One candle at a time.
FIREBALL launch history, replayed in chronological order. Signals see closed candles only. Orders fill at the next minute’s open with modeled costs. Creator-fee estimates enter the treasury as volume accumulates; fees are not profit. Move the slider to inspect what the bot knew at each step.
FIREBALL / SOL
Rolling 30-minute checkpoints
Same fixed rules throughout. No parameter fitting, no resets between checkpoints.
| End UTC | Buys | Sells | Fees SOL | Cash SOL | Marked P/L |
|---|
Token obligations
50% goes to the simulated holders immediately. The remainder releases over 24 hours. Sales do not erase tokens owed. This sample ends before that full cycle.
What changed
Fee income arrives as observed volume accumulates, with a one-minute collection lag. The paced profile uses 5–10% entries, with five-minute spacing early and three-minute spacing thereafter. Older inventory can sell into profitable local rallies even after a newer buy. Both profiles retain the 0.5% dip trigger; no daily purchase cap or fixed SOL buy ceiling.
Data, assumptions and limits
GeckoTerminal minute candles from the launch bonding curve and migrated PumpSwap pool. Historical trade direction is unavailable for most of this range: both presets deliberately use candle-only entry and exit rules. The comparison uses the earlier candle-only rules with identical estimated fee income. Estimated fees are external contributions, never counted as strategy profit. The rate follows Pump’s published fee schedule; these are not verified receipts. A 4-candle warmup precedes entries. This historical experiment is not a live bot or an untouched out-of-sample test.
Download full results and source hashes