Why Your Bundle Checker Shows Over 100% (And How We Fixed the Bug)

A bundle checker's whole job is to tell you what share of a token's supply went to early buyers. So when a number comes back over 100%, something is wrong — not with the token, with the tool. We hit this exact bug while building our own Bundle Checker, and fixing it taught us something worth writing down.

Where the number comes from

A bundle checker reconstructs a launch by reading the token's earliest transactions and tracking which wallets received tokens. Add up what those wallets received, divide by the token's initial supply, and you get a percentage. Simple, until one of those "wallets" isn't a buyer at all.

The account that isn't a buyer

On Pump.fun, unsold supply sits in a program-owned bonding-curve account until someone buys it. That account receives a huge share of supply the moment the token is created — often 70-90% of it. If a checker's logic treats "any account whose balance went up" as a buyer, it counts the curve itself as the single largest "early buyer" on every token. Add a few real buyers on top of that and the total sails past 100%.

Why excluding it isn't as simple as it sounds

The obvious fix is: find the curve account, exclude it. We built exactly that — and it worked, until we tested a token where a second program-owned account was also moving tokens around mid-launch, acting as an intermediate distributor. Our first-pass exclusion logic only knew about one such account, so this second one slipped through and got counted as a buyer again. Same bug, different shape.

The fix that actually held up: a real buyer's token balance only ever goes up during a buy transaction. Any account whose balance goes down during that same transaction is — by definition — distributing tokens, not buying them. That rule excludes every distributor account automatically, without needing to guess how many of them exist or hardcode any specific address.

What this means for reading any bundle checker

• A number near or over 100% almost always means curve inventory got miscounted, not that buyers somehow acquired more than existed.
• Ask, or check, whether a tool excludes program-owned accounts specifically — not just "the largest holder."
• Same-slot activity and supply percentage are two different signals; a bug in one doesn't necessarily affect the other.

The same underlying pattern — a program-owned account looking like an ordinary wallet — shows up in holder distribution data too, not just bundle detection. We cover that case in why your holder checker might be lying about whales.

You can see this in action on our own Bundle Checker — paste any Pump.fun mint and the wallet table shows exactly who's counted and why.