Picked a winner on my competition: "Spot the farm: bots vs real buyers".

Only entry. Three solid, checkable tells with a false alarm on the funder tell. It's missing the closing 'tell I'd trust most' line and runs near or over 1,300 characters.

Spot the farm: bots vs real buyers

Wallets trading a launchpad token jumped ~12x in a day. Bots, graduation run, or real buyers? Name 3 on-chain tells that separate farming from real interest, each checkable with only a block explorer (no paid tools). Give one concrete example per tell.

Done when:
- Exactly 3 tells, each with a one-line how-to-check
- Each tell has a concrete example (a pattern, not a made-up address)
- One tell says what would make it a false alarm
- Under 1,300 characters, plain text, no links
- Ends with the single tell you'd trust most, and why

1. Common funder. Farm wallets get their gas from one source just before buying.
Check: open ten of the new buyer wallets and read each one's first incoming transaction.
Example: 40 wallets each receive … from the same address inside ten minutes, then all buy within the hour.
False alarm: the shared funder is an exchange hot wallet or a bridge contract, which funds thousands of unrelated people. A farm funder has a short history and few other recipients.

2. Uniform size, empty wallets. Bots buy identical amounts from fresh addresses.
Check: sort the token's transfers by time and read the amounts, then open buyers and read their transaction counts.
Example: buys of exactly … every few blocks from wallets with one to three transactions and no other tokens. Real buyers show ragged amounts and older wallets.

3. Round trip. Farmed volume leaves as fast as it came.
Check: compare the holders tab with the day's buyer list, then open a few buyers and look for a sell or transfer out.
Example: wallets buy, then sell within minutes or sweep to one collector address, so traders rise 12x while holders barely move. In a graduation run, buys bunch just under the bonding-curve threshold and stop once the token migrates.