Spread compression is not destroying market making. It is widening the moat.

Every analysis of HFT profitability focuses on per-trade economics. That is the wrong variable. XTX eliminated last-look entirely in 2017 — deliberately accepting worse per-trade economics — and grew its direct client base from 20 to 100 counterparties in 12 months. Bilateral volume rose 50%.

A firm with near-zero incremental cost per trade does not lose when the spread compresses. It answers compression with volume scale. The firms that cannot do this are squeezed out. The ones that can compound the barrier annually.

The compression narrative is the consensus. The volume-scale mechanism is what the on-record interviews and SEC filings actually show.

The fair-value model accuracy lever — the only one of the three that compounds on proprietary live flow data — and why a 2026 entrant structurally cannot replicate it is in the full piece.