A timestamped public record of every signal and trade produced by IBIS, a private quantitative trading system. Entries are pushed here automatically when they occur — before their outcomes are known — so the record can be watched forming in real time rather than assembled after the fact.
Nothing in this repository is investment, financial, trading, or legal advice, or an offer or solicitation of any kind. It is a research record of an experimental system trading a simulated (paper) account. Do not make investment decisions based on it. If you do anything with these records, you do so entirely at your own risk.
- signals.md — every signal-day the system has computed: direction, confidence, price at first detection.
- account.md — the paper account: open positions with their exit rules stated up front (max hold, target, stop), and every closed trade with its outcome.
- EVIDENCE.md — the system's auto-generated self-grading report: independent-episode counts, resolution status, baseline comparisons, statistical power thresholds, and uptime. Its verdict is computed from the data, not written by a person.
- notes/cross-domain-test.md — a documented negative result from testing the prediction core on non-financial data, including the false positive the grading system caught.
- Each entry was published at the git timestamp shown in the commit history, which is recorded by GitHub, not by us.
- Positions are opened with their exit rules declared in advance and are closed only by those rules (stop, breakeven lock, or time expiry) — never by discretion.
- The record includes everything: losing trades and expired signals stay.
- No real money. This is a paper account with simulated fills at observed prices. Slippage and liquidity in live execution would differ.
- No proven edge. The system's backtest results are in-sample by definition; this ledger exists because those numbers shouldn't be trusted until an out-of-sample record accumulates. That takes months. As of the first publication there are zero closed trades.
- Not literally immutable. Git history can in principle be rewritten by the repo owner (force-pushes are disabled, but ownership is ownership). What the ledger offers is third-party timestamps and tamper-evidence: anyone watching the repo would see a rewrite.
- No methodology. How signals are generated stays private. This record is only about whether the calls, made in advance, are any good.
The repo went live on 2026-07-06. The signal rows and the open position dated before that were backfilled from the system's local log at first publication — their detection timestamps are as recorded locally, but their publication timestamp is the first commit. Everything after 2026-07-06 is published as it happens. The one open position's outcome (BTC short, max hold through 2026-07-23) was unknown at publication time.