1. Revenue nowcast
Revenue point estimates per company per quarter, with attribution by region, cabin and route type, plus a running track record of forecast accuracy.
Data inputs
- Schedules;
- FlightAware data;
- Altus fares;
- bookings (?);
- card data;
- TSA and airport passenger counts;
- cabin seats map;
- competitor capacity;
- historical load factors.
How
Model passenger revenue bottom-up at carrier × route × month (or carrier × region × cabin × month), built as operated seats × load factor × yield. Aggregate to company level. Calibrate against reported revenue each quarter and track forecast error vs consensus and vs prior forecasts.
2. Capacity execution & grounding monitor
Weekly capacity-gap report (“Airline X scheduled +7%, operated +3%, gap concentrated in GTF narrowbodies”), plus grounded-fleet exposure by airline, engine family and age.
Data inputs
- schedules;
- FlightAware at tail level;
- aircraft type and engine family;
- operator;
- aircraft age;
- last-flight date;
- historical schedule-to-operated ratios;
- disruption events (weather, strikes, ATC).
How
Track scheduled vs operated capacity (seats, ASKs, frequencies) by carrier, route, region and engine family. At tail level, classify each aircraft as active/ temporarily inactive / grounded based on days since last flight vs baseline. Aggregate to operator, engine family and aircraft age cohort.
3. Booking curve & fare dynamics
Forward read on whether headline fares will hold. Catches late discounting before it shows up in reported yield.
Data inputs
- Altus fares at fixed booking windows (26 WBD → 0WBD);
- bookings;
- historical fare curves by route and season;
- competitor fares;
- holiday and event calendar;
- cabin map.
Methodology
For each route and departure date, compare current fare curves against same-route, same-season, same-window historical curves. Classify the shape: early strength + late flattening; inventory tightening; late discounting; premium tightening; etc.
Output
Per-route and per-carrier curve-shape signal, flagged as confirming or contradicting headline fare-scrape strength.
4. Surprise, revision & ranking stack
Per company, three views — “vs consensus” estimate and bridge, revision probability and size, and cross-sectional rank within the universe.
Data inputs
- Outputs of Projects 1, 2 (and 4 once available);
- consensus estimates;
- guidance history;
- fuel and FX;
- stock move into prints;
- short interest;
- valuation;
- balance sheet;
- momentum.
How
One pipeline, three layers:
-
Layer A decomposes the forecast vs consensus into revenue / fuel / labour / FX / mix surprises and converts to P(beat) and P(guidance change).
-
Layer B predicts size and probability of consensus revisions over the next 2 / 4 / 8 weeks.
-
Layer C produces a weekly cross-sectional ranking on residual returns, stripping out market, sector, oil and FX betas.
5. Macro scenario & route economics engine
Fast way to translate a macro view (oil, FX, demand) into sized stock-level expressions and pairs.
- Ranked carrier exposures to a given shock;
- EBIT/EPS sensitivity;
- suggested pair expressions
Data inputs
- Jet fuel curve / Brent / crack spreads;
- FX;
- route distance;
- aircraft fuel burn;
- fares and yield;
- load factors;
- hedging disclosures;
- historical oil-shock and event windows.
Methodology
Build a simplified route contribution model (revenue minus fuel, crew, airport, maintenance, ownership). Plug into a scenario engine for shock fuel, yield, capacity or FX. Combine with historical event-window residual returns for base-rate priors. Use the output to generate macro-hedged long/short pair candidates.
6. Signal-agreement view
Makes conviction explicit. Tell if a thesis is high-conviction across signals vs when it is fragile and signal-dependent.
Dashboard per company showing signal-by-signal direction, confidence and historical reliability, with conflict patterns flagged (“fares strong + bookings weak = discounting risk”; “capacity down + fares up = supply-led yield support”).
Data inputs
Outputs of all preceding work-streams.
How
For each company-quarter, build a matrix of (signal, direction, confidence) showing which sources agree and which disagree. No single aggregate score, display the disagreement explicitly.
7. Utilisation (almost done)
Turns raw flight data into direct fundamentals for civil aero names (engine OEMs, MRO providers) and a maintenance cost view for airlines.
Read-through to engine OEM flying-hour revenue, shop-visit demand, spare-parts demand, and airline maintenance cost exposure, by engine family.
Data inputs
- FlightAware;
- Cirium for aircraft-to-engine mapping;
- Block hours;
- aircraft age;
- operator;
- region.
Methodology
Tail-level utilisation metrics (EFH, cycles, stage length, days active) mapped to the OEM. Compare against baselines to identify engine families with structural utilisation strength or weakness.