DEV Community: Larkmile road notes

Pricing rental car excess as a tiny data problem (with real Tasmanian numbers)

Larkmile road notes — Sat, 14 Nov 2026 16:29:52 +0000

Rental car excess is one of those decisions people make tired, at a counter, after a flight. It's also a very small data problem: a daily price, the excess you still owe, and the number of days. You can reason about it in ten lines of Python.

The hard part isn't the maths. It's getting numbers you can trust. Most hire companies in Tasmania don't publish their counter cover prices at all; they quote them when you book. One that does is Bargain Car Rentals, so I'll use its published terms as the worked example.

The inputs

For a standard car, Bargain publishes:

Option	Price	Excess you still owe
Standard excess	included	A$5,500
Gold Package	A$34.99 a day	A$2,500
Platinum Package	A$44.99 a day	A$0

For context, the standard excess at other Hobart and Launceston companies that publish it sits in the same band: A$5,200 at Redspot, A$5,400 at YesDrive, A$5,500 at East Coast Car Rentals (A$6,600 for its larger vehicles).

The model

Two numbers per option: what you pay for the cover, and your worst case if the car comes back damaged (cover plus remaining excess).

from dataclasses import dataclass

@dataclass
class Option:
    name: str
    per_day: float   # A$ a day, 0 if you keep the standard excess
    excess: float    # A$ you still owe after damage

def trip_cost(opt: Option, days: int) -> tuple[float, float]:
    """(what you pay for cover, worst case if the car comes back damaged)"""
    cover = round(opt.per_day * days, 2)
    return cover, round(cover + opt.excess, 2)

bargain = [
    Option("Standard excess", 0, 5500),
    Option("Gold Package", 34.99, 2500),
    Option("Platinum Package", 44.99, 0),
]

for days in (5, 10):
    for o in bargain:
        cover, worst = trip_cost(o, days)
        print(f"{days:>2} days  {o.name:<17} cover A${cover:>8,.2f}  worst case A${worst:>9,.2f}")

Output:

 5 days  Standard excess   cover A$    0.00  worst case A$ 5,500.00
 5 days  Gold Package      cover A$  174.95  worst case A$ 2,674.95
 5 days  Platinum Package  cover A$  224.95  worst case A$   224.95
10 days  Standard excess   cover A$    0.00  worst case A$ 5,500.00
10 days  Gold Package      cover A$  349.90  worst case A$ 2,849.90
10 days  Platinum Package  cover A$  449.90  worst case A$   449.90

The interesting line isn't the cheapest one. It's the gap between the two packages: A$10 a day buys A$2,500 less exposure. Over ten days that's A$100 for A$2,500. Whether that's worth it is a personal call, but now it's a visible one.

The field the model is missing

Excess is only half the story. The other half is what the top cover still leaves out, and that doesn't fit in a float.

Reading the terms of 11 companies with a Hobart or Launceston depot: 10 of 11 still charge you for roof and underbody damage even with their top cover. Bargain's Platinum Package is the exception. Tyres and windscreens are sold as separate daily add-ons at some companies (Redspot lists a tyre waiver at A$6.21 a day and a windscreen waiver at A$17.60 a day).

Then there are road rules, which matter a lot in Tasmania:

East Coast Car Rentals, Hertz and Thrifty don't allow unsealed roads.
Bargain allows graded, well-maintained gravel, never the Western Explorer Road.
Redspot allows gravel but raises your liability on it.

So the honest data model is closer to this:

@dataclass
class Option:
    name: str
    per_day: float
    excess: float
    still_yours: list[str]      # e.g. ["roof", "underbody"]
    gravel: str                 # "no", "graded only", "allowed, higher excess"

The first two fields you can optimise. The last two you have to read.

Separate excess policies

The other branch is a standalone policy instead of counter cover. Two things change the maths:

You pay the hire company first, up to the excess, and claim it back later. Five of the companies above say so in their terms.
The bond still goes on your card. A separate policy doesn't remove it.

Also check the policy limit against the excess, and whether it covers roof and underbody. Some travel insurance and credit cards already include hire car excess cover, as the ACCC points out, so you may not need to buy anything.

Where the numbers come from

All figures above are from each company's own terms, read on 10 November 2026. The full table for 15 companies, with a calculator that does the comparison for your dates, is on Larkmile.

Factual information, not financial advice. Read the rental terms and the policy's PDS before you decide.

Which Hobart Airport car hire desks are open when you land? Modelling published hours in Python

Larkmile road notes — Wed, 11 Nov 2026 15:44:37 +0000

If your flight lands at Hobart Airport at 22:45 on a Saturday, how many car hire desks are still open? Comparison sites usually list each company's hours one row at a time, which doesn't answer that. So we turned a published table of depot hours into a small Python model you can query by day and minute.

The input is the depot table in Larkmile's Hobart Airport car hire comparison. It covers 14 companies with a Hobart Airport depot on their own sites. Nine are at the airport and five are a short shuttle ride away in Cambridge. The company pages were read on 25 October 2026, and the airport and shuttle details on 28 October 2026. Hours change, so treat everything below as a snapshot of published hours, not a promise that a desk will be staffed.

Step 1: decide what you can't model

Before writing any code, go through the rows and flag the ones that don't turn into clean intervals. Three of the 14 don't:

Drive Car Hire publishes no hours. It "says it services all flights".
Astra Car Rentals: hours "not published".
YesDrive: "its page says 7am to 11pm, seven days, but its footer shows 7:00-18:00".

The honest choice is to leave these three out of the counts and say so, rather than guess. That leaves 11 companies with hours you can encode.

Two more need a decision:

Alice Car Rental has seasonal hours: 08:00-20:00 from December to April and 08:00-18:00 from May to November. We model the May to November hours because that's the season the data was read in. Swap them for a summer trip.
Hertz lists Sunday to Friday as 07:00-24:00. We treat 24:00 as end of day (minute 1440).

Step 2: encode the hours

Most companies publish one set of hours for Sunday to Friday and another for Saturday, so the data structure follows that pattern:

SUN_FRI = [6, 0, 1, 2, 3, 4]   # 0 = Monday ... 6 = Sunday
SAT = [5]
ALL = list(range(7))

def h(s):
    hh, mm = map(int, s.split(":"))
    return 24 * 60 if hh == 24 else hh * 60 + mm

desks = {
    "Avis":       ("terminal",  [(SUN_FRI, "07:30", "21:45"), (SAT, "07:30", "20:30")]),
    "Budget":     ("terminal",  [(ALL, "04:00", "23:30")]),
    "Europcar":   ("terminal",  [(SUN_FRI, "07:00", "23:00"), (SAT, "07:00", "21:30")]),
    "Redspot":    ("terminal",  [(ALL, "07:30", "22:00")]),
    "SIXT":       ("terminal",  [(SUN_FRI, "07:30", "22:30"), (SAT, "07:30", "19:00")]),
    "Bargain":    ("forecourt", [(ALL, "07:00", "21:00")]),
    "Hertz":      ("airport",   [(SUN_FRI, "07:00", "24:00"), (SAT, "07:00", "22:00")]),
    "Thrifty":    ("airport",   [(SUN_FRI, "07:00", "23:30"), (SAT, "07:00", "21:00")]),
    "Alice":      ("shuttle",   [(ALL, "08:00", "18:00")]),   # May to November
    "Apex":       ("shuttle",   [(ALL, "08:00", "17:29")]),
    "East Coast": ("shuttle",   [(SUN_FRI, "07:00", "20:45"), (SAT, "07:00", "17:30")]),
    "Drive Car Hire": ("precinct", None),   # no hours published
    "Astra":          ("shuttle",  None),   # no hours published
    "YesDrive":       ("shuttle",  None),   # two conflicting sets of hours
}

def open_at(day, minute, where=None):
    names = []
    for name, (loc, rules) in desks.items():
        if not rules or (where and loc != where):
            continue
        if any(day in days and h(a) <= minute < h(b) for days, a, b in rules):
            names.append(name)
    return names

The where field keeps the depot type from the source table. "Terminal" means a desk or kiosk inside the terminal (Avis, Budget, Europcar, Redspot, SIXT). Bargain is on the terminal forecourt. Hertz and Thrifty list an airport depot on Addison Drive. The shuttle depots are in Cambridge.

Step 3: sweep the day

for minute in range(0, 24 * 60, 30):
    names = open_at(2, minute)   # a Wednesday
    print(f"{minute // 60:02d}:{minute % 60:02d}", len(names))

Run it for a weekday and for Saturday, and the shape is clear. Counts cover the 11 modelled companies:

Time	Weekday: open	Saturday: open
04:00-06:59	1 (Budget)	1 (Budget)
07:00	6	6
08:00-17:00	11	11
19:30	9	7
21:00	7	4
21:30	7	3
22:00	5	1 (Budget)
22:30	4	1 (Budget)
23:00	3 (Budget, Hertz, Thrifty)	1 (Budget)
23:30-23:59	1 (Hertz)	0

Sunday runs on the Sunday to Friday hours for every company in the model, so it matches the weekday column.

What the sweep shows:

Before 07:00, there's one option. Budget opens at 04:00 every day, and nobody else in the model opens before 07:00.
Saturday night is much thinner than a weekday night. At 22:00 there are five desks open on a weekday and one on a Saturday. Hertz (22:00), SIXT (19:00), Europcar (21:30) and Thrifty (21:00) all close earlier on Saturdays.
After 23:30, it's Hertz or nothing. On Sunday to Friday, Hertz's listed hours run to midnight. On Saturday, no modelled desk is open after Budget closes at 23:30.

Step 4: filter by where the desk is

If you don't want to wait for a shuttle late at night, count only the desks inside the terminal:

open_at(2, h("23:00"), where="terminal")   # ['Budget']
open_at(5, h("21:30"), where="terminal")   # ['Budget', 'Redspot']

At 23:00 on a weekday, Budget is the only desk open inside the terminal. Hertz and Thrifty are open too, but they're listed as an airport depot rather than a terminal desk. At 07:00, only Budget and Europcar are open inside the terminal. Avis, Redspot and SIXT open at 07:30.

What the model leaves out

Pick-up hours are only half of the problem. The comparison also lists return rules that don't fit an interval model:

Avis and Redspot have key drop boxes at their terminal desks. The Avis one works while the airport is open.
Budget and Apex Car Rentals don't take after-hours returns.

So the company with the earliest opening (Budget, 04:00) won't take a return after hours. For a very early flight out, that matters more than the pick-up sweep. A fuller model would store pick-up and return windows separately.

A few other limits:

Published hours, not staffing. The model counts what each company's page says. It doesn't know about late-flight arrangements beyond what's published.
Three companies aren't counted. Drive Car Hire says it services all flights, so the real late-night count could be higher than the table shows.
Snapshot dates. Company pages were read on 25 October 2026. Rerun the sweep after updating the hours.

Extending it

The same structure works for any airport where you can collect published hours: one tuple per day pattern, a where tag, and None for rows you won't guess. Useful next steps:

add a returns field (drop box or none) and sweep it separately
give seasonal companies such as Alice Car Rental a date range per rule
feed in your actual flight time and print the list, not just the count

The point of modelling it this way is the None rows. A count you can trust beats a longer list that quietly assumes hours nobody published.

Larkmile road notes