HypurrTrade Proof of Work
Historical research, source checks and open gaps. These receipts are not live P&L, proof of profitability, a security audit or approval to trade.
- Strategy verdicts
- 3 NO PASS, 2 screen-only; none promoted
- Independent out-of-sample
- None — all evaluated periods reused or consumed
- Hosted / release readiness
- Blocked
- Verified public trade ledger
- Unavailable / not yet supported
Verify on Hyperliquid
One click re-downloads the official Hyperliquid builder-fill files and checks they still match our published fingerprints.
What this confirms: The daily files Hyperliquid publishes for this builder address still have exactly the bytes our receipt fingerprinted, and our published receipt and manifest are internally consistent.
What it does not prove: This does not prove strategy profitability, a complete public trade ledger, or that every HypurrTrade trade is represented. It is not a blockchain transaction or smart-contract proof.
Hyperliquid's official daily builder-fill files are exchange-published records of builder fees, which Hyperliquid processes onchain. HypurrTrade has published no separate hash-anchor transaction.
Separate on-chain hash anchor: not published
What this page is, in plain words
This is not crypto mining. It is a public notebook of what we tested, what failed and what is still unfinished. Builder activity comes from files Hyperliquid itself publishes for our builder address 0x0e024A4faD828e8f976fdCcbEEaAb72d4205be3c; each file carries a fingerprint (a SHA-256 code). The button above re-checks those fingerprints for you.
Bot orders also carry the public HTBT order prefix you can look up on Hyperliquid.
Advanced: check by hand
Download any official daily file linked in the builder-fill table below and compute its SHA-256 with any tool; it should equal the hash shown next to it. All receipts are indexed in /proof/manifest.json.
Matching fingerprints prove a file was not edited. They do not prove a strategy makes money. No separate on-chain hash anchor is published.
Status legend
- NO PASS
- The diagnostic failed its own evaluation gates. Not a candidate for live use.
- Screen only — no promotion
- An exploratory screen. No strategy was selected or promoted.
- Historical diagnostic — no trading approval
- A historical receipt with reused or consumed periods and no independent out-of-sample evidence.
- Current serving revision — not attested
- The reference SHA identifies the reviewed source snapshot, not the code serving this page.
- Hosted / Phase 6 readiness — blocked
- Required hosted verification is incomplete; these receipts authorize no release.
- Verified public trade ledger — unavailable
- No verified live trade, fee or user statistics are published.
Engineering / release checks
- Reviewed reference source SHA
- eb1ce1e7977e7911425769d1099946e37b1ec65a
- Current serving revision tests / build
- Not attested
- Hosted / release readiness
- Blocked
The reference SHA identifies the source snapshot reviewed for these receipts, not the current serving revision. No current-revision test/build attestation is published here. Local code checks are separate from strategy evidence and hosted release readiness.
Research receipts
Already checked-in reports only; no new market data or research run. Links below are curated public summaries, not raw internal reports. Each records the original report’s SHA-256 fingerprint. Positive historical figures are included for completeness, never as a claim that a strategy is profitable or Live-ready.
Four-coin candidate screen
No winner selected
Historical diagnostic — no trading approval
- Period (UTC)
- 2025-10-19 → 2026-10-18; 365 completed daily bars
- Universe
- BTC, ETH, SOL, HYPE
- Cost assumptions
- $10,000 initial capital; 11.5 bps per side (4.5 taker + 5 builder + 2 slippage); 1.0× gross baseline. Stress costs not run in this screen.
- Reported result
- Velocity Ranker −3.12% (max drawdown −6.31%, 43 trades); Carry-Aware Trend +46.20% (max drawdown −20.44%); Crowd Fade Carry −33.25%.
- Evidence limits
- Historical screen only. Comparing candidates consumes comparative information. The most recent Carry-Aware segment was −2.60%. Prior four-coin results were not fully replayed after the quantity-accounting correction; these are original reported figures, not current-engine replication.
- Source report fingerprint (SHA-256)
- 00ca221bffc0361225582bf5271d16cad180cd54bd77b69a766e9a8f2ae1b7a2
- Fingerprints the checked-in internal report (docs/research/candidate-screen-2026-10-20.md), not this public summary and not the current serving revision. Reviewed at reference SHA eb1ce1e7977e7911425769d1099946e37b1ec65a; current serving revision: Not attested.
Fixed broad eight-coin screen
Screen only — no promotion
Historical diagnostic — no trading approval
- Period (UTC)
- 2025-10-19 → 2026-10-18; 365 completed daily bars
- Universe
- BTC, ETH, SOL, XRP, DOGE, BNB, AVAX, LINK; no HYPE
- Cost assumptions
- $10,000 initial capital; 11.5 bps per side (4.5 taker + 5 builder + 2 slippage); 1.0× gross baseline. Stress costs not run in this screen.
- Reported result
- Velocity Ranker −2.17% (max drawdown −7.92%, 69 trades); Carry-Aware Trend −9.38%; Crowd Fade Carry +19.31%. Stress costs were not evaluated.
- Evidence limits
- Already-used dates, not a fresh holdout. Fixed-basket survivorship limitations remain despite trailing-volume eligibility. Later paired accounting comparisons used different funding fingerprints and conventions; they are not exact reproductions of these original figures.
- Source report fingerprint (SHA-256)
- 0ebd9b76424b99ed60e87a77b716a2061d0c30eeb6865f708398cc4f633da2fe
- Fingerprints the checked-in internal report (docs/research/broad-universe-screen-2026-10-20.md), not this public summary and not the current serving revision. Reviewed at reference SHA eb1ce1e7977e7911425769d1099946e37b1ec65a; current serving revision: Not attested.
Structural walk-forward diagnostic
NO PASS
Historical diagnostic — no trading approval
- Period (UTC)
- 2025-10-19 → 2026-10-18; final segment 2026-07-19 → 2026-10-18
- Universe
- BTC, ETH, SOL, HYPE
- Cost assumptions
- $10,000 initial capital; 11.5 bps per side (4.5 taker + 5 builder + 2 slippage); 1.0× gross baseline. 14/18 bps stress and 0.5× gross sensitivity.
- Reported result
- Selected funding-aware baseline: train +15.79%, validation +30.19%, evaluated final −2.61% (max drawdown −8.32%, 38 trades). All six gross/cost scenarios were negative. Five of seven gates failed.
- Evidence limits
- Chronological 50/25/25 selection used train/validation only. The final quarter has now been evaluated and is consumed, not available as fresh evidence. Prior HYPE/walk-forward figures were not fully replayed after the accounting correction.
- Source report fingerprint (SHA-256)
- 65a1c951bc64a0168dbe46b6a86ee1c5b62a82250fd3fb9623533e5e5ffccc79
- Fingerprints the checked-in internal report (docs/research/walkforward-2026-10-20.md), not this public summary and not the current serving revision. Reviewed at reference SHA eb1ce1e7977e7911425769d1099946e37b1ec65a; current serving revision: Not attested.
Longer-history structural comparison
NO PASS
Historical diagnostic — no trading approval
- Period (UTC)
- 2024-10-20 → 2026-10-20 UTC end-exclusive; 730 daily bars
- Universe
- BTC, ETH, SOL, XRP, DOGE, BNB, AVAX, LINK
- Cost assumptions
- $10,000 initial capital; 11.5 bps per side (4.5 taker + 5 builder + 2 slippage); 1.0× gross baseline. 14/18 bps stress and 0.5× gross sensitivity.
- Reported result
- A failed training (−10.59%); B and C failed validation (−22.19% and −23.23%). No candidate selected. Reserved final 2026-04-20 → 2026-10-20 and final stress scenarios/gates were NOT EVALUATED.
- Evidence limits
- Corrected signed-quantity/cash accounting; long-only keeps cash. Reserved final overlaps previously studied dates and would be reused evidence. Same-close fills are optimistic. Pre-order gross drift reached 1.2159×; the paid close-observed cap is not a continuous intraday guarantee.
- Source report fingerprint (SHA-256)
- 9d81e2929b4ec768d2735a7a90152adb68024d711b5571b5bd8000a43686fddb
- Fingerprints the checked-in internal report (docs/research/round2-long-history-2026-10-20.md), not this public summary and not the current serving revision. Reviewed at reference SHA eb1ce1e7977e7911425769d1099946e37b1ec65a; current serving revision: Not attested.
Long-only Carry-Aware structural diagnostic
NO PASS
Historical diagnostic — no trading approval
- Period (UTC)
- 2024-12-06 → 2026-10-20 UTC end-exclusive; 683 common daily bars
- Universe
- BTC, ETH, SOL, HYPE
- Cost assumptions
- $10,000 initial capital; 11.5 bps per side (4.5 taker + 5 builder + 2 slippage); 1.0× gross baseline. 14/18 bps stress and 0.5× gross sensitivity.
- Reported result
- Train +4.94%, validation +1.90%. Contaminated final 2026-05-01 → 2026-10-20: +25.74%, max drawdown −13.48%, 23 completed closes. HYPE contributed 92% of net P&L. Coin/quarter concentration failed; close-count gate inconclusive.
- Evidence limits
- No independent OOS: final overlaps previously evaluated 2026-07-19 → 2026-10-18. Train/validation reuse Round 2 dates on a different universe. All six cost/gross scenarios were positive (+11.91% to +25.74%) but confer no approval. Same-close fills remain optimistic. HYPE listing history limited the common sample to 683 days.
- Source report fingerprint (SHA-256)
- 3cb44f8bf75458519564ab20ff9e1b7b32b9f6a450f239c663b37e2bb46ffaf1
- Fingerprints the checked-in internal report (docs/research/longonly-walkforward-2026-10-25.md), not this public summary and not the current serving revision. Reviewed at reference SHA eb1ce1e7977e7911425769d1099946e37b1ec65a; current serving revision: Not attested.
Builder-fill activity receipt
Date-level aggregates computed from Hyperliquid public builder-fills daily files (official Builder Codes data), verified by hash and schema before aggregation. Per-wallet rows, addresses and per-user P&L are discarded at parse time.
- Window (UTC)
- 2026-10-22 → 2026-10-24 (3 verified day(s))
- Builder-tagged fill records
- 4
- Aggregate notional proxy
- $207.9
- Builder fees (source units)
- 0.207902
- Retrieved at
- 2026-10-25T09:30:23.568Z
- Counts are builder-tagged fill records (rows); each row may be a partial fill, so counts are not unique orders.
- The aggregate notional proxy is sum(abs(px*sz)) across fill rows, not executed order value or volume net of cancels.
- Builder-tagged fills alone do not prove net profitability, a successful user experience, unique users, or all HypurrTrade activity.
- Days whose file is missing are labelled missing, never counted as verified zero activity.
| Date (UTC) | Status | Fill records | Notional proxy | Builder fees (source units) | Raw file SHA-256 |
|---|---|---|---|---|---|
| 2026-10-24 | Verified | 1 | $13.69482 | 0.013694 | 376a2cb5204b5c0aa49c32f2d26c28a5094dcadaf0ff138934beca17672ab97eOfficial daily file (2026-10-24) |
| 2026-10-23 | Verified | 2 | $57.07552 | 0.057074 | 1bb02e87d98aa81adea5f90d37d54429a918d9a4f8aebe5db3669713a60aeee4Official daily file (2026-10-23) |
| 2026-10-22 | Verified | 1 | $137.13462 | 0.137134 | 9eaa196d00e1ce9897cdb2830e42f6c1895e40654d4e09fd8fe170b67884257eOfficial daily file (2026-10-22) |
Reproducibility
- Source and method
- Hyperliquid public builder-fills daily files (official Builder Codes data); receipt method builder-activity-receipt@1.
- What a hash proves
- A raw-file SHA-256 proves the retrieved daily file's bytes matched the receipt at retrieval time, and that aggregates were computed from exactly those bytes after schema and number validation.
- What a hash does not prove
- A hash does not prove the file is complete or correct at source, does not prove profitability, unique users or all HypurrTrade activity, and says nothing about the current serving revision or hosted readiness.
- Regenerate the builder receipt
- From a checkout of this source tree, run:
- bun scripts/proof/builder-activity.ts --days=3 --out=public/proof/builder-activity.json
- The script re-fetches the official daily files, validates LZ4 structure, schema, numbers and dates, and rewrites the aggregate receipt. It never writes raw fill data.
- Machine-readable manifest
- Every receipt on this page is indexed in /proof/manifest.json (proof-of-work-manifest@1) with artifact paths, source fingerprints, verdicts and caveats.
Security / custody boundaries
Source/documentation statements at the reference snapshot; not a hosted security attestation.
- Source design: trading runs in an open browser tab, with browser-held agent-key signing rather than server-side order signing. This registry loads no trading engine or wallet controls.
- Hyperliquid trading-agent permissions are intended for trading, not withdrawals. This is a custody-boundary statement, not a security audit or a guarantee against browser compromise or trading losses.
- Live configuration changes require the existing owner-wallet signature and matching verified configuration. This milestone has not exercised a wallet or verified a hosted signing flow.
- Phase 5 validation/shadow remains contained and unavailable. Referral payout settlement remains disabled in source.
- Browser trade submissions are refused in the current source. Legacy public records are unverified and cannot establish live P&L, fees, users or a verified ledger.
Current evidence gaps
Verified public trade ledger: Unavailable / not yet supported
- Phase 6 hosted/release readiness is blocked: required integrity RPC preparation is unapplied; isolated hosted verification, staged-job concurrency/recovery and runtime/IP behavior remain unverified. No production SQL or release is authorized by these receipts.
- Phase 5 validation/shadow remains contained. Historical gates, local SQL simulations and passing code tests are not independent prospective evidence.
- Verified public trade ledger: unavailable / not yet supported. No verified live trade, fee or user statistics are published here.
- The latest long-only result is NO PASS with a contaminated/reused final segment and no independent OOS. Prior periods are consumed; original screens have accounting/input caveats and do not prove executable live parity.
- Velocity Ranker policy mismatch remains unresolved: project guidance restricts daily-ranked deployment and lists only tested Live options, while source includes daily Live paths and the selector includes Velocity Ranker. Selector availability is not Live validation or permission to activate it.
- No independent security audit, hosted release certification, real-wallet signing verification or independently verified live performance is asserted by this registry.