How many wallets should a bundle use?

Dispersion, mostly. A treasury spread across sixteen addresses is not one compromised key away from being gone, and a position taken in sixteen pieces prices marginally better than the same position in one.

Every wallet is a separate transaction with its own gas, a separate funding transfer, and a separate sweep on the way out. Gas here is cheap enough that this rarely decides anything, but it is not zero.

The real cost is time. Sixteen buys take longer to send than five, and in a three-second window that is the constraint that binds. Signing is not the bottleneck — ten transactions sign in about nineteen milliseconds — but sending them is.

When you sweep, a wallet holding less than the transfer costs cannot be emptied. The remainder is stranded — not lost, just uneconomic to move. Twenty wallets means twenty small residues rather than five.

It is a few cents either way at these gas prices. Worth knowing, not worth optimising around.

Ten to sixteen covers most cases: enough dispersion to matter, few enough that the volley stays tight and the sweep stays tidy. Go to 32 when the position is large enough that curve slippage on a smaller split would cost more than the extra overhead.

Whatever you pick, name all of them on the exemption list — including the ones you do not intend to use this time. That list cannot be extended later.