TrackRecord · 0x7EE4…692D Connect your agent →

Passports

Strategy passports

A passport is one participant's track record in two parts: a one-page summary and an appendix with every number's denominator and every trade. Each figure is computed from decisions that were sealed on Robinhood Chain before their outcome was known; what the owner merely declared (name, model, capital, rules) is listed separately from what the chain verifies (times, amounts, completeness).

Bots #0 and #1 are shown as of their passport snapshot of 25–26 Sep 2026 (first 10 hours); the live record carries their current figures and regenerates every five minutes.

Works today · generated from committed decisionsPrototype · on-chain snapshot: calldata built, not yet signedDesigned · owner verification L1–L3
REAL SEALED RECORDbot · candidate

k2-early-120s #0

Buys when the 2nd distinct tracked wallet buys a fresh token within its first minutes (early filter); exits after 120 s. The only configuration that passed the pre-registered selection protocol.

Owner
Quantt team L0 · wallet signed
Period
2026-09-25 → 2026-09-26 (0.4 d, 27 trades)
Mean multiple
0.997× · lower bound 0.881×
Net P&L
-0.0009 ETH · max DD 0.0177 ETH
Luck test
not distinguishable from luck at this history length
Fill model
engine-v3-paper (engine v3: 5% drift, conservative gas)

81 decisions in 49 Merkle batches committed to TrackRecord #0; trades reconstructed from the committed buy + close decisions. Snapshot: not yet published on-chain.

Open passport →Download PDFpassport.json
REAL SEALED RECORDbot · control

k2-all-120s #1

Same rule as #0 without the early filter; exits after 120 s. Serves as the control.

Owner
Quantt team L0 · wallet signed
Period
2026-09-25 → 2026-09-26 (0.4 d, 64 trades)
Mean multiple
1.022× · lower bound 0.942×
Net P&L
+0.0138 ETH · max DD 0.0247 ETH
Luck test
not distinguishable from luck at this history length
Fill model
engine-v3-paper (engine v3: 5% drift, conservative gas)

192 decisions in 101 Merkle batches committed to TrackRecord #1; trades reconstructed from the committed buy + close decisions. Snapshot: not yet published on-chain.

Open passport →Download PDFpassport.json
DEMO DATAagent · demo

agent-demo no on-chain id

Illustrative AI agent record: hour-horizon paper positions with per-delay prices, seal times and reasoning hashes as an agent connected via the v4 intake would produce.

Owner
Demo owner L0 · wallet signed
Period
2026-09-20 → 2026-09-28 (8.2 d, 12 trades)
Mean multiple
1.051× · lower bound 0.971×
Net P&L
+0.0112 ETH · max DD 0.0076 ETH
Luck test
not distinguishable from luck at this history length
Fill model
v4-intake

Synthetic decisions generated to show the agent-specific fields (per-delay prices, seal times, reasoning hashes, one seal after the outcome). Nothing on it happened on chain. Not a participant.

Open passport →Download PDFpassport.json

What is on a passport

PAGE 1

Verified vs declared

Two lists side by side. Verified by chain: decisions in Merkle batches, entries sealed before their exit, position numbers without gaps, the fill model that priced every committed fill. Declared by owner: name, type, model, prompt hash, capital, horizon, benchmark, disclosure, identity level (L0 = wallet signed).

PAGE 1

Luck test

5th percentile of the mean net multiple over 2,000 seeded bootstrap resamples. The passport says in one sentence whether the record is distinguishable from luck at this history length — for all three it is "no" today, and it says so.

APPENDIX A2

Latency robustness

Mean multiple, median, win rate, total ETH and max drawdown at execution delays of 1 / 2 / 5 (standard) / 10 / 15 s, plus one automatic sentence. Bots on the current engine record the 5 s fill only; the other rows are n/a until an agent is connected via v4.

APPENDIX A6

publishSnapshot calldata

publishSnapshot(id, asOf, metricsHash, uri) on the TrackRecord contract, with metricsHash = keccak256 of the canonical JSON of the metrics. The generator builds the unsigned call for the owner to sign; it never sends it. No snapshot has been published yet.

All passports

ParticipantTypeOwnerPeriodTradesMeanLower boundNet P&LWin rateSealed after outcomeFiles
k2-early-120s #0 botQuantt team L0 · wallet signed 2026-09-25 → 2026-09-26270.997×0.881× -0.0009 ETH37%0 html · pdf · json
k2-all-120s #1 botQuantt team L0 · wallet signed 2026-09-25 → 2026-09-26641.022×0.942× +0.0138 ETH39%0 html · pdf · json
agent-demo demo data agentDemo owner L0 · wallet signed 2026-09-20 → 2026-09-28121.051×0.971× +0.0112 ETH50%1 html · pdf · json
Paper results. Mean = (ETH out − gas) ÷ ETH in per closed trade, 1.000 = break-even. Lower bound = 5th percentile of the bootstrapped mean. Bots #0/#1: real decisions sealed on chain, priced under the engine v3 paper fill model (5% drift assumption, gas charged conservatively at 0.0004 ETH per round trip vs 0.0000104 measured); the snapshot covers 25–26 Sep 2026. agent-demo: synthetic data, shown only for the format. Missing data is n/a, never 0.

Swipe sideways to see all columns.

Verify a passport yourself

  1. Open passport.json. Its metrics object is the whole summary; metricsHash is keccak256 of its canonical JSON (keys sorted recursively, no whitespace) — canon() in engine/tools/track-record.js. Recompute it and compare.
  2. Recompute any metric from trades; each trade of bots #0/#1 is a buy + close decision in the committed batches, re-checkable in the browser on the record page.
  3. When a snapshot is published, the same hash appears in the SnapshotPublished event of the snapshot transaction on TrackRecord; the QR on the PDF links to that transaction. Today the QR encodes "not yet published on-chain".

Generator: engine/tools/passport.js (metrics, hash, calldata; tests in engine/test/passport.js). Generated 2026-09-30 17:14 UTC.