Volume-weighted average price at which ZEC entered each pool. When ZEC is shielded into Sapling at $50, that batch has a $50 cost basis. Outflows reduce the balance but keep the average.
The spot price line shows the current ZEC price for comparison — pools with an average below spot are "in profit" (holders acquired cheaper than current price). Transparent pool includes coinbase rewards (mined ZEC) and deshielding.
Ironwood has no per-pool flow columns, so its acquisition price is reconstructed from the per-transaction migration attribution (each inflow priced at its migration day).
When ZEC crosses into a pool it is priced at that day’s close, and each pool carries the volume-weighted average of everything that entered it. Shield 100 ZEC into Sapling at $50 and Sapling’s basis moves toward $50. A pool trading below spot is, on average, held above cost.
Outflows do not reprice. Value leaving reduces the balance and leaves the average where it was, because this data cannot tell WHICH coins left — a shielded pool does not expose that. So the basis is the average of what went in, not of what is still there, and the two diverge when outflows are selective.
Transparent is not a cohort. It holds mining rewards priced at the day they were mined alongside everything that has ever been deshielded, so its basis is a blend of two unrelated populations and the most easily over-read line here.
Ironwood is reconstructed, not read. It has no per-pool flow columns yet, so its basis is built from per-transaction migration attribution, each inflow priced at its migration day. That is a different provenance from the other three and worth knowing before comparing them closely. For the chain-wide number see realized price.