Method v1.0

How a grade is made.
All of it, in the open.

The weights, limits and bands on this page are loaded from the same file the grading engine runs on. If the page and the engine ever disagreed, the page would be the one that changed.

One number, 0 to 100.

Each check scores 0–100. The grade is their weighted average. A check that could not run is left out and the remaining weights are rescaled, so missing data can never raise a grade. After averaging, the hard limits are applied. The result is the grade.

Six checks, fixed weights.

Hard limits override everything.

When one of these is true, the grade cannot rise above its ceiling, however well the coin does elsewhere. Every limit that applied is printed on the read.

Five verdicts.

Verdicts describe how much the evidence on chain supports the coin's story. None of them means "safe", and none of them is a recommendation to buy or sell.

Everything at one block.

A read starts by fixing the newest block. Curve reserves, holder balances, the exit test and the creator's sells are all read at that height, so the numbers in one report can never contradict each other. The block number is printed on every read and every share card, so anyone can query the same contracts at the same height and get the same values.

Where the data comes from.

  • Robinhood Chain RPC (public endpoints, rotated): the Pons factory, the bonding curve, token metadata, balances and the simulated sell.
  • Blockscout explorer API: the holder list, holder count, and every transfer to and from the creator's wallet.
  • fxtwitter: age and follower count of the X account written into the token.
  • RDAP: the registration date of the website's domain.

No numbers are written by a language model. Sentences in a read are assembled from the measured values by code.

What you can check yourself.

Enforced by the chain

  • Every finding links to its transaction, wallet or contract
  • The read is pinned to a block you can query
  • Creator tax and launch terms are read from the contract, not from a website
  • The exit test is a real call to the curve's sell function

Kept by ALETHEIA

  • The record is stored by our server and shown in full, misses included
  • Follow-ups are measured by the server at 24 hours and 7 days
  • Weights only change with a new method version, listed below

What a read can't see.

  • Who owns a wallet. She reports what wallets did, never who is behind them.
  • Off-chain promises, private deals, or what a team says in a group chat.
  • The future. A clean coin can still go to zero, and a weak one can still run.
  • Coins that were not launched through Pons. For now, ALETHEIA reads Pons coins only.

How we read our own coin.

$ALETH gets no exception. Once the contract is live it is graded by the same engine, with the same weights and limits, and its reads go on the same public record as every other coin. If it reads badly, that stays up too.

Changelog

  • v1.0 · first public method: six checks, six hard limits, five verdicts, 24h and 7d follow-ups.