Commercial
What the market is doing.
Every scheduled flight carries a load factor that changes day to day, because passenger demand does. This page shows what drives it.
Last 90 days simulated to 20 Sep 2026
- Achieved, all history 84.96% distance weighted
- Long-run target 83.67% what it is calibrated to hold
- Difference +1.29 pts across 263 days
Network load factor
Distance weighted, every scheduled flight that day.
96.74% high · 80.02% low
90 days, averaging 90.09% against a 83.67% long-run target, marked · 5,521,155 seats flown
Drivers
What moved demand on 20 Sep 2026.
Every term that goes into demand, as the percentage it adds or takes away, averaged across all 66 stations.
Terms in force
Holidays show the departure side only. Local and Trend are each city's own drift.
Common market factors
One European move lands differently on Vienna, Munich and Paris: each city has its own sensitivity.
Stations
Demand by city.
Each against its own baseline, so zero is normal for that city rather than for the network.
By region
Seat weighted.
Furthest from normal
Most out of step today, either way.
All 66 stations
Ordered by how far demand has moved. Click any station to open it.
Markets
More frequency, thinner flights.
Ten times the seats is not ten times the market, so the busiest pairs carry the thinnest flights.
139 directed markets
One dot per market. Bigger is more flights, darker is more connecting traffic.
Seats a day, log scale
Fitted: -3.7 points per tenfold increase in seats.
Calendar
9 holiday windows in the next six weeks.
Each window lifts departures and arrivals separately. Bar length is the window, height is how hard it pushes, colour is which way.
departures lead arrivals lead
Flights
297 flights priced for 20 Sep 2026.
Split by cabin size and departure time. The 777-300ER flies in three cabins, so a flight number is not a fixed number of seats.
Load factor, spread across the network
Five-point bands. Average 90.6%.
100% high · 72% low
Last engine pass: tick, ok, 276 profiles rewritten at 2026-09-20 00:07:37Z .
Fullest today
BR EVA Air · B7 Uni Air
Method
How the number is made.
One index per city
Each station gets a demand score built from several layers: a long-term trend, the yearly season, shared market conditions, the city's own quirks, and small day-to-day noise. Real seasonal events, cherry blossom in spring, Hokkaido's ski season in winter, are tied to their actual dates rather than repeated automatically, so a peak in spring does not force a matching one in autumn. Holidays are added on top, counted separately for departures and arrivals.
Cities are not independent
No city moves on its own. One global condition affects the whole network, and six regional conditions each affect the cities in that region, by their own fixed amount. So when the European market shifts, Vienna, Munich, Paris and Milan all move with it, just not by the same amount.
Markets, not cities, carry demand
Demand on a route blends how busy the departure city is with how busy the arrival city is, weighted toward the departure side. Adding more seats to a route grows its market, but not in proportion, so heavily served routes run a bit thinner per flight. Routes used mostly for visiting family or work travel mostly skip the tourist season, since that kind of trip does not follow the calendar the same way.
Then it lands on an aeroplane
Each day's demand for a route is split across the flights that operate it, weighted by aircraft size and departure time, with a bit of randomness so two flights on the same morning are never identical. Cabin layout varies too: the 777-300ER fleet flies three different configurations, so a flight number is not always the same number of seats. Demand beyond one flight's capacity spills onto another. The result is what vAMSYS shows as that flight's load factor.
Reproducible
Every random number in the simulation comes from one starting seed, so nothing depends on what order it was calculated in. Any single day can be recalculated on its own and comes out identical every time, and replaying the whole history reproduces it exactly.
Calibrated, and checkable
One constant is solved so the network holds its long-run target load factor of 83.67%. What it actually achieved, and the gap, are at the top of this page. Nothing else is tuned to fit.
A simulation, not a record of flights flown. What was actually flown is on the statistics page. Times UTC. Read 2026-09-27 21:14Z.