Because nobody paying it is paying it. Six epochs a day compounds to roughly 1,498,000% a year. Funding that from a 0.7% fee take would require about 2.5× the token's market cap in trading volume every single day, forever. The arithmetic is real; the outcome is not available.

So the rule is inverted: pay a fixed share of the reserve rather than a fixed share of your stake. The reserve then decays geometrically instead of hitting zero, and the rate becomes something the protocol discovers rather than something it promises. 0.44% survives only as a ceiling. See the curve.

A fixed share of whatever the reserve holds, every four hours, indefinitely. Right now that is 0.190% per epoch. Left entirely alone it decays — in 90 days with no refills it would be 0.040%. BUY cycles push it back up.

Over the long run it settles at fee inflow ÷ total staked. That is the real yield of the protocol, and it is the one number nobody here can inflate.

The model decides, and the chain proves it decided first. Each cycle the sealed model outputs one action; the operator wallet publishes its hash, waits for a block, then executes — and the execution transaction reveals the preimage atomically. If the revealed decision does not match the commitment, the trade reverts.

We do not claim contract-level autonomy, because that would still be a lie. We claim something checkable instead.

Sealed. But the modelDigest on every receipt commits to it. The moment the preimage is published, every cycle ever run is verified at once — and if it is never published, the digest still proves the configuration never changed. The model card lists the four ways to catch that being false.

No, and not as a matter of policy. BoxVault has no owner. It exposes no withdraw, no rescue, no sweep, no upgrade path, and no pause switch on withdrawals. There is a test in the repository that fails the build if any of those names ever appear in its ABI.

Staking is non-custodial: tokens sit in the vault contract against shares your wallet owns, and unstake works in the same block you staked. No team wallet is anywhere in the path.

Three things, all in the contract. The target must be listed by an on-chain registry the operator does not control. The size is capped at 10% of intake per cycle and 25% per rolling day. And the only way to sell a hunted position returns ETH to the treasury, which itself has no exit.

A bad call costs one slice of one cycle. It cannot cost the intake.

The loss gets a receipt in the same format as a win. Failed executions are explicitly abandoned on-chain with a reason and counted in abandonedCommits — currently 0 — so a bad cycle shows up as a recorded failure rather than a gap in the numbering.

When the reserve is unusually large relative to what is staked, the drip would exceed 0.44% of stake. The ceiling clamps it and the remainder stays in the reserve rather than being paid out.

This is deliberate: without it, a single fat BUY could be dumped on whoever happened to be staked that one epoch, which rewards timing rather than holding. Withheld to date: 0.00 SIG across 0 capped epochs. It is withheld, not lost — it funds later epochs.

No. $SIG is a memecoin experiment with no promise of profit, yield, or utility — the rebase is explicitly a share of a reserve that can decay toward nothing. Nothing here is a recommendation, and nobody involved is a licensed advisor.

Trade it like the toy it is, or better, just read the ledger.