Scoring, strategy and risk
How tokens are scored
Each new token is observed for about 20 seconds, then scored every 10 seconds until it qualifies, is rejected, or expires. Ten dimensions are scored 0–100, each with a confidence:
| Dimension | What it looks at |
|---|---|
| Liquidity | SOL in the bonding curve, price impact of your size |
| Volume | 60-second volume and acceleration, discounted for wash trading |
| Momentum | Price change, net SOL flow, distance from the peak |
| Transactions | Buy share, pace, number of distinct buyers |
| Holders | Holder count and concentration |
| Smart money | History of early buyers seen by your app |
| Bonding curve | How far along the curve is |
| Token age | Enough evidence without being stale |
| Dev activity | Creator holding and selling |
| Market | Launchpad-wide buy/sell ratio |
Missing data is never treated as good: it lowers confidence instead.
Hard stops
Some findings reject a token no matter the score, including dev dumps, bundled launches, wash trading, dangerous mint features (freeze or mint authority, transfer hooks, permanent delegate), extreme concentration and thin liquidity. Manual buys are blocked by hard stops too.
Presets
Conservative, Balanced and Aggressive set the score threshold, position size and exits together. Changing any value switches to Custom.
Risk limits
Max SOL per trade, total exposure, concurrent positions, daily loss limit, cooldown after a loss, and a SOL reserve for fees. The kill switch stops new entries and sells everything.
Exits
Stop loss, take profit, trailing stop, max hold time and creator-dump detection run every two seconds — also when the window is closed to the tray.