Proof of Process
Dated public notes and SHA-256 fingerprints to verify. A fingerprint identifies a file; it does not certify its interpretation or engine profitability.
Deployed, observed and archived results
One shared engine Checked on 14 September 2026 at 23:15 UTC: vault entries and execution disabled, no position or open order. Activation remains the owner's decision. This is a dated observation, not a permanent guarantee.
Dated archives · previous models These historical results belong to their original versions and datasets. They do not measure the deployed release and do not guarantee a future edge.
Research and engineering activity
GitHub delivery pendingDocumented release
F378
Paper observation
Recorded commits
861
Since 2026-09-07
Latest activity
Engineering activity and technical checks do not establish profitability.
Research records
- Replay bench2026-09-22 · F400
Checks incomplete; no promotion
- Replay bench2026-09-22 · F400
Recorded technical checks passed
- Replay bench2026-09-22 · F400
Observation in progress; no financial conclusion
- Replay bench2026-09-22 · F400
Observation in progress; no financial conclusion
- Entry-policy research2026-09-22 · F400
Recorded technical checks passed
- Public evidence publication2026-09-22 · F353
Recorded technical checks passed
- Entry-policy research2026-09-22 · F400
Evidence inconclusive
- Entry-policy research2026-09-22 · F400
Evidence inconclusive
Historical release documents
One shared engine
F331 candidate · F332 integration
Dry-run and vault decisions use the same verified source and configuration. Real execution prices and fees may differ. The vault remains disabled.
Qualification in progress
Prospective trial · profitability unproven
The candidate was developed using three already studied trades. The higher-cost stress scenario remains negative. Software tests and deployment do not establish a statistical edge.
Research cycle
Collect → replay → compare → observe
Recorded decisions connect collection, test bench and workshop. Every change must preserve its data, version, costs and validation criteria. Documented failures remain in the archives.
Dated archives · previous models
Dated archives · previous models September 3, 2026
These historical results belong to their original versions and datasets. They do not measure the deployed release and do not guarantee a future edge.
Dated archives · previous models
15m Historical Positions
6 540
Historical dataset 2021–2026 (Binance)
Simulated liquidations
0
Previous simulation only
Bias identified
Bias identified
Selected at close, not predictive
Simulated liquidation rate
0.00%
Historical observation, no guarantee
HyperNatt dry run
Simulated trade history. Each reported trade gain follows the journal convention and may exclude entry fees. Realized account PnL is shown separately. This table is not evidence of real execution.
x402scan counters unavailable.
On x402scan, select “All Time” to compare.
Dated archives · previous models
6,540 positionsTable A — Candle Close Rejection Signal Audit (2021–2024)
Statistical Confirmation at 15m Close (TRAIN n=4,782)
Lab TRAIN · Bias identified. These historical results belong to their original versions and datasets. They do not measure the deployed release and do not guarantee a future edge.
Dated archives · previous models · 664
Dated archives · previous models · 1758 / -7.7 bp
These historical results belong to their original versions and datasets. They do not measure the deployed release and do not guarantee a future edge.
Dated archives · previous models · 664
Dated archives · previous models · 6540 / 2021-2026 / Binance Spot
These historical results belong to their original versions and datasets. They do not measure the deployed release and do not guarantee a future edge.
Dated archives · previous models · 656
Dated archives · previous models · 645 / n_fills=0 / 2026-09-03
These historical results belong to their original versions and datasets. They do not measure the deployed release and do not guarantee a future edge.
Dated archives · previous models · 670
Dated archives · previous models · 2026-09-06 / FLAT
These historical results belong to their original versions and datasets. They do not measure the deployed release and do not guarantee a future edge.