FILOMAP — THE SIGNAL JOURNEY
01 RAW FILING
A 13F lands at the SEC.
An institution discloses its quarterly holdings — rows of CUSIPs and share counts. Public, mandatory, and almost unreadable. Just one of
02 SIGNAL DETECTION
The filing becomes a Signal.
A detector resolves the raw rows — entity, ticker, action — into one canonical object. Plain English, anchored to the source document. This is the moat object.
03 CONVICTION ENGINE
One conviction score.
Breadth, recency, position size and historical accuracy collapse into a single number on the Signal — a result, not a panel of dials. Strong rises; noise fades.
04 DASHBOARD RANKING
The Signal takes its rank.
The Intelligence Layer ranks Signals by conviction, recency, reinforcement, capital and overlap. The strongest rise to the top of your morning read.
05 USER ACTION
You follow it.
One tap creates a follow relationship — to the Signal, the investor, or the ticker. Tracking, not owning. The Signal stays canonical; your interest is a reference.
06 ALERT ENGINE
A change fires an Alert.
When the Signal's state changes, an Alert fans out to everyone following it — carrying the source filing and both clocks. The Alert is a notification about the Signal, not a new truth.
07 OUTCOME ENGINE
Measured against what happened.
The Signal's clock starts when the event occurred — not when it was filed. 30, 90, 180 days later the outcome is recorded. Leakage-free, sample size shown. History, never a prediction.
08 COHORT INTELLIGENCE
The cohort learns.
The outcome updates the base rate of every similar Signal — its cohort. The engine learns which patterns hold and which don't, across thousands of resolved Signals.
09 COMPOUNDING LOOP
Intelligence compounds.
Sharper cohorts raise the accuracy of the next conviction score. Filing to Signal to conviction to outcome to cohort to conviction. The moat is not a feature — it's a loop that's harder to catch with every filing.