AI

Multibuy Convergence: How the Score Works, and How It Gets Faked

Multibuy convergence is a signal about who is buying at once — not proof they are real. Here is where the score comes from and where it lies.

· 6 min read · Blackhat Empire

What Multibuy Convergence Actually Measures

A "multibuy" alert is not one wallet buying a lot. It is several distinct wallets buying the same token inside a short window. Convergence is the scoring layer on top: it tries to answer how unusual that cluster is, not just that it happened.

Most scoring engines stitch together a few inputs:

  • Wallet count inside the window — five wallets in 60 seconds is a different event than five wallets across four hours.
  • Wallet age and history — fresh wallets with no prior trades score very differently from wallets with a real on-chain footprint.
  • Size distribution — one whale plus four dust buys is not the same shape as five similar-sized buys.
  • Prior co-occurrence — wallets that have repeatedly bought the same token together get flagged as a known cluster.
  • Liquidity and holder context — a $5k pool does not "confirm" anything the way a deeper pool does.

When those line up, convergence scores high. That is the whole product. It is a clustering heuristic, and like every heuristic it can be fed.

The Honest Limit of the Signal

Convergence describes coordination. It does not describe intent. A cluster can be smart money rotating early, or it can be one person with forty wallets. On-chain, those two look nearly identical in the first thirty seconds.

That is not a flaw in the tool. It is the nature of the data. Treat convergence as a question, not an answer.

Where It Gets Faked

1. Wallet fan-out from one operator

The oldest trick. One entity funds many wallets from a common source, then buys in sequence. Convergence goes green. What is missing is independent conviction — it is one decision wearing many hats.

What to check: funding trails. Wallets created in the same block, funded from the same address, or funded minutes before buying are not independent. Cross-reference on GMGN to see wallet creation and funding patterns before you trust the count.

2. The dust cluster

Twenty wallets each buying a tiny amount can trip a naive wallet-count threshold while moving almost no money. The score looks strong; the capital does not.

What to check: size distribution, not wallet count. If the largest buy is small relative to the pool, the "convergence" is decorative.

3. Timed buys against thin liquidity

On a shallow pool, a few coordinated buys produce a violent candle. That candle gets screenshotted, posted, and treated as confirmation. The buys did not predict the move — they were the move.

What to check: liquidity depth versus buy size. If the buys are large relative to the pool, you are looking at price impact, not discovery.

4. Recycled wallet sets

Some clusters are the same wallets rotating through token after token. Once you have seen a wallet set converge on ten launches that died, the eleventh is not a signal.

What to check: prior co-occurrence. Repeated co-buying is a pattern — and patterns get sold to you.

5. Screenshot laundering

A real convergence event gets cropped, stripped of context, and reposted hours later on a token that has already moved. The data was true. The framing is a lie.

What to check: timestamps. Always pull the live view on GMGN rather than trusting a forwarded image.

How to Read Convergence Without Getting Played

  • Sequence matters. Early convergence into an untraded pool is a different animal than convergence after a 40% candle.
  • Independence is the whole signal. One operator with many wallets is not convergence. It is a costume.
  • Structure beats count. Five real wallets with history outrank fifty fresh ones.
  • Context or nothing. A score without liquidity, holder count, and time-since-launch is noise.

For the underlying metric definitions we use, see the metrics reference. For how alerts are tiered and what each channel actually fires on, see alerts. And before you act on any cluster, re-read the rules.

Where This Fits in the Empire

Multibuy feeds are one input among many, not a verdict. If you want to watch convergence live, the relevant streams are @empiresolmultibuys, @empirebscmultibuys, and @empirerhmultibuys, alongside @gmgnxsolkolcluster for wallet-cluster context on Solana. The public groups — BH GMGN SOLANA (@gmgnx_solana), BH GMGN BSC (@gmgnx_bsc), BH GMGN BASE (@gmgnx_base), and BH GMGN ROBINHOOD (@gmgnx_robin) — are where people compare notes in real time.

The full channel list lives at blackhat.finance/channels.html, and the Telegram folder is here.

The Part Nobody Says Out Loud

Convergence scoring exists because humans are pattern-hungry and a green number feels like permission. It is not permission. It is a compressed summary of who bought, when, and how much — with all the ambiguity that implies.

Memecoins are extremely high risk. Most go to zero. A high convergence score does not reduce that base rate; it just tells you a specific kind of activity happened. Your job is to figure out whether that activity is independent conviction or a staged event — and to size accordingly when you cannot tell.

If you cannot explain why a cluster is buying, you do not have an edge. You have a screenshot.

Community

Stay connected across the chains:

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