On-chain research
We publish the cases we can document. Every transaction listed in these studies was re-fetched from the chain and checked by slot, timestamp and balance movement — not copied from a summary.
Each report states its own limits. Coordination is not ownership, and on-chain data alone cannot establish who controls a wallet.
Launch tricks we have caught →
Six ways a launch is made to look fair or safe, one verified case each.
Who bought before Cupsey: 9 wallets, 324 coins, and the wallet that funded his public wallet
Cupsey's public Solana wallet looks clean on its own. Following its money tells a different story. Its first SOL came from a wallet that 9 other wallets paid 3,191.8 SOL into. From August to November 2026 those 9 wallets bought 343 of the coins Cupsey bought before he did, a median of 4 minutes ahead, and on 324 of them they sold after his buy: into the people buying after him. Many of their tokens were sent to them by the same wallet that funded his public wallet. Every step below is a transaction you can open.
Read the evidence →
A $202M market cap on 1.6% of the supply: how a self-lock made a launch look safe
Locked supply usually reads as a safety sign. On $uni the creator bought almost the whole supply in the first transaction and locked 98.4% of it back to itself, so only 1.6% could trade. That thin slice carried a market cap of about $202 million at the top, and within 41 minutes the creator had taken back more SOL than it paid at launch. Here is how it worked, step by step.
Read the evidence →
How one wallet hid 59% of a launch behind 208 throwaway wallets
Every first buyer of $LAUNCH seemed to have its own funder, and our first check called it clean. Following the money up ten layers showed one wallet behind the creator and the four biggest first buyers. This is how the trick works, why it fooled us, and what we changed.
Read the evidence →
8 Robinhood Chain coins bundled at launch: how to spot it
On Robinhood Chain a bundle does not hide in a block. It hides in the funding: one transaction pays many fresh-looking wallets minutes or seconds before the coin exists, and those wallets buy most of the supply in the first seconds. In 6 of the 7 coins funded this way, the same transaction also paid the coin’s creator.
Read the evidence →
What is a bundle in crypto? Bundles, bundlers and bundled launches explained
A bundle is one party buying a coin’s launch through many wallets at once, so it looks like real demand. Here is what bundled, bundler and bundling mean, why people do it, and how to check any coin in a minute.
Read the evidence →
One wallet, the whole curve, one transaction: the KIMCHI launch
Bundle checkers look for many wallets buying together. KIMCHI had no crowd to find: one wallet bought everything that was for sale in the first transaction. That is why it looked "clean" to bundle checks, ours included, until we changed how we read launches.
Read the evidence →
13% in a place no buyer can reach: the Dacity launch bundle
Dacity looked clean on some tools and 7–59% bundled on others. Reading the creation block itself settles the launch question: 13.4% went to three wallets in positions only a bundle can reach.
Read the evidence →
Same tokens, to the last unit: the fingerprints of a Proxima launch
Bundling tools sell "organic looking" launches. They still leave marks: wallets handed identical shares to the last unit, and bots that sell on a fixed rule.
Read the evidence →
Why every site shows a different bundle % (and which one to trust)
Bundle numbers disagree because each site answers a different question. Know which one you are reading before you trust it.
Read the evidence →
Two seconds ahead: a Solana side wallet front-buying five launches
Five launches, five front-buys, one direct 20 SOL transfer. Here is the full evidence — transaction by transaction — and what it does and does not prove.
Read the evidence →
Smart wallets or swap routers? Reading contract flags on Robinhood Chain
A contract flag is not a trader-versus-router test. On Robinhood Chain, eight of the wallets we track all report as contracts — and all eight are ordinary accounts delegating to the same smart-account implementation.
Read the evidence →
Three wallets, 22 seconds: a funding trail that does not look independent
Four supposedly separate wallets, all of them funding and drawing from one wallet, mostly in round numbers — and three transfers landing inside 22 seconds. The transfers are verified on-chain; the conclusion is left to you.
Read the evidence →
Coordination or coincidence: why most of what we find is not evidence
Any clustering tool can draw you a graph. The useful question is how much of it survives scrutiny. So here is what our own detection produces — including the part we deliberately refuse to call evidence.
Read the evidence →
KOL wallets explained: side wallets, bundling and shilling on-chain
If you follow traders on X, here is what is happening on-chain underneath the timeline — with our real numbers, including the uncomfortable ones.
Read the evidence →
How to check if a crypto KOL has side wallets (step by step, free)
Anyone can do this check with a block explorer and patience. Here is the order that works, the evidence that counts, and the look-alikes that fool people.
Read the evidence →
Before you copy a wallet: six checks to avoid becoming exit liquidity
A wallet with a huge profit figure can be a bad one to copy. These checks take a few minutes and each one can change your mind.
Read the evidence →
How to check a FOMO trader’s wallet before you follow them
A FOMO profile is a starting point, not proof. Here is how to turn a handle into on-chain facts and what to ignore along the way.
Read the evidence →
What goes into a case study
- Every transaction cited has a signature and slot we re-read from the chain.
- Recorded trade timestamps are marked separately from chain-verified ones.
- Aggregates are labelled as aggregates — never presented as a single transaction.
- Labels in our dataset are attribution, never treated as verified identity.
- A limitations section is mandatory. If the evidence is thin, we say so.