The engine
How it works
Observable on-chain activity becomes a behavioral profile. No identity claims. No black boxes.
01
Read the public ledger
We fetch publicly available transactions for an address from Ethereum, Solana or TRON. Read-only. No wallet access ever.
02
Normalize the data
Transfers and program interactions become a normalized model: direction, tokens, token balances, swap protocols (Jupiter, Raydium, Pump.fun, Orca), timestamps and block numbers.
03
Extract features
We compute real metrics: wallet age, active days, holding durations, turnover, token diversity, protocol usage and activity patterns.
04
Score the behavior
Deterministic formulas produce explainable scores for risk, degen, patience, diamond hands, conviction, activity and more.
05
Classify the personality
Behavioral profiles like Night Trader, Meme Degen, Diamond Hands and Smart Money are derived with confidence levels.
06
Write the WalletVibe
A readable profile is generated — deterministically, with optional AI polish that never invents raw facts.
What we never do
SIGBOT never claims to identify the real-world person behind a wallet. We analyze behavior, not people. Profiles are probabilistic inferences from public data — they describe how an address behaves, not who owns it. We do surface publicly verifiable links: exchange hot wallets we recognize, the swap protocols an address trades on, and first-deposit funding sources. Recognizing a Binance hot wallet is a fact about the ledger, not a claim about a person.
Some on-chain accounts sign transactions on behalf of programs (launch bots, fee payers, treasury accounts) without moving their own balances. For those we report honestly what the ledger shows — program interactions and the fee payer doing the signing — and we never invent an owner where none is provable.
We never ask for private keys, seed phrases or recovery phrases. Payment is handled via public receiving addresses only.
Personality classifications
Every personality is derived from deterministic traits — holding periods, frequency, asset mix, activity timing, P&L direction — combined into a confidence-weighted classification. Primary and secondary traits are reported together.