Trading Places

See what the informed money sees.

The big boys leave tracks. Time to follow them.

Here's how ↓
The Database

Seventeen years of market behavior, quantified.

Every insider trade, every institutional position, every lobbying filing, every corporate disclosure. Parsed, structured, and cross-referenced. The system watches what people do with their money — not what they say.

$0
In tracked insider transactions across the database.
0
Institutional position snapshots from 199 quarters of 13F filings.
0
SEC filings parsed and stored.
0
Unique insiders tracked across 13,049 companies.
0
Coordinated buying clusters detected in historical data.
0
Behavioral fingerprints built from historical clusters — 47,242 with known outcomes.
The Thesis

Information moves through markets in layers.

By the time a story hits the news, three groups have already acted. The insiders who run the company. The institutions managing hundreds of billions. The politicians and lobbyists who see policy before it's announced. When those layers move together, the pattern becomes worth examining.

Layer 1 The Insiders

Company executives and directors are legally required to disclose their trades within two business days. When multiple insiders at the same company start buying — or selling — in close succession, that's the earliest visible signal. The system clusters these filings and compares them against historical patterns.

87,141 clusters detected · 47,242 with resolved outcomes
Layer 2 The Institutions

Hedge funds and investment managers disclose their positions every quarter. The filings are delayed, but the scale is enormous — this is where serious money confirms or refutes what the insiders are doing. When institutional flow aligns with insider clusters, signals strengthen. When they diverge, something unusual is worth understanding.

57,158,629 position snapshots · 13,466 institutions · 199 quarters
Layer 3 The Capitol

Members of Congress disclose their trades under the STOCK Act. Corporations disclose their lobbying activity by bill and issue. The system tracks both — and watches for the pattern where a company lobbies a specific issue, the relevant committee members trade the stock, and the policy announcement follows weeks later.

26,008 congressional trades · 152,575 lobbying filings · 10,323 clients
The System

A system, not a prompt.

Most AI trading tools ask a model to search the internet and form an opinion when you type a question. Trading Places is different. The work is already done before you ask.

Always running. The system doesn't start when you ask a question. It's been working since before you logged in — collecting filings, scoring patterns, comparing against 17 years of historical data. Continuously. While you sleep.

Already computed. When you look up a ticker, you're not waiting for an AI to form an opinion. The cluster was already detected. The pattern was already matched against 85,000 historical cases. The base rate was already calculated. You're reading results, not requesting them.

Data first, AI second. The engine underneath is pure math on public filings. Deterministic. Reproducible. Every number traces back to a specific SEC document you can verify yourself. AI interprets the results in plain language — but if the AI disappeared tomorrow, the engine would still run.

Gets smarter on its own. The system tracks every signal against its real-world outcome. Base rates get sharper. Thresholds get tighter. No manual tuning. Just outcomes.

The Agents

Under the hood, a team of specialized agents handles the work. Each has one job, and they run whether you're watching or not.

The Butler
Data Collection

Runs on a schedule. Pulls fresh filings from SEC EDGAR, price data, news articles, and disclosure databases. Never sleeps, never improvises — just delivers data on time, every time, to the rest of the team.

The News Hound
Coverage Tracking

Monitors how stories break and spread. Tracks the velocity of coverage across outlets and flags when a ticker goes from silent to saturated — a transition the system compares against historical cases to show how similar patterns have played out.

The Analyst
Investigation

When a signal fires, the Analyst investigates. Pulls the relevant filings, the matching historical patterns, the institutional flow, the news context. Produces a structured briefing — what happened, what's similar, what typically follows.

The Auditor
Self-Correction

The conscience of the system. Tracks every signal's outcome. Identifies false positives and missed calls. Proposes calibration adjustments when the data disagrees with the model. Keeps the system honest against itself.

The Broker
Your Interface

Your agent on the outside. Watches the tickers you care about, reads the briefings the Analyst produces, and translates everything into plain language when you ask. You talk to the Broker. The Broker talks to the system.

Case Files

Three patterns the system recognizes.

These are pattern examples drawn from public SEC filings. They illustrate how the system cross-references disclosures that anyone can read — but almost nobody does. Educational examples, not forecasts.

Case 1 · The Divergence

When the narrative and the insiders disagree.

A recurring pattern in the historical dataset: coverage saturation across major financial outlets, retail enthusiasm cascading through social platforms, and sometimes high-profile political endorsement — all while the people running the company quietly sell. The system is designed to surface exactly this kind of situation, where the narrative and the insider behavior aren't telling the same story.

Narrative Layer
Coverage velocity spikes to Tier 1 saturation. Major financial outlets running stories within hours of each other. Social media amplification cascading through retail-heavy platforms.
Influence Watchlist
Sometimes a sitting official, senior appointee, or other high-visibility actor publicly praises the company. Historically, executive branch praise of named public companies is uncommon enough to flag — the system watches for it because of the pattern it tends to precede.
Insider Layer
Meanwhile, the people running the company are selling. Tens of millions in insider sales over 90 days. Multiple executives. No corresponding insider purchases. No cluster of buying to offset the distribution.
Divergence
Three layers, three different stories. Narrative: euphoric. Endorsement: unprecedented. Insider behavior: distribution. The system surfaces the divergence as the primary signal — not because any one data point is decisive, but because the layers aren't agreeing.
The system doesn't tell you what to do. It shows you when the story and the behavior are pointing in different directions. What you do with that is up to you. But you can't weigh that divergence if you only see the story.
Case 2 · The Pivot

The distressed-company sector rename.

A recurring pattern: a small-cap company far below its peak, in a deteriorating core business, announces a strategic pivot into whatever sector is currently attracting speculative capital — blockchain, cannabis, metaverse, quantum, AI. The stock spikes. The system recognizes the signature because it has resolved the same way dozens of times.

Distress Profile
Stock far below its historical peak. Core business shrinking. Retail footprint closing. Flagship assets being sold off. Revenue in multi-year decline. A company in the final act of a long fade.
The Pivot Announcement
An 8-K filing declares a strategic shift into the current speculative sector and a corporate rename to reflect it. No announced customers. No disclosed technology. No prior expertise in the space.
Financing Structure
A convertible note from an undisclosed institutional investor, often sized at multiples of the entire market cap. The convertible structure allows the investor to convert debt to discounted equity — a mechanic that historically has interacted with price spikes in specific, documented ways.
Insider Signal
Zero insider purchases in the 90 days before the announcement. In many cases, quiet insider sales. The people closest to the "pivot" aren't putting their own money behind it.
Historical Base Rate
Across the cohort of historical sector-rename pivots the system has catalogued: 94% returned to their pre-announcement price within 180 days. 41% faced SEC enforcement action within three years. The cohort spans nearly a decade and covers every major speculative sector cycle in that window.
The system doesn't characterize anyone. It recognizes the pattern and shows you the numbers. When ninety-four out of a hundred similar announcements resolved the same way, that's not a prediction — it's a statistic. The pattern was the signal. The history was the context. The decision is still yours.
Case 3 · The Silent Divergence

When coverage and insider behavior disagree.

One of the most consistent patterns in the historical dataset: sustained analyst bullishness running alongside sustained one-way insider selling. The system is built to make this kind of divergence visible, because the individual filings rarely are.

Pattern Component — Insider Layer
A company where insider transactions over a six-month window are heavily skewed to the sell side, with few or no corresponding purchases — even as the share price rises. Every Form 4 filing is public and disclosed within two business days, but the aggregate shape of the pattern only emerges when hundreds of filings are assembled.
Pattern Component — Founder and Long-Tenure Exits
Sales clustered around founders, long-tenured executives, or board members — particularly when those exits coincide with governance changes like board departures. The system tracks these signatures separately because historical data shows they behave differently from routine insider sales.
Pattern Component — Institutional Layer
Quarterly 13F filings showing large institutional holders reducing or exiting positions while the coverage narrative remains constructive. The filings are delayed by quarter, but the scale makes them meaningful when layered over the insider timeline.
Pattern Component — Coverage Layer
Sustained bullish analyst posture, minimal sell ratings, broadly positive financial-press coverage. The system measures this with the coverage-shape tools applied across wire feeds, IR releases, and social channels.
The Signal
Coverage and insider behavior pointing in different directions over a sustained window. The system surfaces the divergence and shows how similar historical patterns resolved — not as a forecast, but as context. The raw data for each component was public throughout. The missing piece was assembly.
The system doesn't predict the future. It assembles public filings into a picture that's otherwise spread across hundreds of documents over months, and shows how similar historical patterns have resolved. What the picture means, and what to do about it, is up to the reader.