Thursday, October 8 · typical odds for a Thursday (too early for a weather forecast) · issued Sun 12:00 PM ET · 6 nonstops

Thu, Oct 8 looks like a normal day here: all 3 nonstops are 15–19% likely to arrive 15+ min late.

Safest bet that day OH 5567 PHL 09:49 → CLE 11:32 · 15% chance 15+ min late Riskiest AA 5581 PHL 15:13 → CLE 16:44 · 19% chance 15+ min late · usually 73% on time

Every nonstop on Thu, Oct 8

12-month record →

Ranked by the chance of arriving 15+ minutes late (typical odds: the weather this date has had in past years), with each flight’s on-time record over Aug 2025 – Jul 2026 beside it. Times are local. Pick a flight for its forecast. Click a column to sort. Dimmed: airlines we don’t forecast yet, shown with their 12-month record if they usually fly that weekday.


OH 5567 PSA	PHL 09:49	CLE 11:32	15%	4%	2.1%	no record
OH 5287 PSA	PHL 16:50	CLE 18:26	18%	7%	2.0%	no record
AA 5581 American flown by PSA Flight history →	PHL 15:13	CLE 16:44	19%	6%	1.6%	73%
AA 5847 American flown by Piedmont no forecast: 12-month record	PHL 08:37	CLE 10:27	14%	5%	1.5%	86%
AA 5862 American flown by Piedmont no forecast: 12-month record	PHL 17:15	CLE 19:06	19%	8%	7.1%	81%
AA 5861 American flown by Piedmont no forecast: 12-month record	PHL 12:50	CLE 14:38	20%	4%	2.8%	80%

How these are worked out. Each flight’s odds are our forecast for that day (typical odds until the weather forecast covers it); flights of airlines we don’t forecast yet use their 12-month record and are marked. Makes the connection: the first flight’s 12-month lateness curve, scaled so its chance of arriving 15+ min late matches the day’s forecast, at the layover minus the 45-minute minimum (90 on separate tickets). Arrives on time: first flight not cancelled × connection made × second flight not cancelled × second flight within 15 minutes, each from the day’s forecast.

About this data: forecast odds from our model for Thu, Oct 8 (newest published update); 12-month records from US DOT on-time data, Aug 2025 – Jul 2026, all US airlines; regional flights count under the airline that sells them. JSON