MELT's First 24 Hours: How Did Buying Support Change as Price Rose?

In short

A review of MELT's first day on Solana, using buy-trade share, net buy share, and wallet participation to examine two possible entry assessment windows and two exit assessment windows.

MELT's first day brought together several things that are easy to conflate in a new token market: a rising price, heavy trading, and growing wallet participation, followed by weaker buying support and a deep decline.

One detail stands out. In some windows where conditions were weakening, more than half of the recorded trades were buys, yet sell volume in dollar terms was greater. A single green percentage could give a very different impression of the market.

We extracted MELT's complete first 24 hours from Signyx's on-chain database to examine two questions along the timeline: Where was there a case for assessing an entry more closely? And when did that case begin to weaken?

This is a historical case study. A, B, X1, and X2 are observation windows selected after reviewing the full price path. They illustrate a decision process; they were not signals actually sent at the time and do not constitute a validated trading strategy. Each observation uses only data from completed windows preceding its marked time. We do not calculate hypothetical trading returns.

First, define the token, time period, and data

All times in this article and its figures are UTC.

ItemScope of this study
NameMeltface / MELT
NetworkSolana
Full contract address5PmU19opZfFbRTVHzhFdUDp5Z4RdaMxVh2uHxqbGpump
Creation and initial pool launchNovember 8, 2026, at 12:38:33
Observation periodNovember 8 at 12:38:33 to November 9 at 12:38:33
Trade records52,016 parsed DEX trade events across 45,305 distinct transaction signatures
Participating addresses6,406 wallets with recorded trades, of which 6,080 made at least one buy
Total traded volumeApproximately $5.856 million, measured across those parsed trade events

MELT was about 43 hours old when we selected it, so its first-day observation period was complete. Trading was active enough, and the price path included both a substantial rise and a decline, making it useful for examining how a move developed and then lost support.

Trade events, transaction signatures, and wallet addresses are three different things. One on-chain transaction may produce multiple trade events, and one person may control multiple wallets. Total traded volume is therefore not the same as new capital entering the token, and wallet counts are not counts of unique people.

Price, traded volume, buy-trade share, and net buy share in this article are all recalculated from the same set of parsed trade events. Changes in the number of holders are reconstructed separately from historical token balances.

A modest narrative, a dramatic price move

The description associated with this contract on its Pump launch page is only: “face under pressure. $MELT on Solana”. It can be read as a character-driven meme built around trading pressure, facial expression, and emotion. That is an interpretation of public metadata, not evidence of established commercial value.

The material available for verification does not establish the strength of its long-term community, the team's ability to execute, or any durable revenue. This case therefore focuses on whether trading participation and buy/sell flows continued to accompany the price move after the narrative attracted attention.

About 2 hours and 49 minutes after launch, MELT reached its highest qualifying one-minute volume-weighted average price (VWAP) of the first day, approximately $0.00038246. To qualify, a complete minute needed at least $500 in traded volume. This avoids elevating a tiny trade into a conspicuous price peak.

The last complete minute that still reached $500 in traded volume was November 9, 01:23–01:24. Its VWAP was approximately $0.000005041, or 98.68% below that earlier peak. That minute contained six trade events totaling about $1,199. It is a historical reference point for the decline, but it does not imply that a large position could have been sold at that price.

Trading was extremely thin near the end of the 24-hour period. The final minute with any recorded trade was November 9 at 12:32, but it contained only one trade worth about $0.02. We do not use that tiny trade to calculate an end-of-day decline or treat it as a price with meaningful market depth.

MELT's first-day price and minute-by-minute volume: trading and price activity receded after the active phase

Figure 1: Minute VWAP is reconstructed from parsed trade events; minutes without trades are left blank. The peak is known only in hindsight and should not be treated as a sale price that could have been identified in real time.

Entry assessment A: At 13:06, participation broadened alongside the rise

Price moved quickly after launch. The more useful question was whether buying in dollar terms and the breadth of participation were changing along with it.

By 13:06 on November 8, the preceding complete five-minute window, 13:01–13:06, showed:

MetricValue
Five-minute traded volume$94,238
Buy-trade share54.5%
Net buy share+4.9%
Addresses making their first buy of this token254
Net change in holder addresses under the database's historical-balance definition+112

In the previous five-minute window, traded volume was about $36,370, there were 102 first-time buyer addresses, and net buy share was −1.9%. Comparing these adjacent windows tells us more than simply observing that price rose again: trading activity increased, more addresses bought MELT for the first time, and buying volume in dollars regained the lead.

That makes 13:06 a retrospective window for assessing a small exploratory entry. The case comes from several observations lining up, not from any fixed percentage automatically becoming a buy signal. Those 254 addresses could include bots or multiple wallets controlled by one person; they are not necessarily 254 new users.

Price had already risen by then, so an invalidation condition would still need to be defined in advance. If participation stopped broadening or the recovery range broke down again, would that call for an exit? Defining only when to buy, without defining when the original thesis no longer holds, leaves room for a substantial giveback even after an early rise.

Entry assessment B: At 14:46, what changed in participation after the pullback?

The second window worth examining came after a pullback.

In the five-minute window ending at 14:36, net buy share was about −0.4%, and the number of holder addresses under the database's balance definition fell by 96. By 14:41, net buy share had recovered to +3.3%, but holder addresses were still down by 48. Dollar flow improved first; the participation measure had not yet moved with it.

By 14:46, the next five-minute window showed:

MetricValue
Five-minute traded volume$172,399
Buy-trade share55.6%
Net buy share+1.8%
First-time buyer addresses244
Net change in holder addresses under the database's balance definition+214

The notable combination here was a price recovery, continued net buying in dollar terms, and a return to positive holder-address growth. First-time buyer addresses also increased from 119 in the window ending at 14:36 to 244.

Thus, 14:46 can serve as a second window for assessing an entry: observe the pullback, then ask whether participation data accompany the recovery. Still, five-minute traded volume had fallen from the preceding window, and net buying held only a 1.8% edge. This was a reason to reassess, not proof that price would continue upward.

MELT's active phase: price, buying and selling in dollar terms, and first-time buyer addresses

Figure 2: The horizontal axis marks the end of each measurement window. Price uses the preceding complete minute's VWAP; both shares and the first-time buyer count use the preceding complete five minutes. Actual availability also depends on chain confirmation, indexing, and computation delays. A, B, X1, and X2 are retrospective observation windows.

If buys were over half of all trades, why watch for selling pressure?

This is why two derived metrics used by Signyx are useful together:

Buy Trade Share

Number of parsed buy-trade events ÷ (number of parsed buy-trade events + number of parsed sell-trade events)

For this study, the unit counted is a parsed trade event, not a distinct on-chain transaction signature. The metric asks: Were more of the recorded trades buys?

Net Buy Share

(USD value of buys − USD value of sells) ÷ total USD traded volume

This asks: Did buying or selling account for more dollars in this window?

In the five-minute window ending at 15:01, MELT's buy-trade share was 55.9%, while its net buy share was −5.0%. There is no contradiction: more buy events can still represent fewer dollars of buying.

That window warranted attention, but it was not enough on its own to declare the rally over. Price rose further afterward and reached this study's qualifying peak during the complete minute from 15:27 to 15:28. This nearby counterexample matters: one negative reading can indicate slowing or disagreement; a sustained deterioration carries more weight.

Exit assessments X1 and X2: At 15:31 and 15:36, the holding case weakened in successive windows

The five-minute window ending at 15:31 showed:

  • Net buy share fell to −11.7%;
  • Buy-trade share fell to 48.3%;
  • Holder addresses under the database's balance definition fell by 101;
  • Traded volume was still about $170,000.

Selling clearly led in dollar terms, while the number of holder addresses also declined. Trading was still busy, but activity alone no longer made a sufficient case for continuing to hold. In hindsight, 15:31 was a window for assessing whether to reduce risk, trim a position, or take some profit.

In the next non-overlapping five-minute window, ending at 15:36, net buy share remained negative at −9.6%, and holder addresses fell by another 115. The persistence made the weakening of the earlier rally's buying support more notable, giving another reason to assess an exit.

Waiting for confirmation came at a cost, however. At 15:31 and 15:36, the preceding complete minute's VWAP was already about 27.1% and 47.1% below the peak identified in hindsight, respectively. These observations could not have placed a sale at the top, and they do not establish that a position could have been sold at those minute VWAPs. Real execution would also depend on slippage, fees, latency, and available depth.

After the decline, small rebounds can turn the percentages green again

In the five-minute window ending at 16:26 on November 8, the percentages looked favorable again: buy-trade share was 65.3%, and net buy share was +12.6%.

Looking only at the percentages might suggest that buyers had regained control. But the window's traded volume was only $1,861, with just 12 addresses buying MELT for the first time. That is a very different level of activity from the recovery phase, when five-minute volume exceeded $100,000 and first-time buyer addresses numbered in the hundreds.

The denominator therefore matters when judging a rebound. Once a percentage turns positive, is there meaningful volume behind it? Has participation recovered? A modest amount of buying can make a quiet window look strong; sustained support would require further evidence from subsequent windows.

Six five-minute windows: trade-count share, dollar-flow share, and participation can tell different stories

Figure 3: Every column is derived from the same parsed trade-event data. The 15:01 window shows a divergence between trade counts and dollar amounts; 16:26 shows positive percentages against low traded volume.

What can this one case actually tell us?

MELT's first day included a rising phase accompanied by broader participation and positive net buying in dollar terms. The later collapse does not erase that earlier move; nor does the earlier move mean that every later dip was worth holding through.

Three lessons are supported by this retrospective analysis:

  1. An entry case needs several observations to align. Improvements in price, buying in dollar terms, and participation together are more worth investigating than any one reading alone.
  2. An exit case should be assessed against the original holding case. Net selling across consecutive windows, falling participation, and weakening price reduce confidence in the original thesis.
  3. Percentages need an absolute scale. Buy-trade share and net buy share answer different questions. When trading is sparse, two positive readings are still not enough to show that the move has recovered.

These are interpretations of one historical sample, not universal thresholds, a success rate, or a validated token-selection strategy. We also lack sufficient evidence to attribute the eventual deep decline to a particular trade, wallet, or act by the project team.

When viewing a token on Signyx, one approach is to compare buy-trade share, net buy share, and participation over the same time window, then put your own entry and exit conditions into a strategy and keep testing them against live market data in simulation. This case is more useful for proposing and testing rules than for choosing, in hindsight, a set of parameters that would always have made money.

Methodology
  • Data scope: Creation and initial pool events for the full Solana contract address above, parsed trade events over its first 24 hours, historical token balances, and the market snapshots needed for context.
  • Price: Minute VWAP = total USD traded volume in the minute ÷ total token quantity traded in that minute. This is not the instantaneous high of a candlestick. The partial first minute is excluded from the research baseline; the peak qualifying minute must have at least $500 in traded volume.
  • Timing and availability: A marked time t is the end of the measurement window. Five-minute metrics use only [t−5 minutes, t); for example, 13:06 covers records from 13:01 up to, but not including, 13:06. This is the theoretical calculation boundary. Actual availability depends on chain confirmation, indexing, and computation delays. The selected windows are retrospective teaching examples, not evidence of historical alerts.
  • Participation: A first-time buyer address means an address making its first observed buy of this token in the available history. It does not mean a newly created wallet or a distinct person. Counts from overlapping five-minute windows cannot simply be added together.
  • Holder addresses: Counts are reconstructed from the database's historical balance events; a window's net change is the difference in positive-balance address counts at its two endpoints. Coverage and classification may differ from the website's holder count. We do not infer a count of actual users or common control of token supply from this measure.
  • Metrics not used: Parsed trade events and aggregated snapshots show an unexplained difference. Every traded-volume figure, event count, and trade-based share in this article uses the parsed trade-event basis. Liquidity, smart-money, whale, and linked-wallet metrics that have not been individually verified are not used to reach entry or exit conclusions.
  • Price versus market cap: Price declines in this article are calculated from price. Signyx Trending tracks multiples using market cap; the two should not be substituted for one another, particularly if supply changes.
  • Research limits: A retrospective study of one case cannot establish predictive ability. We have not modeled executable depth, fees, slippage, or latency, and we do not infer realizable returns from minute VWAPs.

This article is research, not financial advice. Figures are snapshots taken at the times stated, from the sources described in the article; markets move quickly and data can be delayed or incomplete. Do your own research and never risk money you cannot afford to lose.