Crypto risk · Research note 01
Does Ethereum diversify Bitcoin downside?
The daily 50/50 mix had a deeper drawdown. Covariance explains why adding an asset failed to reduce risk.
2024–2025 sample · 9 Sep 2026
Read the analysis ↗The research library
Follow the argument.
Market risk and company fundamentals, examined through explicit comparisons and source data.
Original research
Equities · Research note 02
Microsoft: strong business, insufficient margin of safety
A $427 base valuation, a $340 entry threshold, and the cash recovery needed to support both.
Year ended 30 Jun 2026 · 9 Sep 2026
Read the analysis ↗Historical notes
The archive, with corrections.
These 41 notes retain their original dates. Unsupported findings have been withdrawn; each page retains its complete correction record. They are not validated performance studies.
41 historical notes
- Market Analysis
2026-02-25
Home Depot Q4 FY2025: why the reporting calendar mattersA shorter reporting period changes the sales comparison. A dated correction separates reported results from unsupported influencer scores.
- Influencer Studies
2026-02-25
SaaS and AI agents: the early-February catalyst window is not establishedA dated correction records the verified Claude Code Security release and withdraws unsupported sector returns, creator tiers and causal re-rating claims.
- Methodology
2026-02-25
NVIDIA earnings: separate results, expectations and the tradeA historical follow-up with primary results, a definition change and three qualitative scenarios. Unsupported event-study returns withdrawn.
- Market Analysis
2026-02-24
Gold and Bitcoin: define the instrument before testing the safe-haven claimA dated correction withdraws an unsupported cross-asset comparison and safe-haven cohort while preserving the original question and source limits.
- Influencer Studies
2026-02-24
Novo Nordisk REDEFINE 4: a clinical announcement is not a verified trading scorecardA dated correction preserves the company-reported trial design and endpoint boundary while withdrawing unsupported call, price and post-trial panel results.
- Market Analysis
2026-02-24
Tariff chronology, 20–24 February 2026: the three-reversal scorecard is withdrawnA dated correction distinguishes signed actions from their effective date and withdraws the unsupported call and hedge outcomes.
- Methodology
2026-02-23
AI infrastructure capex: separate company guidance from an unregistered cycle studyA dated correction preserves Amazon's reported capex and cash-flow definitions while withdrawing unsupported basket, event-study and trading claims.
- Market Analysis
2026-02-23
Software-event claims: the contrarian result needs its registerA dated correction preserves first-party company schedules and a product announcement while withdrawing unsupported price, basket, return and trading claims.
- Market Analysis
2026-02-23
Home Depot Q4 FY2025: the reporting calendar changes the comparisonA dated correction confirms the 13-week/14-week reporting periods and withdraws unsupported macro correlations, price windows and influencer conclusions.
- Methodology
2026-02-22
Earnings-week filters: a calendar is not a verified edgeA dated correction preserves selected first-party reporting schedules while withdrawing unsupported report counts, filter results and risk rules.
- Market Analysis
2026-02-22
Cybersecurity and AI disruption: an announcement date is not a causal return studyA dated correction preserves the Claude Code Security announcement and a company filing boundary while withdrawing unsupported price, cohort and causality claims.
- Market Analysis
2026-02-22
Prediction-market probabilities: preserve the contract before reading the forecastA dated correction separates public macro records from an unregistered prediction-market contract, call cohort and allocation study. Unsupported probabilities, returns and thresholds are withdrawn.
- Market Analysis
2026-02-21
GDP and PCE on 20 February 2026: separate the release, unit and vintageA dated correction distinguishes quarterly SAAR GDP from monthly PCE, records later GDP revisions and withdraws unsupported expectations and influencer outcomes.
- Methodology
2026-02-21
NVIDIA earnings: separate the scheduled event from the missing crowding registerA dated correction preserves the event date and later reported results while withdrawing unsupported sentiment, event-return and predictive claims.
- Market Analysis
2026-02-21
Tariff policy on 20 February 2026: separate the Court opinion from the import surchargeA dated correction separates the Supreme Court opinion, White House actions and effective date, and withdraws an unregistered influencer-call scorecard.
- Methodology
2026-02-20
2026 value rotation: the influencer-portfolio comparison needs a registerA dated editorial withdrawal of unsupported influencer-basket and value-tilt return figures. The comparison method is retained without a performance conclusion.
- Market Analysis
2026-02-20
January 2026 FOMC minutes: separate the decision from the market pathA dated correction fixes the minutes release date, separates policy discussion from market-implied probabilities and withdraws unregistered rate-call outcomes.
- Market Analysis
2026-02-20
Walmart Q4 FY26: reported results are not a consensus beat without the snapshotA dated correction separates Walmart's reported GAAP/non-GAAP results and company guidance from unsupported consensus, price-reaction and event-study claims.
- Market Analysis
2026-02-19
Amazon 2026 capex guidance: separate company records from an unregistered selloff studyA dated correction preserves Amazon's 2026 capex outlook and 2025 cash-flow definitions while withdrawing unsupported price, call-cohort and risk-control results.
- Market Analysis
2026-02-19
S&P 500 concentration: historical weights need a dated registerA dated correction preserves benchmark and index-level definitions while withdrawing unsupported historical weights, forward returns and allocation thresholds.
- Influencer Studies
2026-02-19
VIX and influencer advice: define the volatility observation before testing reliabilityA dated correction preserves the VIX definition and daily close while withdrawing unsupported call cohorts, hit rates, alpha and sizing rules.
- Influencer Studies
2026-02-18
Buy-the-dip claims in a tariff regime: the event study needs its registerA dated correction withdraws unsupported dip-event returns and stopout rates while retaining a proposed evidence log.
- Influencer Studies
2026-02-18
Retail-flow claims: aggregate activity is not an influencer resultA dated correction preserves a scoped retail-investing source while withdrawing unsupported full-year flows, basket returns and causal conclusions.
- Influencer Studies
2026-02-18
SpaceX IPO proxies: later filings do not prove a rumor-window tradeA dated correction separates later SpaceX offering records from an unregistered listed-proxy event study and withdraws its return and timing claims.
- Influencer Studies
2026-02-17
Gold's 2025 record: annual demand evidence is not an influencer auditA dated correction preserves a World Gold Council annual snapshot while withdrawing unsupported January returns, creator samples and move-capture results.
- Market Analysis
2026-02-17
Small caps in January 2026: preserve the benchmark before comparingA dated correction preserves the benchmark definition and a separately dated market snapshot while withdrawing unsupported month-end returns, cohorts and allocation effects.
- Influencer Studies
2026-02-17
Quantum valuation claims: reported revenue is not a verified price-to-sales snapshotA dated correction preserves period-specific company disclosures while withdrawing unsupported multiples, creator counts and scenario returns.
- Market Analysis
2026-02-16
Energy versus AI narratives: the custom rotation test needs its registerA dated correction preserves a separately scoped 13 February market snapshot while withdrawing unsupported sector baskets, cohorts, drag estimates and predictions.
- Influencer Studies
2026-02-16
SEC social-media cases: date the allegation before trusting the claimA dated correction using two SEC releases to separate allegations, source dates and procedural status from unsupported enforcement-trend counts.
- Influencer Studies
2026-02-16
Down-market trading advice: the content scorecard needs its archiveA dated correction preserves neutral market-regime definitions while withdrawing unsupported post samples, outcome gaps, survey context and causal claims.
- Influencer Studies
2026-02-15
Crypto influencer returns vs S&P 500: comparison withdrawn pending a registerA dated editorial withdrawal of unsupported 2025 portfolio-return figures; benchmark and execution definitions are required before comparison.
- Influencer Studies
2026-02-15
AI influencer calls: the account register comes before the scorecardA dated correction preserves dated company disclosures while withdrawing unsupported call samples, benchmark returns, creator rankings and concentration claims.
- Influencer Studies
2026-02-15
FOMO trading claims: use an evidence log before trusting urgencyA dated correction that withdraws unsupported cohort returns and prevalence claims and publishes a neutral urgency-evidence log.
- Influencer Studies
2026-02-14
Trading-account red flags: a neutral claim-audit checklistA dated correction withdrawing an unsupported account/post cohort and reframing red flags as questions to investigate, not labels for people.
- Market Analysis
2026-02-14
February 2026 sell-off: dated inputs are not a trading playbookA dated correction preserves source-scoped index and company-guidance records while withdrawing unsupported stress tests, recovery paths and allocation thresholds.
- Market Analysis
2026-02-14
Copy-trading costs: measure the fill path before claiming a return gapA dated correction withdrawing unsupported mirrored-trade performance findings and specifying the fill register needed for a cost study.
- Methodology
2026-02-14
S&P 500 analogs: the historical similarity result needs its registerA dated correction preserves a source-scoped closing-index record while withdrawing unsupported analog selection, forward-return and forecast claims.
- Methodology
2026-02-14
Trading-signal metrics: the register comes before the resultA dated correction withdrawing unsupported provider and signal-cohort findings while retaining metric definitions as a proposed method.
- Methodology
2026-02-13
How to test a trading claim: the evidence register comes firstA proposed evaluation method with an explicit withdrawal of unsupported cohort performance claims.
- Influencer Studies
2026-02-13
Trading-signal costs: the follower fill register comes firstA dated correction preserves general SEC cost guidance while withdrawing unsupported channel, execution, return and sizing results.
- Market Analysis
2026-02-13
YouTube trading-call claims: audit the register before trusting the resultA dated correction withdrawing unsupported channel/call performance findings and retaining a proposed call-audit workflow.
No matching notes. Try another company or topic.