Seven categories · 10 points each · balanced to 70
Every car on the Board gets one number out of 70, built from seven scores of ten. The point of the number is that it is comparable: the same seven questions, asked the same way, of every car we cover. Here is exactly what each one measures, how the two halves are balanced against each other, and the assumptions we hold fixed so the money lines mean something.
The single most important thing to understand about the Board: it ranks the best buy, not the best car. Those are different questions. On raw performance alone, the quick, expensive cars win. But a score of 70 rewards the whole ownership picture — what you pay, what it costs to keep, what breaks, and whether you can actually live with it — so a cheaper, cheaper-to-run car can finish above something faster. When a modest car beats a fast one here, that is the system working, not a claim that it is the better machine.
Three of the seven categories are about driving (The Drive, The Limit, The Mods) and four are about ownership (The Deal, The Upkeep, The Risk, The Daily). A car has to be good at both to win.
The two halves are balanced against each other, and that is worth explaining. Every category is scored out of ten and none of them outranks another. But there are four ownership categories and only three driving ones, so for a long time ownership could swing 40 points and driving 30 — not because we decided it mattered more, but purely because there happened to be more of them. That ratio would have shifted again every time we added a category, with nobody choosing it. So each half is now normalised to 35: ownership out of 40 scaled to 35, driving out of 30 scaled to 35, added together for the score out of 70.
The practical consequence is that the seven scores on a car’s scorecard no longer add up to its total in a single sum, so every scorecard shows both halves under the number. Add the four ownership scores, add the three driving scores, and you can check our working.
Money numbers are only comparable if they are calculated the same way for every car, so we hold a fixed set of assumptions and state them on every page:
Some figures on this site carry a mark. They are there because a research note that says "we could not source this" is worthless if it never reaches you, and because a number with no mark should be one you can check:
Where a figure had no honest basis for an estimate at all, the cell reads Not captured and the row simply carries less. We would rather publish a gap than fill it with another car's number.
What it costs to get in, and what it gives back when you sell. This is the used buy-in price against how the car holds its value — a cheap car that depreciates hard and an expensive one that barely moves can end up in very different places here. Buy-in is the KBB private-party figure for the scored trim and year.
How it feels on a good road. This is a feel-and-refinement score, and deliberately not a power-to-weight ranking or a stopwatch reading. A car with a worse power-to-weight number can score higher if it is more satisfying to drive, and it works the other way too. We decide this weighting on purpose: the question is whether you will enjoy the drive, not whether it wins a drag race.
What it costs every month once it is yours — fuel, tires, insurance, and routine service, all under the fixed assumptions above. As with insurance, the rank order is the reliable part; the precise dollar deltas between cars are directional. This is the category that quietly decides a lot of value cases, because it is money you spend whether or not you ever push the car.
What can break, how likely it is, and what the bill looks like. We start by counting the catastrophic failure modes a car has in the version we scored, then ask whether a pre-purchase inspection would actually find them. That is why the repair figure alone does not set the score: a rare-but-catastrophic failure and a common-but-cheap annoyance can land on the same number, and a flaw an inspection cannot see is worse than a dearer one it can. We show the worst realistic repair in dollars so you can see what you are signing up for, and where a known problem is tied to a specific year or version, the scope line does real work: the same car can carry more or less risk depending on which one you buy.
Could this honestly be the only car you own? This is livability rather than raw space. Seats and doors set the band, cargo places the car inside it, and ride, noise and cabin break the tie, which is how two cars with identical boot space can finish a point apart. Two seats cap this score no matter how good the car is elsewhere; back doors, real cargo room and everyday comfort lift it.
How it behaves when you actually push it — engagement and composure at the edge, autocross and track pedigree, and seat time. Crucially, this is not a grip-and-braking ranking. A lighter, more playful car pulling less lateral g can out-score a grippier, faster one that is numb or nervous when pushed. We show the grip and braking figures for context, but the score is about how the car behaves at the limit, not the size of the number on the skidpad.
What a build costs, how far the platform goes, and what it costs you back in reliability. Some cars unlock a lot of power for very little money; others take real money to move at all, or wall out early. We look at the cost of the first meaningful step, the ceiling of the platform, and how much risk you take on climbing toward it.
Knowing what a category measures still does not tell you what a 6 is. So here is the ladder we score against, with the car and the figure that anchors each rung. Every rung below is held by a car we have actually scored, and we score new cars by finding the rung their figures earn rather than by picking a number that feels right.
Two things worth saying plainly. No car we have scored has ever earned a 10 in any category, and only one has ever gone below a 3, so those rungs are defined in our internal rubric but no car sits on them and we will not invent one to fill the row. And the rungs are derived from the whole Board rather than decided in advance: we read all 154 scores, worked out what the numbers were already doing, and wrote that down. Where the Board contradicted itself, we say so on the car's own page instead of quietly smoothing it over.
What we read first: KBB private-party buy-in for the scored trim and year. Retention is a modifier, never a route: a dear car cannot buy its way up this category by holding value, or an appreciating exotic would outrank a cheap car and the Board would stop ranking best buy. The price band leads: the Supra holds 87% at $42,000 and scores 6, the 370Z holds 92% at $18,000 and scores 8.
What we read first: TWO reads, each out of 5, summed. Steering feel is what the front tires tell you; powertrain character is whether the engine is a reason to own the car. Scored SEPARATELY and added, because as one 0-10 read this category returned a 7 for 12 of 22 cars and used 4 of its 11 values. Power-to-weight is context and does not rank: the Genesis Coupe and the Focus RS both run 10.0 lb/hp and now score 5 and 9, which the split explains instead of asserting.
What we read first: Annual running cost at 12,000 miles a year on the fixed assumptions above: fuel at the octane the car requires, tires, insurance, routine service. The figure is a dollar total, so the rungs here are dollar bands and the score follows arithmetic rather than judgment. THE FLOOR IS EXTERNALLY ANCHORED and the spacing is not: a 10 is fixed to a real car costed on our own assumptions ($2,794), while the band width is tuned to separate the cars we actually cover. Anchoring both ends was tried and rejected, because an E60 M5 runs to roughly $9,964 a year and an evenly spaced scale across that range puts almost every car we cover on a 9 or an 8. The cars on this Board occupy a narrow slice of what running a car can cost.
What we read first: How many catastrophic failure modes exist in the version we scored, then whether a pre-purchase inspection (PPI) would find them. The repair dollar is context rather than the driver: the 128i's worst realistic repair is $800 and it scores 7, the Miata ND's is $1,200 and it scores 8, the 370Z's is $900 and it scores 6.
What we read first: Seats and doors, then cargo volume, then ride, noise and cabin as the tiebreaker. This is LIVABILITY rather than raw space, and the Board has always scored it that way: the Nissan Z and the 370Z carry the same 6.9 cu ft and two seats and score 5 and 4, because one rides like a GT and one is stiff, loud and dated.
What we read first: Behavior at the edge, plus track and autocross pedigree and host seat time. SCCA class is context and does not rank: the 370Z, 350Z, G37 and Genesis Coupe are all F Street and score 8, 7, 6 and 4. Grip is context too, and the Miata at 0.90 g out-scoring the Supra at 1.02 g is the category working as designed.
What we read first: Cost of the first meaningful step, then the ceiling on stock internals, then what climbing costs in reliability. The dollar alone does not rank, and the reason is that the figures buy different things: $800 gets the Nissan Z to roughly 420 whp and $8,000 gets the Miata ND to roughly 275 whp. Those are not the same purchase.
