About | Tebada Insider

The data

The SEC requires officers, directors, and 10%+ beneficial owners to disclose every transaction in their company's securities via Form 4 within two business days. EDGAR publishes the raw XML publicly. We ingest those filings continuously and parse each transaction.

The score

For every (ticker, day) we compute a conviction score in [0, 10] that mixes four bounded components:

Sv — log-notional intensity, weighted by recency, optionally normalised by liquidity.
Sf — frequency of buy events with exponential decay (configurable half-life).
Su — number of distinct insiders, saturating monotonically.
Sa — historical alpha of buys for that ticker: the median return after the insider purchase window, shrunk by event count.

Weights are normalised internally so the resulting score is always within [0, 10] regardless of the configured parameters. The full formula is frozen as a contract: changes require 172 snapshot tests and are versioned to preserve reproducibility of past backtests.

The backtests

We run scenarios as deterministic backtests over the historical transaction stream against actual price series. Each run records entries, exits, holding days, P&L absolute and relative, drawdowns, and alpha/beta against SPY. Over 1,000 such runs feed our strategy selection.

The live engine

A daemon polls EDGAR every few minutes during US market hours, recomputes signals, and publishes new buys/sells to a real-time channel. Pro+ subscribers receive them via the dashboard (Server-Sent Events) and Telegram. Optional IBKR integration mirrors orders to a paper trading account; live trading remains restricted pending compliance review.

What we don't do

We do not give investment advice. Signals are general market information, identical for every Pro+ subscriber.
We do not manage your portfolio. The broker integration is opt-in, audit-logged, paper-only by default.
We do not predict the future. Past performance is not indicative of future results.

Open foundations