A 5-Star CSR2 Car Can Be Less Built Than a 4-Star

Players use star count as shorthand for how good a car is. Five stars sounds finished; four sounds like work remaining. We checked whether the rating tracks build strength at all, and the answer is barely.

We already knew star count tracks recorded time within a tier — five-star Tier 5 cars post a median 7.232 against 8.473 for four-star. This is a different question: does a higher star rating mean a more built car?

The ranges overlap almost completely

The medians rise cleanly — 1,189, then 1,642, then 2,264. On medians alone the rating looks like it works.

The ranges tell a different story. Four-star cars run from 933 to 2,447 EVO. Five-star cars run from 785 to 2,699. The five-star range starts *lower* than the four-star range does.

614 five-star cars are less built than the best four-star

Count it directly. The strongest four-star Tier 5 car on the sheet carries 2,447 EVO. The number of five-star cars carrying less than that is 614.

77% of the five-star cars in Tier 5 are out-built by at least one four-star car. In Tier 4 it is 79%.

So the rating is not a build meter. A five-star car on 785 EVO and a five-star car on 2,699 EVO have the same number of stars and are nothing like the same car — the gap between them is larger than the gap between the best and worst four-star car.

What the rating is actually telling you

Putting this together with what we found before, star count appears to describe the car's standing within its tier rather than the state of its build.

That explains both results at once. It tracks recorded time, because a car that ranks highly in its tier tends to be quick. It does not track EVO, because two owners of the same five-star model can have built it to wildly different levels.

In other words the rating is closer to a classification of the car than a readout of your work on it. The game assigns it; you do not earn it by building.

The two cars at the extremes

Naming them makes the point better than any statistic:

The Agera R and the Time Machine carry the same five-star rating and are 1,914 EVO points apart. That gap is larger than the entire four-star class, which spans 1,514 points from end to end.

Meanwhile the four-star Aston Martin Vantage on 2,447 EVO records 7.000 seconds and sits 265th in the game. 514 of the 776 timed five-star Tier 5 cars are slower than it.

So two thirds of the five-star class loses to a four-star car. If the rating described how good a car was, that could not be true.

How this interacts with fusion

Star colour and star count behave differently again, which is worth separating because they share a symbol.

Colour is a build state you control. Fusing a car turns the stars purple and the measured gain is consistent — a median 0.050 seconds across 226 matched pairs, with the purple copy quicker in 218 of them.

Count is not something you build toward in the same way. A car does not gain a star because you fitted more parts, which is exactly why the EVO ranges overlap so heavily. You can take a five-star car from 785 EVO to 2,699 without the count changing at all.

The practical reading: chase purple because it is a real gain you can measure. Do not treat the number of stars as a progress bar, because it is not measuring your progress.

Why this matters when you are choosing

The practical failure mode is comparing two five-star cars and assuming they are equivalently developed.

• Two five-star Tier 5 cars can differ by over 1,900 EVO. That is more than the entire range of the four-star class.
• A four-star car at 2,447 EVO is a better-developed machine than a five-star car at 1,500, whatever the rating implies.
• Use EVO to judge build state and stars to judge class. They answer different questions and neither substitutes for the other.
• Use recorded time for anything that actually matters. It is the only column that is a result rather than a description.

The same trap as PP, from a different direction

This is the third number on the car screen we have found to be misleading in a specific, identifiable way, and the pattern across all three is now clear.

PP counts fitted parts and is blind to Elite Customs, which moves a car a median 474 EVO without touching PP at all. Star count classifies the car and is blind to how far you have built it. EVO measures build strength and tracks real time closely, but stops discriminating at the very top where everything is built out.

Each is useful for the thing it measures and misleading for everything else. The mistake is not trusting any of them; it is assuming they are three attempts at the same measurement with one being most accurate.

Both columns sit side by side for every car in the record database, so you can sort by EVO and watch the star ratings jump around as you scroll.