Skip to content
Whale Atlas — Follow the signal.
Home
FULL RECORDS
Search insider filingsAll Congress filings
MORE TOOLS
WatchlistGold IntelligencePeopleCharts & technicalsCompany financialsOfficial newsMy saved workspace
YOUR RESEARCH
HomeSearch insider filingsAll Congress filings
More research tools
ExploreGold IntelligencePeopleInvestigate an ideaCharts & technicalsMore evidenceCompany financialsOfficial newsMy research & helpWatchlistMy saved workspaceLearn the basicsHow signals workData & coverage
RESOURCES
LearnMethodologyData sources
Built on public evidence.

Primary sources. Clear reasoning.
No black boxes.

Our approach
Public research workspacev1.0

Loading data…Scanning SEC filings…Checking Congress reports…Following the signal…

HomeSearchCongressWatchlist
© 2026 Whale AtlasFree public-filing research. Not financial advice. By using this site you agree to the Terms of Use.
FAQCompare trackersMethodologyData & correctionsTermsPrivacy
NO BLACK BOXES

How Whale Atlas reads the evidence

Every score is built from public filings, with visible weights and an honest account of what’s missing.

Real purchases onlyGrants, gifts, option exercises and sales never count as buying.
Known at the timeA trade counts only after its filing became public.
Missing ≠ zeroGaps stay visible as gaps, never as “no activity”.
  1. Public filingSEC Form 4, House & Senate reports
  2. ClassifiedPurchase, sale, grant, gift…
  3. ScoredVersioned, repeatable rules
  4. Shown with evidenceEvery point links to a filing

The Signal Score, piece by piece

Seven factors, 100 points in total. Model insider-experimental-1.1. Factors marked “up to” are scaled down for older trades by the recency curve below, then rounded — a company’s shown points can be less than the maximum.

Purchase valueup to +25

How much was bought, weighted by how recent the trade was.

Distinct buyersup to +20

How many separate insiders or buying groups bought, weighted by recency.

CEO purchaseup to +15

The chief executive bought shares, weighted by how recent that purchase was.

Cluster buying+15

Two or more independent buyers within 30 days.

CFO purchaseup to +10

The chief financial officer bought shares, weighted by how recent that purchase was.

Holdings increaseup to +10

A purchase grew the insider’s stake by 10%+ or opened a new position, weighted by recency.

Repeated purchases+5

The same buyer came back on another day.

What counts as a purchase

  • Open-market or private purchases of common stock (SEC code P); the code does not by itself prove the venue
  • Separate buyers merged when they report the same holding
  • Grants, awards, gifts and option exercises
  • Disclosed trading plans (Rule 10b5-1) and dividend reinvestment
  • Preferred shares, debt, derivatives and unclear securities
  • Amended or replaced filings, until they’re reconciled

Recent trades weigh more

Share of a purchase’s value that counts, by trade age.

0–14 days
100%
15–30
90%
31–60
75%
61–90
55%
91–180
25%
180+
0%

Value tiers award 2, 6, 12, 20 or 25 points at any, $25K, $100K, $500K and $1M of weighted buying.

Congress is a small confirmation

Each distinct member with an eligible stock purchase in the last 180 days adds weight, capped, and it is only 5% of the combined score. Congressional trades alone can’t create a score. Amounts stay as the reported ranges, and a spouse’s trade is never relabeled as the member’s own.

The Home page’s Congress list is a separate, shorter-window view: it ranks the largest Congress trades filed in the last 60 days, for browsing recent activity. It doesn’t use the 180-day window above, which is only for this score.

Evidence alignment labels

Combines insider trades, Congress trades and daily price momentum using explicit rules. Version 1.1.

Full bullish alignment

Insiders and Congress are both buying, nobody eligible is selling, and daily momentum (RSI 50–70, MACD above signal and zero) agrees.

Bullish · Partial

Momentum agrees with at least one buying source and nothing opposes it, but the full three-way rule isn’t met. The card shows what’s missing.

Mixed evidence

The sources disagree or don’t line up. An oversold or overbought RSI on its own never counts as confirmation.

Leaning bearish / Bearish

Momentum is weak (RSI 30–50, MACD below signal and zero) and one or both groups are selling. Any opposing purchase blocks the label.

Not enough evidence

A required feed, a clear security match, or fresh price history is missing. Missing data is never treated as a zero.

“Full” means every rule lined up across the sources we have, not complete market coverage. The strict screen can correctly return no matches; we never pad it with neutral stocks.

See the strongest signals

Honest limits

  • These are research rules, not predictions. They haven’t been proven profitable, and no success rate is claimed.
  • People buy and sell for many reasons. A filing doesn’t reveal why.
  • Disclosures can be late, incomplete or amended. Congress coverage is partial (Senate: last 12 months of electronic reports).
  • Public officials are not rated or ranked. Always check the original filing.
Explore the data sources
Combined research score (private research mode)
60%Insider activity
35%Technical setup
5%Congress confirmation

Combined score = available component scores × their weights ÷ the sum of available weights. Missing components are left out, not scored as zero. Example: insider 80 and technical 60 with no Congress data gives (80 × 60 + 60 × 35) ÷ 95 = 73, shown with 95% coverage.

On the public site, stored price history isn’t used: public scores come from filing evidence only, and prices and charts are shown by TradingView.

Technical setup factors
FactorWeightPoints awarded
Long trend30Close above the 200-session average
Trend alignment2050-session average above the 200-session average
Medium trend10Close above the 50-session average
MACD momentum1515 for a positive histogram, 8 for a rising negative one
RSI range1515 at RSI 45–70, 8 above 70, 0 below 45
Extension10Close within two ATR of the 20-session average

Needs at least 50 valid daily bars; with fewer than 200, only the factors that can be calculated count and the score is labeled partial. RSI and ATR use 14-period Wilder smoothing; MACD is 12/26/9. Stale bars, gaps or suspicious price jumps switch technical scoring off.

Timing, exclusions and fine print
  • Windows. Alignment uses disclosures from the last 30 days covering trades from the last 90.
  • Freshness. Prices must be no more than two weekdays old. SEC data older than 48 hours, or Congress imports older than seven days, block their feed.
  • Point-in-time. Source files, hashes, retrieval times and model versions are kept. A trade only counts from the moment its filing was public and received; old trades found today aren’t backdated into earlier snapshots.
  • Amendments. Amended filings are held out of scoring, and the originals they may replace are excluded once the amendment is public.
  • Coverage. “None found in imported reports” is not proof nobody traded. Companies listed in the SEC directory without imported filings have no score. See current coverage.
  • Validation. Observations are frozen daily for future testing (5, 10 and 20 sessions, costs included). No outcome statistics are published yet.
  • Further reading. Fidelity on RSI, Fidelity on MACD, and research on congressional returns.