Hot Picks: algorithmic multi-city flight deals

What Hot Picks actually searches

A naive multi-city search explodes: every extra stop multiplies outbound flights, stay windows, and the return home. Hot Picks does not enumerate every combination. It runs a greedy recursive search over the flight graph, keeping only branches that still look cheap after a return-cost lookahead.

• Closed-loop trips. Every result starts and ends at your origin so you are not stranded in the last city.
• Return potential. Before committing to a next city, the engine checks the cheapest known fare home from that airport and drops expensive dead-ends early.
• Hub bias. Airports with dense budget-carrier graphs stay in the search longer because they unlock more second and third stops.

How to read a Hot Pick

The map view is the working tool. Use this page to start a search; submit origin and dates to open the Hot Picks map. Each card is a scored itinerary, not a packaged airline multi-city ticket. Legs are typically separate one-ways, which is how budget carriers price Europe.

Want the engineering detail — C++ core, Python orchestration, scoring weights? Read how the Hot Picks algorithm works.

When to use Combi-Search instead

Hot Picks is for inspiration: you pick an origin and a window, we propose cities. If you already know the cities and stay lengths, use Combi-Search so the planner respects your route order.