I Tracked a $71,000 Sedan's Value Every Week for a Year — It Lost $19,400 by Month 12

Every Monday at 8 p.m., after the kids were down, I opened three valuation sites and a dealer trade-in calculator, logged the numbers in a spreadsheet, and watched $71,000 of German sedan turn into less. Fifty-two Mondays. The spreadsheet still opens on my laptop with the first cell reading "Week 1 — $71,000 (purchase, Apr 3)" and the last reading "Week 52 — $51,600." That is $19,400 gone in a year, and I have the graph to prove it does not fall the way anyone says it does.

the rules of my experiment

Why do this to myself? Cuz in April 2025 I snagged a three-year-old 540i with 24,000 miles — deliberately not new, deliberately past the famous first-year cliff — and I kept reading articles that described depreciation like a smooth staircase. Ten percent the first year, fifteen the next, neat and predictable. My suspicion was that the real curve was lumpy, event-driven, and occasionally cruel, and the only way to know was to measure my own car weekly, like a detrimental hobby.

the cliff nobody schedules

The first cliff arrived in week 6. Nothing happened to my car. But the manufacturer launched the next model year with a facelift and a $2,000 price cut on the outgoing trim, and my sedan's book value dropped $2,800 in seven days. Week 6 to week 7: $63,900 to $61,100. I refreshed the page four times thinking it was a glitch. It was not a glitch; it was a factory in Munich making my car one generation older by press release.

The second cliff, week 19 to 22, was gentler but odd: $3,100 total, driven by a wave of lease returns hitting auction lanes — end of a generous 2022 lease cycle. Supply is destiny. My Tuesday-morning valuations moved when thousands of identical sedans rolled off transport trucks in Newark, not when I drove anywhere.

the plateaus that fooled me

Between the cliffs, the curve did something the buying guides never mention: it went flat. Weeks 27 through 38, twelve straight Mondays, my car's wholesale value moved less than $150 total. I began to feel smug. I recalibrated my mental model toward "the worst is behind us." Then weeks 39 to 44 took another $2,600, mostly mileage-band math — I crossed from the 30k band into the pricing category where appraisers assume fleet use. The car did not change. The category did. Depreciation is not a slope; it is a staircase where someone else decides when the next step arrives.

fifty-two Mondays, tallied

The full year, tallied: $19,400 lost, of which approximately $6,900 came in five specific weeks and the remaining $12,500 bled steadily. Average of $373 a week. My total driving that year: 11,800 miles. Which means the depreciation alone cost about $1.64 per mile before gas, insurance, or a single brake pad — and that is the number I want people to sit with when they finance nearly-new luxury metal on a six-year note.

the financing layer underneath

Here is where the experiment turned from morbid to useful. I had financed $58,000 of the purchase over 60 months at 6.8%, so I owed far more than the car was worth through most of the year — negative equity peaking around $9,000 in month seven, by my spreadsheet's overlap columns. The gap only closed in month eleven, when the balance ultimately fell below the wholesale value. That means for over half that loan, a single total loss event would have left me writing a check to the bank. I priced GAP coverage that week: $340 flat through my insurer. If I had known the exact shape of this curve, I would have snagged it on day one instead of day 220.

january, the refinance window

The year also pushed me into a smart move I nearly skipped. By January my credit score had crossed 760 on the back of paying off some old card debt, and my credit union offered an auto loan refinance at 5.4%. On a $52,000 balance that was $68 a month, approximately $2,300 across the remaining term — nearly 12% of the depreciation pain, recovered with one 40-minute appointment.

the punchline about buying one-year-old cars

The conventional advice says "buy two to three years old and let someone else eat the cliff." My data mostly supports it, with one amendment: the cliff is not calendar-based, it is event-based, and you can see the events coming. Model year announcements, lease-return waves, mileage bands — all of em are scheduled or observable. What the data killed in me is the fantasy that a car is an asset with a gentle glide path. Mine lost 27% of its value in a year and, per mile, cost more than the lease payments on the rental I once despised as wasteful.

week 53

I still log it. Force of habit. This Monday the number read $51,100 — down $500 for a cracked wheel I have not fixed, which is its own slight lesson: condition repairs are worth doing before the appraiser notices, not after. The graph on my laptop has cliffs, plateaus, and one embarrassing spike where a valuation site briefly glitched and told me I was rich. Fifty-three weeks in, the spreadsheet's most honest cell is a note I typed in week 7 and never deleted: "The car does not know what it is worth. The market decides on Tuesdays."