Participate
Prepare a v1.0-compatible record
No official submissions or empirical results exist for v1.0. You can format a v1.0-compatible synthetic or private-evaluation record against the schemas, run the reference checks in the harness, and contact us for a pilot or private evaluation. Any record shown is labeled by status (synthetic / private / public) and is not an official result.
How submission works
- Run your system under walk-forward sealing on the chosen track/split.
- Produce a sealed results file (timestamped + hashed) and a metadata record.
- Submit the form below. Your entry lands as pending review.
- The organizer triages it to public, verified, or private-eval (or requests changes).
Submission tiers
- Public (unofficial) — self-reported on the open/synthetic split. Clearly labeled.
- Verified — the organizer reproduces your result from submitted code.
- Private eval — the organizer runs your system on a frozen hidden split end to end.
⛔
No future leakage. Only evidence timestamped at or before the prediction time t may inform a prediction. Random cross-validation, peeking at outcomes, or training on the resolution window invalidates a submission.
Required files & metadata
- Results file (sealed predictions), or a runnable system for private eval.
- Model/system name, output adapter, training cutoff.
- Track, split, and benchmark version.
- Both attestations below.
Need a private evaluation or have questions? Email yurisyl@gmail.com.