Xommodo: a track record that proves itself
In the first Notes post we showed how a SHA-256 hash chain makes audit logs tamper-evident — and which building block an external anchor adds. Xommodo, our forecast research demonstration, shows both in live operation: forecasts timestamped on the Bitcoin blockchain, verifiable by anyone.
The section “Update August 2026” below states +127.3 % as the total return adjusted for roll costs. This figure is wrong. Roll costs accounted for roughly 8 of roughly 67 percentage points of deviation; the reliable value for the same date is +67.9 %. In the forward window, the simulation also trailed buy-and-hold by 10.3 percentage points. The original statement remains below so that what was published stays readable. How this happened and what follows from it →
What Xommodo is
Xommodo produces quantile forecasts (q10/q50/q90, 10-day horizon) for commodity prices — currently copper, as a public, free preview. The forecasts draw on weak, leading upstream signals; a paper trader turns them into simulated positions. Xommodo is a research demonstration: we show method and track record — not investment advice, not a trading recommendation.
The core: provability before performance
Every forecast is committed before the outcome is known: as an entry in an append-only ledger with a SHA-256 hash chain — the same construction principle as in the first post. Added to this is exactly the building block we described there as a later stage: the external anchor. Xommodo anchors the ledger on the Bitcoin blockchain via OpenTimestamps — this secures the end of the chain as well, and the track record is preserved in full, exactly as it came into being. Raw forecasts and metrics sit alongside, dated; the current head hash is published on the site. Anyone who wants to can check the arithmetic.
Status today (July 2026)
- Public copper preview live on xommodo.com — dashboard and four-panel forecast chart freely accessible
- 20 anchored predictions in the public ledger
- Paper-trader simulation over 1,273 days (to 1 July 2026): $10,000 became $22,687 (+126.9 %); buy-and-hold reached +44.1 % over the same period — a simulated lead of 82.8 percentage points
Simulated and historical results are no indicator of future results.
Update August 2026 — research work completed
Xommodo has examined its research question — can weak, leading signals from public data predict commodity prices verifiably? — and completed the work as planned. The project has been discontinued, and this is exactly where the approach shows its value: a cryptographically anchored track record remains fully verifiable, even after the last forecast has been written.
- Final ledger state: 47 anchored entries, last forecast dated 12 August 2026; head hash and OpenTimestamps anchor remain public.
- Among them a documented gap entry (27 July – 6 August 2026, outage of the capture infrastructure). The gap stands as its own entry in the same hash chain — neither removable nor shortenable after the fact. A reliable track record documents its outages as provably as its hits.
- In August the paper-trader simulation was refined with direction-dependent roll costs from the term structure — the reported total return fell from +135.0 % to +127.3 % as a result. Accuracy takes precedence over the nicer figure. [Corrected on 26 August 2026: +127.3 % was not the correct value but a narrowly corrected wrong one — the correct figure is +67.9 %. See the box above.]
- The interactive views (paper-trader dashboard, forecast chart, comments) have been discontinued; ledger, raw forecasts and metrics remain as a public research archive.
- The results will appear as a paper — including the null results and the method (pre-registered experiments, synthetic controls, tamper-evident publication). Null results are rarely published; that is exactly why we do.
Simulated and historical results are no indicator of future results.
What is public — and what remains working capital
Public is everything that builds trust and requires verifiability: track record, ledger, anchor, metrics. The signal sources, the model architecture and the data pipeline are our working capital and remain so. This separation follows the same thinking as the audit format: transparency where it proves — discretion where it protects value.