Identifying an edge is the easier part. Capturing it is where value is created, and where most failures happen.

That is the through-line of my talk at the Lufthansa Group Data Community Day 2026. A predictive model usually provides only a small statistical edge. The hard part is converting it into value and defending it before it decays.

The deck works through real cases across quant finance, horse racing, credit scoring, real estate, and Big Tech’s ad business: edges that were won, lost, copied, regulated away, or never captured at all.