earnings tracker, IV momentum in Trending IV, and the quality-dip shares screen in Stock Picks — this tab links to them instead of duplicating them.

Use the month grid as a cross-asset participation record: SPY, QQQ, IWM, SMH, DIA and VXUS show whether equity strength is broad or concentrated; TLT, GLD and VIX add rate, defensive and stress context. Click a day to compare all nine instruments. Beneath it, the First & last hour tracker logs SPY / QQQ / IWM from regular-session 30-minute bars: previous close→open gap, 9:30→10:30 ET return/range/volume, 10:30→close recovery or continuation, 3:00→4:00 ET return, open→close result, and whether the day’s high or low first printed in the opening hour, middle, or closing hour. Conditional cards answer what followed an opening move of at least your chosen threshold; gap and prior-close VIX filters add regime context. Green means up except for VIX, where green means volatility eased. Today begins populating after 10:30 and finalizes after the close. Not financial advice.

Cross-market lead-lag signals. Asian cash markets and FX trade and close before the US opens, so an overnight move in a foreign peer is a leading read on its US counterpart — Samsung & SK Hynix selling off in Seoul flags memory names like MU; a yen-carry unwind in Tokyo flags broad US risk. Use 24H / 5D / 20D to separate a fresh overnight shock from a developing swing trend, then click any market card for its path, convention, source, and linked US names. The Japan-carry desk combines official Japan 2Y/10Y yields, the US-minus-Japan 10Y spread, AUD/JPY and EUR/JPY, and an effective-rate differential (EFFR minus the BOJ overnight call rate). Currency strength is a simple equal-weight relative comparison across USD, JPY, EUR, AUD and GBP. The USD/JPY volatility tile is a clearly labeled near-30-day FXY ATM-straddle proxy because the institutional OTC/CME benchmark requires licensed distribution.

Beyond tech, commodities and rates drive their own sectors: crude (energy & fuel-heavy logistics), copper (industrials), gold/silver vs the dollar (metals), nat gas (power), long yields (banks, homebuilders, TLT) and bitcoin (crypto-levered names). Markets that close before the US open are genuine leading reads; concurrent (Europe, cash VIX/yields) and 24h tiles are tagged as co-movement, not a lead. The latest completed sessions refresh from the fixed live market feed while this tab is open; longer histories and correlations remain tied to the verified build snapshot, so different market clocks can still carry different as-of dates.

Each name's constant-maturity ATM ~30-day IV is interpolated between surrounding expirations in total-variance space, which reduces false jumps when the nearest contract rolls. The current read is compared with that name's own daily history using both classical and median/MAD z-scores; the more conservative agreeing magnitude is used, plus percentile, a 5-session ramp, lighter 20-session context, and a short rising streak. Tiers require both statistical elevation and a real percentage expansion in premium. Histories under 60 comparable sessions are labeled provisional and must clear stricter score and z bars. Rising IV says magnitude, not direction, and overlapping momentum inputs are deliberately down-weighted so one climb is not counted three times. Names already rich but no longer climbing remain marked Elevated.

Daily moves first pass an adaptive noise floor based on the ticker's own recent median absolute move (bounded at 0.08–0.35%), so microscopic closes do not create streaks. Counter-day and cumulative-tolerance break bars also scale with that volatility, within conservative bounds. A same-direction close partially heals the tolerance bank and one counter day rather than erasing all prior damage. The displayed streak count is the number of same-direction days; when tolerated counter/flat days occur, the sparkline may span a longer session window. Cumulative return is compounded, not summed. “Just snapped” only includes a material prior run (3+ same-direction days plus length or move significance) with a meaningful reversal, which keeps ordinary two-day noise out of the mean-reversion queue.

A 0–100 sentiment gauge built by CNN from seven equally-weighted indicators of US equity-market psychology. Low readings (extreme fear) have historically preceded rebounds; high readings (extreme greed) often mark overheated conditions. Refreshed each build from cnn.com/markets/fear-and-greed.

Separate infrastructure demand from investment quality. Rising aggregate CapEx can support chip, networking, power, and data-center suppliers, but it can also pressure the buyers’ margins and free cash flow when spending outruns revenue. Start with management’s latest full-year outlook, then compare SEC-reported CapEx growth with revenue growth and the current run-rate before checking supplier backlog/guidance and each ticker’s Grade. Guidance definitions differ by company and remain separate from reported cash CapEx; fiscal years also differ.

Start with the trade posture, then require upstream and downstream confirmation. Wholesale DRAM spot usually moves first; US retail shows whether the pressure is passing through to finished DDR5 kits. Retail confirmation uses the median move and breadth across reasonably stocked categories because a scarce configuration can distort the all-kit composite. Tightening favors memory-supplier pricing-power research; easing can relieve system-builder input costs. Verify the move against supplier guidance, relative strength, earnings risk, and each ticker’s Grade before acting. Spot data: TrendForce / DRAMeXchange; retail data: WhereIsMyRam.

First choose one accelerator so every number on the desk refers to the same GPU model. The chart is daily rental-price history in USD per GPU-hour: the thicker line is the cross-provider median and the thinner colored lines are individual provider / market lanes. A falling line means compute is getting cheaper; a rising line means it is getting more expensive. Use Spot to inspect interruptible capacity, On-demand for standard rentals, or All for the broad blended read. The 7-day and 30-day cards compare like-for-like daily snapshots; “Building” means not enough calendar history exists yet. These are rental list prices, not chip selling prices or provider revenue, and a posted rate does not prove capacity was available. Vast.ai rows are medians of verified rentable offers; CoreWeave, Runpod and Lambda rows are provider-published prices.

Use the two series for different questions. Global net demand is the World Gold Council / Metals Focus quarterly estimate of purchases minus sales, including activity that may not yet be publicly reported. Country holdings are official reported reserves compiled from IMF IFS and central-bank sources; reporting dates vary and can lag. The country table shows each bank’s latest known tonnes, gold’s share of its reserves, and changes over roughly three and twelve months. A rising total supports the structural gold-demand backdrop, but it is not a short-term entry signal for gold or miners — confirm price, real yields, the dollar, valuation and each ticker’s Grade.

Search attention is a participation signal, not a buy signal. Start with the 7-day change to find themes entering public attention, then check whether price, volume, catalysts and fundamentals confirm the move in related assets. A surge without market confirmation can be curiosity, controversy or late-cycle crowding. Interest is Google Trends’ relative 0–100 index, normalized here against the same stock market anchor so themes can be compared; it is not absolute search volume.

Start with the trade posture, then separate input-cost pressure from demand. Gold and silver can reflect the dollar, real yields, safe-haven demand and, for silver, industrial demand; rising cocoa or coffee can hurt exposed buyers, while rising Baltic Dry can signal firmer industrial demand. The desk compares market series over 30 days and monthly series month over month so a noisy one-day move does not outrank a persistent trend. Aged observations are excluded from the decision desk but remain visible in detail. Use the impact label and linked tickers to confirm whether the move is transmitting through relative strength, volume, pricing actions, miner economics, and margin guidance before acting. Items marked proxy track an ETF rather than the native spot benchmark.

Start with the verified event mix, then review the newest financing terms before changing exposure. Share issuance expands the float; convertibles can add both leverage and future dilution; straight debt raises interest and refinancing risk without automatically diluting shareholders. Buyback authorizations are only intent until the company actually purchases shares. Only headlines that explicitly name the issuer and transaction enter the risk totals. Ambiguous ticker associations and insider-sale wording remain in the review ledger. Amounts shown from SEC filings are context and may describe a different reporting period, so confirm the offering price, coupon, maturity, conversion terms, share count, and use of proceeds at the linked source.

Start with the upcoming IPO calendar for the scheduled date, expected raise, proposed share-price range, and sector, then use the trade posture to judge the broader funding window. Dates and terms can change until pricing. Bond issuance shows whether companies can borrow, but an investment-grade-led month is not the same as broad access for riskier issuers. IPO counts are adjusted to a quarter run-rate before comparison with the completed prior quarter, and SPAC share is shown because a SPAC-heavy calendar can overstate conventional equity appetite. Bank deposits indicate funding stability; revolving and NY Fed balances describe household leverage, not immediate default stress. Confirm an open window with high-yield participation, improving ex-SPAC IPO pace, and stable deposits. Treat filing counts as activity rather than unique companies, and inspect the underlying sleeve before changing exposure.

13F: start with share-direction breadth. A stock added by many managers is stronger evidence than a large dollar change from one manager; dollar value also includes price drift, and filings can arrive 45 days after quarter-end. Form 4: code P open-market purchases are separated from code S sales and from grants, option exercises, gifts and tax withholding. Insider sales can reflect diversification or taxes, so no filing supplies an entry by itself. Open the ticker in Grade and confirm the current trend, catalyst, valuation and risk.