Robinhood Chain · 4663
Every trade committed before it ends.
Quantt runs automated strategies on live Robinhood Chain data and writes every entry and exit to an on-chain contract within a minute. Losing trades can’t be hidden. Results can’t be edited. Anyone can check any trade.
| Strategy | Batch | Window (UTC) | Decisions | Merkle root |
|---|---|---|---|---|
| #0 | 41 | 14:02:11 – 14:03:05 | 2 | 0x9c1e…42af |
| #1 | 118 | 14:01:40 – 14:02:38 | 3 | 0x5d07…e1b3 |
| #1 | 117 | 13:59:52 – 14:00:49 | 1 | 0xa3f4…07cc |
| #0 | 40 | 13:57:20 – 13:58:16 | 1 | 0x16bb…9d20 |
Problem
Trading results are screenshots. Screenshots can’t be audited.
Signal channels, bots and AI agents publish profits nobody can verify. Losing strategies disappear, losing trades are left out, costs are ignored, backtests are tuned until they look good. On a chain with 100 ms blocks, bonding curves and tokenized stocks, the gap between a backtest and real fills is especially wide.
How it works
Register, trade, commit, verify
1 · register
Strategy goes on chain first
Registered before its first trade. Registrations can’t be deleted.
2 · trade
Live data, real costs
Paper trades around the clock with curve depth, fees and gas measured on chain.
3 · commit
Before the outcome
Each entry is committed within a minute, before its 120-second exit.
4 · verify
Proof for every trade
Batches are published with Merkle proofs; the contract confirms any trade.
Proof
The contract, in plain sight
No owner, no admin, no upgrades, no funds held. Source verified on the explorer and on Sourcify.
- Contract
- 0x7EE4325B8Beec66be1ccBd5fFeB4b5BDe6E4692D
- Strategy #0
- k2-early-120s · passed the pre-registered selection protocol
- Strategy #1
- k2-all-120s · the same rule without its filter (control)
- Source
- Sourcify · exact match
// take hash and proof from any batch json in the live record verifyDecision(0, index, hash, proof) // → true // change one field of the decision, recompute the hash verifyDecision(0, index, hash', proof) // → false
Results
We publish what failed
All numbers after realistic costs. Selection rules were written down before the out-of-sample test.
| Test | Scale | Outcome | Verdict |
|---|---|---|---|
| Public Telegram signals | 1,086 signals | No configuration profitable after costs | rejected |
| Tracked wallets, all tokens | 3,695 variants | Out-of-sample NO-GO; mean 0.978 | rejected |
| 2nd buyer, early tokens, 120 s | 913 trades | Mean 1.051, lower bound 1.017, median 0.93 | candidate |
| Same candidate, forward, public | from 26 Sep | Every trade committed on chain | running |
For
Builders, agents, and the people who follow them
Strategy builders
Prove results while the rules stay private behind a hash.
AI trading agents
Turn an agent’s claims into an auditable history.
Followers and allocators
See every trade, every loss and every abandoned strategy before paying.
Roadmap
From one engine to an open registry
now
Public paper record
Two strategies committing every trade.
next
Open registry
External strategies and agents, same rules.
then
Tokenized stocks
Stock-token strategies priced by Chainlink.
later
Certificates
On-chain snapshots and a capped live test.
FAQ
What exactly does the chain guarantee?
That each decision existed when it was committed and hasn’t changed since. Entries are committed before their 120-second exit, so losing trades can’t be dropped. Gaps in position numbers are visible.
What doesn’t it guarantee?
Paper fills are our model. The chain proves what the engine claimed at that moment; each decision names token, time and price, so anyone can re-simulate the fill at that block.
Could you hide failed strategies?
No. Strategies are registered before they trade and registrations can’t be deleted.
Why Robinhood Chain?
Retail users, tokenized stocks and agentic trading are arriving there; 100 ms blocks and tiny gas make per-minute commits practical.
Is this investment advice?
No. Paper results are not real-money results.