Learn·Markets
Why did this subnet's price move?
Trading versus emission, decomposed exactly rather than fitted — with the leftover always published.
Published 2026-09-01 · updated 2026-09-28
Exact, not a model
Because the price is the pool ratio, the move can be split algebraically into the τ side and the α side. This is arithmetic, not a regression — there is no fitting and no assumed model.
The parts are computed in log space so that they add to the whole exactly. Adding percentage changes would not.
The residual is always shown
Whatever the named causes do not account for is published as a residual rather than quietly absorbed into the biggest component.
When the named causes explain less than 90% of the move, the answer is flagged as partial. A partial explanation presented as a complete one is worse than no explanation.
Why there is no history before we started
The public lite node prunes old state within minutes — a read a few hundred blocks back already fails there. Attribution therefore runs off snapshots Taosis takes itself and its own index of fills. An archive node keeps every block and could answer for older windows, but Taosis has not back-filled these snapshots: the history starts at the first one.
Until Taosis holds a snapshot of the subnet's pool at least as old as the window you ask about — a subnet it has only just started sampling, or a window longer than its history — the answer is unavailable, and it says so. A deploy does not reset this: the snapshots are kept in Taosis's database.