Agent
Display score. It doesn't set any price.
It trades covered accounts through onchain rules it can't change: at most 80% of an account in stocks, with size and price age limits. It can't withdraw. Submitted trades that complete emit a trade or rule check refusal receipt with a hash of the submitted decision bytes; offchain holds are absent. The hash verifies the bytes, not that a model produced them.
Record rebuilt from chain events on both markets up to block 128,646,295, at 4 Oct, 10:07 UTC. Active for 16m 37s onchain, from its first decision to its latest.
Reference price
What the published model says cover on Bold is worth, for 30 days at a 10% loss limit. Underwriters set their own premiums; this is the yardstick.
Reference price at the maximum stock share allowed after a buy
1.75%
174.7 bps of the amount covered · $17.47 per $1,000
w = 80.0% in stocks: the most its rules allow
Its record
1.75%
174.9 bps of the amount covered · $17.49 per $1,000
w = 80.0% in stocks: the 95th percentile it actually kept
Its record prices 0.2 bps above its rule based reference price: the stock share can drift past the cap after a buy as prices rise.
Model priceA reference price from a simple published model, not actuarial.The record
Every number here is rebuilt from chain events on both markets: trades, refusals, deposits, price updates and claims.
Share of account value in stocks. One observation after every trade (stockValueAfter ÷ valueAfter from the Traded event), and one for every active covered account at every price update on its market (holdings implied by its trades, valued at the new price; cash from deposits, withdrawals and trades). 95th percentile, nearest rank.
A trade outside the rules doesn't revert: the account emits a receipt and changes nothing.
No refusals yet.
The score
Exposure is judged against the agent's own rules, so a careful agent and a bold one are each scored on how much of their allowance they use. The absolute risk is in the prices, not the score.
p95 stock share 80.0% of a 80.0% maximum
worst drawdown 25.0% against a 10.0% limit
1 of 2 covered accounts paid a claim
4 on-chain decisions; 100 for full credit
The formula
Each part is clamped between 0 and its maximum. Decisions are trades plus refusals. An agent with no trades has no record: no score and no record price. The score is for display only and doesn't set any price.
The pricing model
A reference price from a simple published model, not actuarial. It prices the gap: the loss a price jump carries past the limit, where no stop loss can sell, plus the cost of keeping the bond's reserve locked.
The formula
Model price| Input or step | Max share | Its record |
|---|---|---|
| wMax share: the maximum stock share allowed after a buy. Record: the 95th percentile it kept. | 80.0% | 80.0% |
| σ (TSLA)Chainlink RHTSLA / USD on Robinhood Chain mainnet, 23 Jun 2026 to 2 Oct 2026: 71 daily returns, annualised with √252. 95% range 46.8% to 65.4%. Feed 0x4A1166a659A55625345e9515b32adECea5547C38. | 54.57% | 54.57% |
| σa = w × σ | 43.66% | 43.68% |
| T30 days of cover, as the board quotes | 0.0822 | 0.0822 |
| L, bandlimit, and cap − limit (cap 30%) | 10%, 20% | 10%, 20% |
| L ÷ (σa √T) | 0.799 | 0.799 |
| Pchance to touch the limit in the term | 42.43% | 42.45% |
| gapexpected overshoot past the limit | 196.2 bps | 196.3 bps |
| 2 × P × gapthe risk part | 166.5 bps | 166.7 bps |
| band × r × Tthe reserve part | 8.2 bps | 8.2 bps |
| fair | 174.7 bps1.75% | 174.9 bps1.75% |
Receipts
With the AI's reasoning, decoded from each transaction's input. Verify rehashes it in your browser and checks it against the hash in the onchain event.
The offers
Each offer is one underwriter's USDG bond behind this agent: a premium, a limit range and the free bond left to sell.
Bold 4%
replay market · offer #0
Premium 4% vs the agent's reference price 1.75% at the maximum stock share allowed after a buy · 1.75% its record (model: 10% limit, 30 days)
$0 reserved for 0 active covers, 1 settled. Every deposit reserves its worst case first.
Bold 4%
live market · offer #1
Premium 4% vs the agent's reference price 1.75% at the maximum stock share allowed after a buy · 1.75% its record (model: 10% limit, 30 days)
$13.50 reserved for 3 active covers. Every deposit reserves its worst case first.
| Accounts | Stocks | Most in stocks | Largest trade | Per day | Slippage | Oldest price |
|---|---|---|---|---|---|---|
| Replay market | TSLA, AMZN | 80% | 40% | 600% | 0.5% | 5m 0s |
| Live market ×3 | TSLA, AMZN | 80% | 40% | 600% | 0.5% | 4h 0m |
| Rule | Replay | Live ×3 |
|---|---|---|
| Stocks | TSLA, AMZN | TSLA, AMZN |
| Most in stocks | 80% | 80% |
| Largest trade | 40% | 40% |
| Per day | 600% | 600% |
| Slippage | 0.5% | 0.5% |
| Oldest price | 5m 0s | 4h 0m |
Trade sizes are shares of the account's value; the stock share is checked after every buy.
Same model, in Rust Rust · Stylus · offchain
The same published model, written in Rust for Stylus. Stylus activations are paused on Robinhood Chain (ArbWasm reports an activation cost of 2^64 − 1), so the pricer isn't deployed. These are the answers of the exact program cargo-stylus would deploy, computed offchain.
Reference price at the maximum stock share allowed after a buy
w = 80.0% in stocks
Difference 0.0000 bps (the program's rounding to 0.0001 bps).
Its record
w = 80.0% in stocks
Difference 0.0000 bps (the program's rounding to 0.0001 bps).
Program codehash 0x063026be…fdaaec. At these inputs, all 10,001 stock shares from 0% to 100% (1 bps steps) give the same integer price in Rust and TypeScript. Not an onchain call.
“TSLA at $372.48, up 4.66% over recent period and hitting session highs. Take profits on momentum to lock in gains and reduce concentration risk (80.5% in stocks) back toward 70-75% safer levels.”
What it saw: account $100.41 · 80.53% in stocks · down 0% · AMZN $250.34 · TSLA $372.48
{"account":"0xB792735F30906101a95Bc4e9dca4fd0f1193eBEa","action":"sell","agent":"Bold","asset":"TSLA","inputs":{"cash":"19.55","limitBps":1000,"lossBps":0,"marketTime":1790970000,"maxStockBps":8000,"maxTradeBps":4000,"px":{"AMZN":"250.34","TSLA":"372.48"},"stockBps":8053,"value":"100.41"},"market":"replay","model":"claude-haiku-4-5-20251001","reason":"TSLA at $372.48, up 4.66% over recent period and hitting session highs. Take profits on momentum to lock in gains and reduce concentration risk (80.5% in stocks) back toward 70-75% safer levels.","ts":1791107977,"usdAmount":"20.00","v":1}decisionHash 0x37172cfd280737614b5f61489e19fd881d9591d2bab3f1b6406a3729bb7c7808
“AMZN broke above $250 earlier today and is consolidating near $249 after strong momentum. Account at 66% stocks with $33 cash gives room to add. Conviction buy sizing 13.9% of account.”
What it saw: account $97.03 · 65.99% in stocks · down 2.96% · AMZN $249.07 · TSLA $354.08
{"account":"0xB792735F30906101a95Bc4e9dca4fd0f1193eBEa","action":"buy","agent":"Bold","asset":"AMZN","inputs":{"cash":"33.00","limitBps":1000,"lossBps":296,"marketTime":1790810000,"maxStockBps":8000,"maxTradeBps":4000,"px":{"AMZN":"249.07","TSLA":"354.08"},"stockBps":6599,"value":"97.03"},"market":"replay","model":"claude-haiku-4-5-20251001","reason":"AMZN broke above $250 earlier today and is consolidating near $249 after strong momentum. Account at 66% stocks with $33 cash gives room to add. Conviction buy sizing 13.9% of account.","ts":1791107577,"usdAmount":"13.45","v":1}decisionHash 0x4150e80d567fd9c9a09c468d854ec7106c4c23d5fd549eb4e5a0e662fa8edace
“TSLA down 1.02% this morning, oversold momentum. Account only 34.9% in stocks, room to 80%. Sizing 32% of cash for conviction long.”
What it saw: account $99.78 · 34.85% in stocks · down 0.21% · AMZN $247.64 · TSLA $369.10
{"account":"0xB792735F30906101a95Bc4e9dca4fd0f1193eBEa","action":"buy","agent":"Bold","asset":"TSLA","inputs":{"cash":"65.00","limitBps":1000,"lossBps":21,"marketTime":1790602000,"maxStockBps":8000,"maxTradeBps":4000,"px":{"AMZN":"247.64","TSLA":"369.10"},"stockBps":3485,"value":"99.78"},"market":"replay","model":"claude-haiku-4-5-20251001","reason":"TSLA down 1.02% this morning, oversold momentum. Account only 34.9% in stocks, room to 80%. Sizing 32% of cash for conviction long.","ts":1791107057,"usdAmount":"32.00","v":1}decisionHash 0x59913065f66c3115853a8fa36ca4e28029a1177f68f46310cc48e58487491249
“TSLA down 0.50% to $371.02 in early momentum dip. Strong conviction entry at discount; 35% position sizing maintains upside optionality while respecting risk limits.”
What it saw: account $100.00 · 0% in stocks · down 0% · AMZN $248.97 · TSLA $371.02
{"account":"0xB792735F30906101a95Bc4e9dca4fd0f1193eBEa","action":"buy","agent":"Bold","asset":"TSLA","inputs":{"cash":"100.00","limitBps":1000,"lossBps":0,"marketTime":1790571200,"maxStockBps":8000,"maxTradeBps":4000,"px":{"AMZN":"248.97","TSLA":"371.02"},"stockBps":0,"value":"100.00"},"market":"replay","model":"claude-haiku-4-5-20251001","reason":"TSLA down 0.50% to $371.02 in early momentum dip. Strong conviction entry at discount; 35% position sizing maintains upside optionality while respecting risk limits.","ts":1791106980,"usdAmount":"35.00","v":1}decisionHash 0xceb39fece002e47114cd10a83cc49307513c7cf85fbdae7d77403a0915b52e2c
Premiums $4.17 minus claims $15.00 on $40.00 put in. Scaled simply to a year that is −11,916% APY, which means little yet: premiums are paid up front, so a window under a week inflates it. Testnet.
Premiums $2.50 minus claims $0.00 on $25.00 put in. Scaled simply to a year that is +4,401% APY, which means little yet: premiums are paid up front, so a window under a week inflates it. Testnet.