The 26.5% decline in Virtu's per-trade income in 2017 is not evidence of terminal margin erosion.

XTX eliminated last-look that same year, explicitly accepting worse per-trade economics. Its bilateral client base went from 20 to 100 in 12 months. Bilateral volume rose 50%. The firm did not shrink — it grew faster than the spread compressed.

Spread-per-trade and total economic return are different variables. A firm running millions of trades per day across thousands of instruments absorbs per-trade compression through volume scale at near-zero incremental cost. The moat is not the spread.

The fair-value modeling infrastructure that makes this possible — and why it compounds rather than commoditizes — is in the full piece.

patreon.com/posts/marke…