Pump.fun trending: which signals a bot can move, and which it cannot

Trending feeds rank tokens on several inputs at once and none of the platforms document their weights. What can be established is which inputs are observable, which respond to coordinated trading, and which do not respond at all. That distinction is the difference between a campaign with a plausible mechanism and one built on a promise nobody can keep.

Volion Research Updated Jul 30, 2026 4 sections

The signals that appear to matter

Observable inputs to discovery ranking are: trade frequency, unique maker count, traded volume, market capitalisation, holder count and distribution, recency of launch, and social engagement such as replies and follows. Volume bots move trade frequency, maker count and volume directly. They influence market cap indirectly through price. They do not move genuine holder conviction, and no configuration changes that.
SignalObservableBot can move it
Trade frequencyYes, on chainDirectly
Unique makersYes, on chainDirectly
Traded volumeYes, on chainDirectly
Market capYes, derivedIndirectly, via price
Holder countYes, on chainSuperficially
Holder convictionNoNot at all
RecencyYesNot at all
Social engagementYes, platform sidePartially
Weights are undocumented and change without notice. What is stable is which inputs exist and which respond to trading activity.

Holder count deserves the qualifier "superficially". A fleet of wallets holding a token does raise the count, but distribution analysis makes fleet-held supply distinguishable from organic holders, and any ranking weighting distribution quality rather than raw count will read it accordingly.

The shape of activity matters more than its size

At identical volume, activity spread across many wallets and many slots scores differently from the same volume concentrated in few wallets or a single spike. Feeds that weight unique participants separately from volume are measuring distribution, so five hundred wallets trading once is a materially different input than one wallet trading five hundred times, even though the volume line is the same.

This has a direct configuration consequence. If your budget is fixed, you are choosing between fleet breadth, trade size and duration, and those choices produce different signal profiles at the same cost. Wide and slow favours maker counts and sustained ranking position. Narrow and fast favours a volume spike and a brief placement.

Duration interacts with how feeds compute their windows. Rankings that operate on rolling windows reward presence across the window rather than a peak inside it, so a campaign that spends everything in fifteen minutes buys a shorter placement than the same spend across several hours. Which is preferable depends on whether you are trying to be visible during a specific event or trying to hold a position.

The trade-off between wallet count and trade size is worked through in wallet count, and the maker-versus-volume distinction in makers vs volume.

Why nobody can promise you a placement

Trending is a ranking, which means it is relative. You are competing against every other token active in the same window, and the platform can change weights at any time without announcement. A tool can raise the inputs it is able to raise. It cannot control what competitors do, and it cannot control a formula it does not have. Any guarantee is therefore a claim about things outside the seller's control.

Two mechanisms make guarantees structurally impossible. First, ranking is competitive: identical activity ranks differently on a quiet afternoon than during a launch wave, because the threshold is set by other tokens rather than by an absolute number. Second, platforms adjust their ranking logic, partly to reduce the effectiveness of exactly this kind of automation. A weight that mattered last month may not this month.

What you can reasonably expect is that the inputs you paid to move went up, and that you can verify this on chain independently of anything we tell you. That is the honest boundary of the product. A vendor promising trending is selling certainty they do not possess, and the tell is usually that their pricing page contains a guarantee but no measurement.

Platforms are actively looking

Discovery platforms have obvious incentives to discount coordinated activity, and the signals they would use are the ones any chain analyst uses: wallet funding lineage, timing regularity, wallet age distribution, and the ratio of trades to distinct funding sources. Assume filtering exists and is improving. Designing around that assumption is more useful than assuming invisibility.

This cuts against the interests of anyone selling a volume bot, which is precisely why it is worth saying. A fleet funded from one source in a short window, trading at regular intervals in equal sizes, is trivially identifiable. Randomised sizes, spread timing and varied funding paths raise the effort required to identify it. Nothing makes it invisible, because the chain is a permanent public record and analysis tooling improves monotonically.

The practical implication is to treat volume as one input among several rather than a substitute for the others, and to expect diminishing returns as filtering improves. Detailed treatment in clustering risk.

Which of those signals a campaign can move, and at what measured cost per landed swap, is on the Pump.fun volume bot page.

Once a token graduates these platform signals stop applying and aggregator surfaces take over, which is a different ranking problem with different inputs. That transition, and what it costs on PumpSwap, Raydium, Meteora DLMM or Orca Whirlpools, is measured on the Solana volume bot overview. What aggregators index is on DexScreener trending.

Questions

How much volume do I need to trend on Pump.fun?
No honest figure exists, because the threshold is relative to what other tokens are doing in the same window rather than absolute. Any specific number quoted to you is either invention or a description of one particular past moment.
Does Pump.fun publish its trending formula?
No. Neither Pump.fun nor the major aggregators document their ranking weights, and they change them over time. What can be established is which inputs are observable and which respond to trading activity.
Do comments and favorites help?
They are platform-side engagement signals that appear to feed the calculation alongside trade data, so volume alone leaves part of the input set untouched. How heavily they are weighted is not documented, and treating them as a decisive lever would be overstating what is known.
Is a short spike or a long campaign better?
They buy different things. A spike produces a brief high ranking that decays quickly; sustained activity holds a position across more of a rolling window. Choose based on whether you need visibility during a specific event or over a period.
Can trending be guaranteed?
No. Ranking is competitive and the formula is undocumented and changeable, so no seller controls the outcome. A guarantee in this category is a claim about factors outside the seller's control.