How the Pump.fun Ranking Algorithm Works in 2026
A practical breakdown of the signals Pump.fun uses to rank and surface tokens — volume, holders, comment velocity, and bumps — plus how each feed (new, trending, top) weights them differently.
Pump.fun does not publish its ranking weights, but the behavior is observable and consistent. Understanding how the feeds differ — and which signal each one leans on — tells you exactly where to spend effort at each stage of a launch.
Three feeds, three jobs
“Ranking” on Pump.fun is not one list — it is several surfaces, each optimizing for something different. A token’s journey is usually new → trending → top, and each hop rewards a different behavior.
| Feed | Optimizes for | Dominant signal |
|---|---|---|
| New / recently created | Freshness | Recency + bumps |
| Trending / hot | Momentum right now | Volume + comment velocity |
| Top / by market cap | Sustained size | Accumulated market cap |
The signals that feed the score
- Volume velocity — not just total volume, but how much is happening right now. A burst of buys in a short window weighs far more than the same volume spread across a day.
- Holder count & growth rate — breadth of distribution. Rapidly growing distinct holders signals organic demand.
- Comment velocity — an active token page with fresh comments reads as a live community and lifts the trending score.
- Bumps — the explicit visibility action; frequent bumps keep a token near the top of the “new” surface.
- Recency decay — everything is time-weighted. Signals from the last few minutes count for far more than signals from hours ago.
Which lever to pull, and when
Stage 1 — Get out of “new” (first minutes)
Fresh tokens get a brief slot in the new feed. Convert it with immediate volume and bumps so the token registers momentum before recency decay pushes it down the list.
Stage 2 — Break into “trending” (first hour)
Trending is where organic flow lives. It is dominated by volume velocity and comment activity. This is the make-or break window — sustained buys plus a genuinely active chat are what tip a token onto the surface real traders watch.
Stage 3 — Hold in “top” (post-traction)
Once market cap accumulates, the top feed provides durable, lower-effort visibility. By here the flywheel should be organic; your job shifts from manufacturing momentum to sustaining community and narrative.
Practical implications
- Timing beats volume. The same SOL concentrated into a five-minute window outperforms the same SOL dribbled out over hours.
- Push every signal together. They compound — volume without a live chat, or comments without buys, under-performs coordinated pressure on all of them.
- Respect the decay clock. If you miss the launch window, re-entry requires a fresh catalyst, not more of the same slow activity.
This is precisely the problem a rank engine is built to solve: coordinating volume, comments, bumps, and distribution on the algorithm’s clock rather than yours. See Why Most Pump.fun Tokens Fail for how these signals map onto the launches that actually survive.