Signs a launch is already over

A launch is over when new participation has stopped and the remaining flow is existing holders trading with each other. That definition is deliberately about participation rather than price, because participation stops first and price follows some minutes later. The gap between those two moments is where most of the late buying happens.

The signals below are all descriptions of who is trading rather than of what the price is doing. None of them is a prediction, and each one is paired with the observation that would refute it.

What over actually means

There are three different things people mean by a launch being over, and conflating them produces most of the confusion.

The first is mechanical: the curve has completed and the token has moved to a pool venue. That is an event with a timestamp and it is not an ending in any meaningful sense.

The second is attentional: the launch is no longer visible on the screens where people find new tokens, because activity has fallen below the thresholds those screens use. This is usually reversible and often temporary.

The third is structural: no new addresses are arriving, so every purchase from here is made by somebody who is already holding. This is the one worth detecting, because it changes what any subsequent price movement is made of. When a rise happens with no new addresses, the SOL producing it came from inside the existing set.

The one measure that carries most of the weight

If you can track only one thing after the opening window, track the count of addresses buying for the first time, per minute. Almost every other signal on this page is a slower, noisier proxy for that number.

The reason it dominates is arithmetic. A launch that keeps attracting first-time buyers has an external source of flow. A launch where the same addresses trade repeatedly has an internal one, and internal flow is finite by definition: it is bounded by what those addresses hold and are willing to commit.

The measure is not hard to obtain. Every swap names its signer, so a running set of addresses seen so far turns each new transaction into a first-time flag or not. Feeds rarely surface it, but a transaction list on a public Solana explorer contains everything required.

The quiet signals, in the order they appear

These tend to arrive in roughly this sequence. The ordering is a general pattern rather than a rule, and any of them can appear alone.

1. First-time addresses per minute falls toward zero while total trade count holds. The earliest and clearest signal.
2. Median trade size rises while trade count falls. Fewer participants doing larger transactions is the signature of a thinning set.
3. Buy and sell transactions approach parity among the same addresses, which is what circulation looks like when nothing is entering or leaving.
4. Gaps between trades lengthen and become irregular in a specific way: long silences punctuated by clusters, rather than a steady thinning.
5. The largest early holders reduce positions in steps rather than at once, which keeps the chart from breaking while the position leaves.
6. Metadata activity stops. Whatever social or descriptive presence accompanied the launch goes quiet before the trading does.

Signal five is the one most often misread as strength. Stepwise selling produces a chart that looks stable while a meaningful share of supply changes hands, and stability under distribution looks identical to stability under accumulation on a price line alone.

Telling a pause from an ending

Not every quiet period is an ending. Launches go quiet for ordinary reasons: attention moves to something newer, a broader market moment pulls flow elsewhere, or the row simply drops off a sorted screen for a few minutes. The distinction is worth making carefully because the two look similar for the first several minutes.

The last row is the decisive test and it requires patience, because it can only be run when attention returns. A launch that is paused will pull in unfamiliar addresses when it is visible again. A launch that is over will produce more transactions from the same set, which looks like recovery on a chart and is not one.

Why the chart is the last thing to know

A price line reports the last trade. It has no way to express who made it or whether that party had traded before. Consequently a price line can be entirely flat, or rising, while the underlying participation collapses, and it will show nothing until the internal flow runs out.

The lag is not small. If a closed set of holders continues trading among themselves, the chart can look healthy for a considerable period, and the eventual move happens when the last holder willing to buy stops buying. That moment is not visible in advance on the line, and it is visible well in advance in the address data.

This is the strongest practical argument for reading participation rather than price on new launches. The price is a summary of the least informative field, and it is the field every interface makes largest.

Concentration after the fact

After the opening window, concentration becomes readable in a way it was not earlier, because there is enough history to see whether it is changing.

The useful observation is the direction of concentration over time. If the largest holders are holding a rising share, supply is consolidating and the number of parties who could sell is shrinking. If they are holding a falling share while no new addresses appear, the supply is moving to the addresses that arrived during the opening window, which is the pattern with the least favourable arithmetic for whoever arrived last.

Both readings need the same caveat as everywhere else on this site: addresses are not people, so a falling top-holder share can be one party spreading a position across more of its own addresses. Concentration measured on addresses is an upper bound on dispersion, not a measurement of it.

Counts fall before amounts do

Trade count is a more sensitive instrument than traded amount, and it moves earlier. The reason is that participants leave one at a time while the remaining participants often trade larger, so the amount holds up while the count is already falling.

Watching the two together gives a cleaner signal than either alone. Count falling with amount is ordinary cooling. Count falling while amount holds or rises is consolidation into fewer hands. Count holding while the address set is static is circulation, which is the specific condition this page is about.

None of this requires sophisticated tooling. It requires recording two numbers per minute and one running set of addresses, which is well within what a spreadsheet and an explorer can do for anyone willing to spend an hour on a single launch. The transaction and account structures that make this possible are set out in the Solana developer documentation, and a reader who understands the signer field can build the whole measurement from it.

Where the first sell pressure lands

The first sustained selling on a launch is informative less for its size than for where it lands. Selling that is absorbed by first-time addresses is a transfer from the early set to a wider one, and it leaves the launch with more participants than it started the minute with. Selling that is absorbed by addresses already holding is a redistribution inside a fixed set, and it leaves the launch with the same participants and a different internal allocation.

These two produce almost identical price behaviour and completely different situations. On a chart they are one event. In the transaction list they are trivially separable, because every buy names a signer and the signer either has prior history on this token or does not.

There is a third case that deserves separating from both. Selling absorbed by addresses that appear for the first time but share a recent funding source with the sellers is not a transfer to a wider set at all; it is an internal move that looks like external absorption in every field a feed displays. This is the specific pattern where an address-only reading is at its weakest, and the funding check is what recovers it.

The practical instruction is short. When sell pressure begins, stop watching the price and record who is on the other side of it for the next twenty transactions. That single observation will usually settle whether what you are watching is a launch broadening out or a launch changing hands internally, and it is available immediately rather than in retrospect.

A worked decay example

Round numbers, chosen for readability, describing no real launch. Consider six consecutive five-minute buckets after an opening window.

The trade count falls by under forty per cent across the whole sequence, which on a chart of activity looks like mild cooling. First-time addresses fall by effectively one hundred per cent, and median size nearly quadruples. A screen showing only the first column would report a launch that was still busy at bucket six.

This is also the shape that sustained paid activity produces, and it is worth naming rather than implying. A professional Solana volume bot is bought precisely to keep a trade count from collapsing after organic attention moves on, which is a legitimate service with a stated price and is also exactly why a steady count is not evidence of continuing interest.

A checklist for the hour after

• Keep a running set of every address that has traded, and flag first appearances.
• Bucket the data in fixed five-minute intervals rather than eyeballing a stream.
• Record trade count and median size per bucket, not just totals.
• Track the combined share of the ten largest holders across buckets.
• Note when metadata or social activity stops, with the time.
• Distinguish a bucket with zero first-time addresses from one with a few.
• Re-run the check when the row becomes visible again, and compare address composition rather than trade count.

What would change each reading

Every signal above has a specific refutation, and stating them is the difference between a reading and an opinion.

• Zero first-time addresses is refuted by a cluster of new, independently funded addresses arriving together.
• Rising median size is refuted if the count of participants rises alongside it, which indicates larger participants arriving rather than small ones leaving.
• Stepwise holder reduction is refuted if the receiving addresses turn out to be first-time buyers rather than existing holders.
• Circulation is refuted by any sustained period where a meaningful share of buys come from addresses with no prior history on this token.

Writing the refutation down at the time is what makes the method self-correcting. A reading that was refuted an hour later is information about the method, and a reading with no refutation attached cannot generate any.

Where this reading stops

None of this predicts price. A launch that is structurally over by the definition above can still move sharply, because a small closed set can produce large movements in a shallow venue, and because attention can return from outside for reasons nothing on the chain records.

The address-based method also degrades as time passes. Over hours and days, addresses accumulate history elsewhere, first-time flags become less meaningful, and the clean separation between inside and outside the set blurs. The technique is sharp in the first hour and progressively duller after that.

What it delivers, inside its window, is an accurate description of whether a launch is still recruiting. That is a narrower claim than most launch commentary makes, and it is one of the few claims about a live launch that can be checked against the record while it is still being written.