What is coming on Europe’s freight roads, and how likely it is
Twice a day the radar reads the news and the trade press in four languages, the official ban and dosing calendars, the weather at the Alpine passes and the Strait of Messina ferries. Every item gets a chance, with the reasons behind it.
Loading the latest radar…
Where disruptions can hit, and how likely they are
What the holidays already hold
Every event priced on a real trip, with the plan that saves the most
Outside the calendars: strikes, closures, weather
In the official calendars: holiday bans, Kufstein dosing
By date, with the chance it happens
How the chance and the cost are calculated
A starting probability from how far things have gone, times the evidence
odds = prior ÷ (1 − prior)
odds × m1 × m2 × … for each piece of evidence
chance = odds ÷ (1 + odds), at most 98% until it is official
A national association proclaims a truck stoppage for 14–18 December (prior 70%, odds 2.33). Dates are named (×2), four outlets report it (×1.5), the association is national (×1.5), a meeting at the ministry is scheduled (×0.6).
odds = 2.33 × 2 × 1.5 × 1.5 × 0.6 = 6.3, so chance 86%. If the stoppage is then called off, the prior drops to 4% and the item falls off the top of the list.
z = −3.2 + 0.32 × snow in cm + 0.045 × (gusts above 50 km/h); chance = 1 ÷ (1 + e−z). 5 cm of snow gives 16%, 10 cm gives 50%, 20 cm gives 96%. Forecasts lose skill with time, so a day 12 days ahead counts for a quarter and the rest is the season’s base rate. Messina ferry: wind only, 70 km/h gives 27%, 90 km/h gives 73%.
| Stage reached | Prior |
|---|
| Evidence | Odds × |
|---|
Laws in force count as 99% (97% while Italy’s 2027 decree is not published); the official Tyrol dosing calendar as 95%. These numbers are starting assumptions: every item is checked after its date, and the scale is corrected until 30% items happen about 30% of the time.
Chance × what it costs, against what it costs to act
L = what the event costs a truck if the plan stays as it is
C = what the best alternative costs if it happens; C₀ if it does not
keep the plan: p × L · change now: p × C + (1 − p) × C₀
decide on the evening alert: p × C
change now only if p > C₀ ÷ (L − C + C₀)
This is the cost–loss model of decision theory, used to decide whether to act on a forecast (Thompson 1962; Murphy 1977), together with the value of information (Howard 1966): knowing the evening before whether the event happens lets you pay for the alternative only when it is needed.
For each dated event the radar builds one real trip through the place, timed to arrive just after it starts, and runs zhol’s engine twice, without and with the event: EU driving-time rules, tolls by country, €100 per truck-hour of waiting (the Italian legal rate, art. 6-bis D.Lgs. 286/2005), €1.18 per km, late delivery at €60 an hour (€150 for fresh food), waiting at your own depot at 40% of the road rate, €40 to change a plan.
A truck drives at most about 10 hours a day, so a long wait counts at most 10 lost hours per 24. Weather: about 3 hours of chains and queues on a snowy day. Night closures without published hours: 22:00–06:00. All of these are starting assumptions, written in the code and corrected with pilot data.
Instagram, Facebook and TikTok cannot be read automatically without official API access, and X charges for its API. Public Telegram channels can be added once we know which trucker channels matter. When no AI key is configured, a keyword reader in four languages replaces Claude and is less precise.