For most of the last century, predicting what a used car would be worth in three years was a solved problem. Not perfectly, but well enough. A diesel estate from a known brand depreciated along a curve that everyone in the trade had internalised. Financiers wrote that curve into residual value tables. Dealers priced part-exchanges against it. Leasing companies built their margins on top of it. The car was a known quantity, and so was its decline.

Electric vehicles have quietly broken that machinery. Not because EVs are bad assets, but because the assumptions underneath every residual value model were built for a different kind of car. The industry is now discovering, sometimes the hard way, that a used EV is a far harder thing to price than a used combustion car ever was. And the people most exposed to that uncertainty - financiers carrying residual risk and dealers holding stock - are the ones least served by the tools they have inherited.

This is not a temporary wobble that settles once the market matures. It is a structural shift in how predictable a vehicle's value can be. The honest response is not a better single number. It is a different way of thinking about value altogether.

Why combustion residuals were easy and EV residuals are not

A petrol or diesel car ages in ways the trade understands intuitively. Mileage, service history, body condition, brand, model cycle. The drivetrain is mature technology that changes slowly. A three-year-old engine is, mechanically, much like a five-year-old engine. Depreciation is gradual and legible.

EVs scramble almost every one of those inputs.

The battery is most of the value, and its condition is invisible. On a combustion car, the engine rarely fails in a way that destroys resale value within the typical ownership window. On an EV, the battery can represent a large share of the vehicle's worth, and its real condition is not written on the dashboard. State of health degrades with charging habits, climate, fast-charging frequency and age, and two cars with identical mileage can have meaningfully different remaining capacity. A buyer cannot see this. Neither, often, can the dealer taking it in.

Model cycles have compressed. Combustion model generations ran for years with incremental change. EV development moves at the pace of consumer electronics. Range, charging speed, software features and efficiency improve quickly, which means a two-year-old EV can feel a generation behind a new one in ways a two-year-old diesel never did. That perception flows straight into resale value.

Incentives move the floor under the whole market. Purchase subsidies, benefit-in-kind rules, road tax, charging infrastructure rollout, urban access policies - these vary by country and change with the political weather. A subsidy on new cars can suppress used values overnight by making new ownership cheaper than expected. A withdrawn subsidy can do the opposite. None of this is captured by a depreciation curve fitted to historical data.

The technology floor keeps rising. Battery chemistry, thermal management and charging standards keep improving. Every advance makes older vehicles look relatively worse, even when those older vehicles are perfectly good. This is depreciation driven by the frontier moving, not by the car wearing out.

Put together, these forces make used EV values more volatile, more dispersed and more sensitive to outside events than combustion values ever were. The variance is wider, and the things driving it are harder to observe.

Why traditional valuation models break

The standard residual value model is, at heart, a curve fitted to history. It assumes the future rhymes with the past, that the main inputs are mileage and age, and that a single point estimate is a reasonable summary of a car's worth.

For EVs, each of those assumptions is shaky.

History is thin and unrepresentative. The used EV market is young, and the cars that have already cycled through it were sold under incentive regimes and price expectations that may no longer hold. Fitting a curve to that history risks encoding conditions that have already changed.

Mileage and age are no longer sufficient inputs. Without battery state of health, you are pricing the least important part of the car precisely and the most important part by guesswork. Two cars can look identical on paper and differ sharply in real value.

And the single number actively hides the risk. When a model returns one figure, it implies a confidence that is not there. A residual value table that says a car will be worth a specific amount in three years, with no expression of how uncertain that is, gives a financier no way to size the risk they are actually taking. It is precision without accuracy, and in a volatile market that is worse than useless. It is misleading.

The result is a quiet mispricing risk running through the system. Residuals set too high leave financiers and leasing companies exposed when cars come back. Set too low, and dealers leave money on the table and consumers overpay. Either way, the error is invisible until the car is sold, by which point it is too late to do anything about it.

The honest alternative: data-rich, confidence-aware valuation

If the old model fails because it pretends to know more than it does, the fix is not a cleverer way to produce a single number. It is to value vehicles in a way that reflects what is actually known, and how strongly.

Three principles follow.

Value the battery, not just the badge

A credible EV valuation has to engage with battery state of health, charging history where available, and the specific chemistry and thermal design of the pack. A car with strong remaining capacity is a different asset from one with degraded capacity, even if everything else matches. Pricing that ignores this is pricing the wrong car. The data exists, increasingly, in the vehicle itself and in connected data streams. The constraint has been access and standardisation, not absence.

Show a range, and show your confidence

The most important shift is to stop pretending. A used EV should be valued with an explicit confidence interval, not a single point. "This car is worth somewhere in this range, and here is how sure we are" is both more honest and more useful than a lone figure. A financier can size residual exposure against the downside of the range. A dealer can decide how much room to leave in a part-exchange. Uncertainty, made visible, becomes something you can manage rather than something that ambushes you later.

Confidence should also move with the evidence. A common model with deep transaction history and good battery data deserves a tight interval. A scarce model in a market that just changed its incentive rules deserves a wide one. A valuation that always sounds equally certain is not measuring anything real.

Show the sources

When a number changes, the people relying on it need to know why. A valuation that can point to its inputs - comparable sales, battery condition, the model's market position, the incentive context - is one you can interrogate and trust. A valuation that arrives as an oracle is one you can only take on faith, and faith is exactly what the EV transition has made expensive.

What this means for dealers and financiers

For dealers, the practical lesson is that EV stock carries a different risk profile from combustion stock, and pricing it with combustion-era confidence is a way to get hurt. A car that looks like a bargain may be carrying battery degradation you cannot see. A car that looks overpriced may have a genuinely healthy pack. The dealers who do well will be the ones who can read the real condition of the asset and price the uncertainty, not just the average.

For financiers, the stakes are larger because the exposure is leveraged across a book. Residual value risk on EVs is wider and more correlated with external shocks than the old tables assume. Policy changes can move a whole segment at once. The defensible position is not to guess the future more confidently. It is to price residuals with explicit uncertainty, stress-test against the downside of the range, and update as evidence arrives, so that no single policy shift or battery surprise quietly erodes a book that looked safe on paper.

The figures will vary by market and segment, and anyone offering one tidy depreciation number for "the used EV" is selling certainty that does not exist. The reckoning is not that EVs are bad assets. It is that they expose how fragile single-number, history-fitted valuation always was. EVs simply made the cost of that fragility too large to ignore.

A closing thought

The used car market is moving from a world where value was a known curve to one where value is a probability distribution shaped by battery condition, technology cycles and policy. That is uncomfortable for tools built on the old certainty. But it is also an opportunity for anyone willing to price honestly.

This is the conviction behind how we think about valuation at VehIQ: a value should come with a confidence interval and its sources attached, because in a market this dynamic, a number you cannot question is not a number worth trusting. The future of pricing is not more confident guesses. It is honest uncertainty, made useful.