AI

Multibuy Convergence: How the Score Is Built, and How It Gets Faked

Multibuy convergence is a useful signal until someone learns to stage it — here is how the score works and where it breaks.

· 6 min read · Blackhat Empire

What "Convergence" Actually Means

A multibuy convergence score is not a measure of quality. It is a measure of clustering — how many distinct wallets buy the same token inside a short window, weighted by how independent those wallets appear to be.

Strip the branding off any AI agent or scanner that advertises this and you usually find the same skeleton underneath:

  • Wallet count in a rolling window. Ten buyers in ninety seconds scores higher than ten buyers over six hours, because the second pattern is just normal traffic.
  • Wallet age and history. Fresh wallets with one or two transactions carry less weight than wallets with a track record.
  • Size relative to the wallet's own baseline. A wallet that normally buys 0.4 SOL and suddenly buys 4 SOL is more interesting than a whale doing whale things.
  • Graph overlap. If the "independent" buyers all funded from the same source, or traded the same five tokens in the same order, the score should collapse. Good models penalise this. Lazy ones ignore it.
  • Entry proximity. Buys spread tightly around the same price band suggest coordination or a single decision executed by many hands.

That is the honest version. The output is a probability that something coordinated is happening — not a probability that the token survives. Those are different questions, and conflating them is how people lose money.

Where the Score Gets Faked

Convergence is one of the easiest signals to manufacture, because the inputs are cheap.

Wallet farming. Create fifty wallets, fund them from a handful of sources, let them age a few days, then buy together. Any model that counts wallets without weighting funding lineage gets fooled instantly.

Funding-graph laundering. Route the seed money through an exchange deposit, a bridge, or a chain of transfers before it reaches the buyer wallets. The on-chain link still exists, but it is now several hops deep, and shallow graph analysis stops seeing it.

Timing theatre. Bots can schedule buys to land in a tight band, which maximises the "tight window" component while the actual intent is to trigger scanners and the humans watching them.

Size spoofing. One large buy plus twenty dust buys looks like broad participation. It is one participant and a lot of decoration.

Score gaming across tokens. Some wallets farm convergence scores on small tokens specifically to build a reputation profile, then use that profile on the token that matters. The history is real. The signal is not.

If a convergence score rises but holder distribution stays concentrated, the score is describing activity, not ownership. That gap is the tell.

What a High Score Does Not Tell You

It does not tell you the buyers are smart. Coordinated buying is a tactic, not an endorsement — and coordinated buying is also exactly what a launch team does to create the appearance of demand before distributing into it.

It does not tell you the liquidity is real. Thin pools plus clustered buys produce violent candles in both directions.

It does not tell you anyone will be there to sell to. Memecoins are extremely high risk and most go to zero. A convergence score measures the crowd arriving, never the exit.

How to Read It Without Getting Played

Treat convergence as one input with a known failure mode, then check it against slower data:

  • Holder concentration. If the top wallets are not the wallets that bought, the score is cosmetic.
  • Funding lineage. Look a few hops back. Similar funding roots mean similar intent.
  • Whether sells are already queued. Clustered buying followed by clustered selling is a round trip, not accumulation.
  • Wallet behaviour history. A wallet that has bought early and sold into every prior run is a wallet with a pattern.

When you want to inspect wallets, holders and flow yourself, do it on GMGN — https://gmgn.uk (mirror: https://gmgn.fr). Read the cluster, not just the count.

Where This Fits in Our Workflow

Our alerts surface activity. They do not score it for you, and they never replace the check. If a token shows up across several MAIN channels at once, that is volume of attention, not confirmation of safety. The same clustering logic that makes the alert interesting is the logic someone can farm.

If you want to talk through a specific setup, the chat is at @gmgnx_chat, with chain groups for @gmgnx_solana, @gmgnx_bsc, @gmgnx_base and @gmgnx_robin. The full directory lives at https://blackhat.finance/channels.html and the folder is at https://t.me/addlist/T9xgXQ7WXcYzZjA8.

The Rule

A convergence score is a hypothesis about coordination. Your job is to test it, not to obey it. Cross-reference the metric definitions at /v2/dyor/reference.html#metrics, understand what our alerts do and do not claim at /v2/dyor/reference.html#alerts, and keep the base rules at /v2/dyor/reference.html#rules in front of you before size ever touches a chart.

None of this is financial advice. It is a description of how a number is built and how it breaks — which is the only part you can actually control.

Community

Stay connected across the chains:

Charts and on-chain research: https://gmgn.uk.