Proof of which crypto callers actually make money — and a place to backtest it.
Score them against what the chain actually did. Kaizen Analytics timestamps every call, pulls the price history for that window, and computes eleven families of metrics including win rate, return multiple and dump behaviour. A simulator then replays a strategy with slippage and take-profit ladders applied.
Because they are selected. A channel posting its winners is telling the truth about each individual call and lying about the distribution, and there is no way to audit that from the outside.
The only honest measure is every call, timestamped when it was made, scored against the price history that followed. That removes the ability to choose which trades count.
Different questions that a single number hides. Win rate says how often a call went up; return multiple says by how much; dump metrics say how fast it fell afterwards, which is where most of the damage to followers actually happens.
A caller with a high win rate and brutal dump behaviour is not a good caller, they are a fast one. Splitting the measurement is what makes that visible.
Because a backtest without slippage is a fantasy, particularly on low-liquidity tokens where the act of buying moves the price against you.
The simulator models slippage and take-profit ladders, so a strategy is evaluated on something closer to what a real position would have returned rather than an idealised fill at the posted price.
Timing. If the same wallets consistently buy shortly before a call is published, the call is not a signal, it is an exit.
Checking wallet activity against publication time turns a suspicion into something measurable.
Because the output of this system is reputational. Leaderboards that rank people by performance are worth attacking, and an unauthenticated or replayable endpoint is an invitation.
Every request carries an HMAC signature and a nonce with a time window, so a captured request cannot be resent later. The engine is FastAPI, the dashboard is Next.js, and the signing applies across the whole boundary.
A launched analytics product pairing a FastAPI engine with a Next.js dashboard, with signed, replay-protected APIs end to end.
| Scoring | 11 metric families including win rate, return multiple, dump |
|---|---|
| Data | On-chain activity and DEX price history |
| Simulation | Slippage modelling and take-profit ladders |
| Detection | Smart-wallet front-run timing checks |
| API security | HMAC-signed requests with nonce replay protection |
| Engine | FastAPI |
| Dashboard | Next.js leaderboards and strategy workspace |
Score every call, not the ones the caller shows you. Timestamp each call at publication, pull the price history for the window that followed, and compute win rate, return multiple and dump behaviour separately. A single headline number hides callers whose winners are fast and whose losers are total.
Because on low-liquidity tokens your own order moves the price. A backtest that fills at the posted price overstates returns, sometimes by a wide margin. Kaizen Analytics models slippage and take-profit ladders so a simulated strategy resembles what a real position would have done.
A guard against a captured request being sent again later. Each request carries a one-time nonce with a short validity window, stored so a repeat is recognised and rejected. Signing alone proves who sent a request; replay protection proves they meant to send it now.
You can detect the timing pattern. If particular wallets repeatedly buy a token shortly before a call is published, that ordering is measurable from on-chain data. It does not prove intent, but it turns a suspicion into evidence that can be shown on a leaderboard.
Say it in plain language; the wallet does it safely.