How to find new memecoins early

You find them on new-pair feeds: a cross-chain launch feed, a screener's new-pairs page, or a per-venue tracker. You filter by pool liquidity, pool age and a contract safety check rather than by trending position. Setting that up takes about ten minutes. What it does not do is make early buying work, and the reason is arithmetic rather than skill.

The maths you are trading against

Start here, because every method below is downstream of it.

CoinGecko published research on the lifespan of pump.fun tokens, updated 23 June 2026, covering 18.67 million tokens launched between 14 January 2024 and 18 June 2026. Across that set, 68.67% — 12.8 million tokens — recorded their last trade on the day they were created. Roughly 80.37% were dead within two days. Only 4.55%, about 850,000 tokens, were still trading past 90 days. CoinGecko flags its own caveat: the 90-day number is understated, because the study tracks bonding-curve trades only and misses activity after a token migrates to a DEX.

Take the first figure seriously. It is not a claim that most memecoins underperform. It is a claim that for around two-thirds of them, the entire trading history begins and ends inside a single calendar day. "Early" in that population is not an entry point ahead of a run. It is the only point.

A separate count put pump.fun's cumulative launches at about 11.9 million since January 2024, as of 10 June 2026, of which 18 tokens ever exceeded a $10M market cap and 96 exceeded $1M. That total does not reconcile with CoinGecko's 18.67 million, which is worth knowing in itself: these are estimates from different indexing methods. What both agree on is the shape. Millions of launches, low hundreds of outcomes.

Graduation rates tell the same story from the other end and are covered in why so few memecoins graduate. The short version: published rates range from 0.198% (a study of 832,941 mints between 8 May and 10 June 2026, which observed each mint for only about six minutes and states explicitly that its figure is a lower bound) up to "fewer than 2%" reported on 10 June 2026. The defensible range for a true 24-hour rate is roughly 0.5% to 2%, and any single number quoted without its measurement window is close to meaningless.

Why "early" at the launch moment is adverse selection

The instinct behind the query is that being earlier is being better positioned. At the launch moment specifically, the opposite holds, and it holds for a structural reason.

New launches are contested by automated buyers in the same block the pool goes live. Those bots read the mempool or the launchpad's program directly, they size their orders against the curve, and they are not waiting for a feed to update. By the time a new pair renders in your browser, several seconds have passed and a human has had to look at it. You are not competing for the first fill. You are competing for whatever is left after software with a latency advantage measured in milliseconds has taken what it wants. Sniping, explained covers the mechanics and why retail loses that race by default.

That is the ordinary version. The adverse-selection version is worse. Among tokens that are still buyable at the price you see, a disproportionate share are the ones the fast money declined — and a share of the rest were constructed so that early buying is the product. A cluster of wallets funded from one source can take most of the early supply at launch and sell into the first wave of organic buyers, which is exactly the wave a new-pair feed puts you in. Bundled launches and sniper wallets sets out how to see that pattern in the holder data.

So the honest framing of "early" is this: you are buying what the fastest participants either could not reach or did not want. That is not a reason never to open a new-pair feed. It is a reason to stop treating earliness as an edge.

Where new launches actually surface

There are four kinds of surface, and they differ in coverage rather than quality.

Cross-chain launch feeds show tokens from multiple venues and chains in one stream. That is the only view in which you can see that Solana, Robinhood Chain, Base, BNB and Monad venues are all producing launches simultaneously, and the only one where a venue's share of activity is visible. The Meme Central cross-venue launch feed is ours and is built for exactly this; it reflects tokens indexed by Meme Central rather than the whole market, which is a real limitation and one you should apply to every aggregator including this one.

Screener new-pair pages index liquidity pools rather than launchpads. DexScreener's new-pairs view is the most widely used and covers 60-plus chains as of 31 August 2026. Because it indexes pools, it sees a token when a pool exists — which on a bonding-curve venue is at graduation, not creation. Most guides do not mention that timing difference.

Per-venue trackers are the launchpad's own interface. These are the earliest possible view of tokens on that venue and blind to everything else.

Terminals and bots sit on top of the same data with execution attached. Several are owned by the venues they surface — Padre is owned by pump.fun, Moonshot is owned by or affiliated with Jupiter. A terminal with a stake in what you discover is a conflicted discovery tool.

Whichever you use, the volume is the problem. pump.fun alone was producing around 42,000 tokens in 24 hours as of 10 June 2026. No human reads that. Filtering is not an optimisation of the method; it is the method.

The filters that do the work

Three filters remove almost everything, and they are all pool properties rather than opinions.

Minimum liquidity. Set a floor and treat everything under it as unbuyable rather than as an opportunity. Thin pools are not cheap entries; they are positions you cannot exit near the displayed price, because your own sell walks the price down through your fill. The right floor depends on your size — the working question is what fraction of the pool your order represents, not the pool total in isolation.

Pool age. Filter for a minimum age rather than a maximum. This is counterintuitive on a page about finding tokens early, and it is the single most useful setting on the list. A token with an hour of history has survived the first hour, which is where the majority of failures happen. You give up nothing that was reachable anyway, because the first minutes belonged to the bots.

A contract safety check that runs before you look at the chart. On Solana that means checking mint and freeze authority and LP status; on EVM chains it means honeypot, trading tax and ownership checks. Per-token pages on Meme Central carry a per-chain safety report — GoPlus for EVM chains, RugCheck for Solana — so the check happens in the same place as the discovery. Whatever tool you use, understand that a passing score is a check for known failure modes and not a verdict, which is the argument in what a token safety score actually measures.

Notice what is not on the list. Trending position, because on DexScreener those surfaces are influenced by Boosts, a paid product. Volume alone, because on a young pair it is trivially manufactured by trading against yourself. Social buzz, because it is the cheapest thing here to fake.

What the first hour of holder growth tells you

Holder count is the most misread number on any screener, and the shape of its growth is more informative than its level.

Organic growth in the first hour is irregular — arrivals cluster around whatever caused them, and the wallets have unrelated funding histories and prior activity. Manufactured growth is smooth, or it is a step function, and the wallets share a funding source, were created in a batch, or hold nothing else. A hundred holders where thirty wallets were funded by the same address in the same five minutes is not a hundred holders.

The distribution matters more than the count. If the top ten addresses hold most of the supply, the price you see is set by whoever chooses to sell first, and every other holder is the exit liquidity for that decision. How to read a token's holder distribution covers what the concentration numbers actually mean and where the standard displays mislead — including that a launchpad's own contract, a locked LP position and a burn address all appear as large "holders" and none of them are people.

One correlate worth knowing precisely because it is so easy to misuse: in the May–June 2026 dataset, launches advertising a Telegram graduated at 1.485% versus 0.166% without — an 8.94× differential (Cox hazard ratio 5.40), and an initial market cap above the platform default was the strongest single predictor of graduation (hazard ratio 4.51). Neither is a strategy. Adding a Telegram link is free, so the correlation is picking up something about deployer effort and coordination rather than any property of the token — and every one of these signals is equally available to somebody constructing a launch designed to be bought.

A workflow that takes ten minutes to set up

1. Choose one feed and one chain to start. Cross-chain breadth is useful later. Learning what normal looks like on one venue is worth more at the beginning than covering five.
2. Set a minimum liquidity floor and a minimum pool age. Ten to thirty minutes of age is enough to remove most of the instant failures. Write the numbers down so you are applying a rule rather than a mood.
3. Sort by age, not by gainers. A gainers list is a list of things that have already moved, which is the opposite of what you asked for.
4. Run the contract check before opening the chart. Mint and freeze authority on Solana; honeypot, tax and ownership on EVM. If it fails, close the tab. Do not proceed to look for reasons the failure does not matter.
5. Read the holder distribution, not the holder count. Concentration in the top wallets, funding relationships between them, and whether the large positions are contracts or people.
6. Check the pool, not the promise. Whether liquidity is locked or burned, and for how long. A lock is verifiable from outside; a promise in a Telegram channel is not.
7. Size the position for total loss. Not as a caution. As the modal outcome, on the numbers at the top of this page.
8. Set an alert and stop watching. Continuous monitoring is how people talk themselves into positions their own filters rejected. The alert-and-webhook side of this is in how to monitor new token launches in real time.

What this doesn't tell you

It does not tell you which tokens to buy, and no version of this article ever will.

More usefully, it does not tell you that any of the above improves your expected outcome. The filters remove a category of clearly bad tokens. They do not identify good ones, because the evidence that a token was good is generated after the point at which you would have had to buy it. Nothing in the published data supports the idea that a screening method turns the base rates at the top of this page into a positive expectation, and anyone selling you one is selling the method rather than using it.

Two limits worth stating plainly. The CoinGecko survival data is pump.fun-specific, and pump.fun is not the whole market — Robinhood Chain venues in particular now generate comparable fee volume, and their survival characteristics have not been studied the same way. And venue-level launch and graduation figures move week to week; our per-venue launch and graduation data is refreshed rather than written, and it reflects tokens indexed by Meme Central, not the whole market.

The broader context is in why most memecoins go to zero. For scale: CoinGecko put total memecoin market capitalisation at a peak of $150.6bn in December 2024, down to $47.2bn in November 2025 — a roughly 69% drawdown at the category level, in a period when launch volumes were rising.

Frequently asked questions

Is there a way to find memecoins before the bots do?

No. Bots read the chain or the launchpad program directly and act within the same block; any interface you refresh in a browser is behind them by design. The realistic question is not how to be first but whether being first was ever the thing that mattered, and on the survival data it was not.

What is the best filter to set on a new-pair feed?

A minimum pool age, which sounds contradictory on a page about buying early and is not. Most failures happen in the first minutes, and those minutes were never available to you anyway. Adding a minimum liquidity floor and a contract safety check removes most of what remains.

Does a high holder count mean a token is safe?

No. Holder count is cheap to manufacture by funding many wallets from one source, and it says nothing about concentration. A token can show hundreds of holders while ten addresses control most of the supply. Read the distribution and the funding relationships, not the headline number.

Why do boosted tokens appear at the top of screeners?

Because Boosts are a paid product. DexScreener has no subscription and no token; it is monetised through Boosts and Ads, and a Boost purchases a time-limited increase in visibility within trending and discovery surfaces. The badge is public, so the placement is disclosed. It buys attention, not legitimacy.

How many new memecoins launch per day?

pump.fun alone was producing around 42,000 tokens in 24 hours as of 10 June 2026. No credible source aggregates a reliable all-chain daily total across every venue, so treat any single cross-chain launch figure you see with suspicion, including in this article's absence of one.

---

Before you buy anything this young, check whether the pool can be pulled

The one commitment a buyer can verify from outside, without trusting anybody, is whether the liquidity behind a new pool is locked. Team Finance — built by TrustSwap, which also builds Meme Central — locks LP tokens for a fixed term on Ethereum, Robinhood Chain, Polygon, Base and BNB, and where a token's liquidity is locked that shows as a verified badge on its page in the feed. It settles exactly one question. It does not stop a deployer selling the supply they kept, which is the more common way holders of a new token lose money, and it is not a substitute for any of the checks above.

---

Nothing here is financial, legal or tax advice. Memecoins are extremely high-risk: most lose most of their value, and the majority of tokens launched never reach a decentralised exchange at all. Never spend money you cannot afford to lose entirely. Meme Central does not recommend any specific token. Data described as Meme Central's own reflects tokens indexed by Meme Central and is not whole-market data.