InsiderEdge

Institutional-grade insider-buy signals for your AI agent — from free SEC data, scored by what actually predicts returns.

Add it to any MCP-capable agent (Claude Desktop, Cursor, Cline, …) by pointing it at https://insideredge-mcp.pfitztech.workers.dev/mcp:

{
  "mcpServers": {
    "insideredge": {
      "url": "https://insideredge-mcp.pfitztech.workers.dev/mcp"
    }
  }
}

Upgrade to Pro

What it does

Most insider trackers just dump raw Form 4 filings — and the research says that's noise. The alpha is in opportunistic open-market buys, especially clustered ones in small caps. InsiderEdge classifies and scores those filings with a research-backed conviction model, detects insider cluster-buys, and overlays 13F fund-holding changes as confirmation. It exposes the result as tools any agent can call — ask "show me the strongest insider cluster buys this month" and get a ranked, explainable list.

Tools

Free

health FREE
Server status, your current plan, and the disclaimer.

resolve_ticker FREE
Resolve a ticker to its SEC CIK and company name.

get_insider_activity FREE
Recent Form 4 transactions for a ticker, each open-market buy conviction-scored.

get_insider_cluster_buys FREE
Ranked insider cluster buys by cluster strength (top-line list).

Pro

screen_insider_conviction PRO
Multi-factor conviction screen — filter by role, conviction, value, opportunistic-only.

get_13f_changes PRO
A fund's quarter-over-quarter 13F deltas (NEW / ADDED / TRIMMED / EXITED).

get_fund_consensus PRO
Which funds newly opened or added a ticker — smart-money confirmation.

create_alert PRO
Webhook/email alert that fires on new cluster buys matching your criteria.

list_alerts PRO
List your registered alert rules.

delete_alert PRO
Delete one of your alert rules by id.

Pricing

Tier	Price	What you get
Free	$0	20 reads/day, top-line cluster list, 7-day history. No alerts or 13F overlay.
Pro	$29/mo	Unlimited reads, full conviction detail, screens, 13F confirmation overlay, full history, and alerts/webhooks.
Founder	$19/mo	Everything in Pro — first 50 users, locked for life.

Upgrade to Pro

How the conviction score works

Anyone can echo raw Form 4 data; the edge is scoring which insider buys actually predict returns. The weights below are grounded in peer-reviewed finance research, not folklore.

Stage 0 — Gate. Keep only open-market purchases (transaction code P). Option exercises (M), grants, gifts (G) and sells are dropped before scoring; sells are tracked separately as a cluster-sell warning, never as a bullish input.

Stage 1 — Per-trade conviction (0–100), a weighted sum of six factors:

Factor	Weight	What it captures
Opportunistic vs routine	0.35	Routine (same calendar month yearly) ≈ no alpha; opportunistic carries it. The single biggest lever.
Open-market buy quality	0.15	Gated to code `P` — a clean, single-motive buy signal.
Insider role	0.15	CEO/CFO weighted highest, but "local" officers and division heads are kept — often the most informed.
Net-purchase-ratio	0.15	Shares bought ÷ shares already held — not raw dollar size, which is a weak standalone signal.
Recency	0.10	Decay over ~90 days; the signal persists roughly 1–6 months.
Small-cap tilt	0.10	Predictability concentrates in micro/small caps; large caps are discounted.

Stage 2 — Cluster strength. Flag a ticker when ≥3 distinct insiders buy within a rolling window (default 30 days). Strength = distinct_insiders × avg_conviction × log(net_value), with a bonus when an independent director is in the cluster. Breadth of distinct insiders matters more than one big trade.

Stage 3 — 13F overlay (confirmation, never the trigger). Count how many low-turnover funds newly added the ticker last quarter and surface it as a smart-money badge, with a small capped multiplier. 13F filings lag up to 45 days, so this is context only.

The research behind the weights

Lakonishok & Lee — Are Insider Trades Informative? Buys predict, sells don't; the effect concentrates in small firms.
Cohen, Malloy & Pomorski — Decoding Inside Information (NBER w16454). Opportunistic trades carry the alpha; routine trades carry ~none.
Lakonishok, Jeng & Metrick (Wharton / Rodney White). Open-market purchases as the clean signal; use net-purchase-ratio over raw dollars.
Cluster Trading of Corporate Insiders. Cluster buys roughly double the signal versus single-insider buys.
Hvide & Nielsen — Flying Below the Radar (JFE 2026). Below-the-top "local" insiders are often the most informed.
The Evidence Investor — what insider trades (and non-trades) tell us. Signal persistence over the 1–6 month horizon.
Systematic 13F Hedge Fund Alpha (SSRN). Conviction + consensus beats the market only for low-turnover managers.

Signals, not investment advice. Nothing here is a recommendation to buy or sell any security. Insider-alpha research uses older samples and SEC enforcement has thinned opportunistic trading; small-cap liquidity is thin and slippage is real; 13F data lags filings by up to 45 days. Re-validate before sizing positions and do your own due diligence.

Built on free SEC EDGAR data (Form 4 + 13F).