How to Read a Dyno Sheet: Correction Factors, Smoothing and Honest Numbers

A dyno sheet looks like a fact. It has axes, a curve and a number at the top, and it gets screenshotted into forum posts as though it settled something.

It is better understood as a measurement taken under a set of chosen settings. Change the settings and the same car produces a different sheet, honestly, with nobody lying. As of September 2026 the most useful skill an owner can have here is not reading the peak number — it is reading the conditions and settings printed around the edges of the graph, because those decide what the peak number means.

A typical dyno session runs roughly $300 to $600 depending on region and how much tuning time is included, which is enough money to be worth understanding what you are buying. What that session actually consists of is covered in dyno session cost and what happens.

What a chassis dyno actually measures

A chassis dyno does not measure horsepower. It measures how quickly the car can accelerate a known mass, or how much force the car produces against a known resistance, at the tyre contact patch. Horsepower is then derived from force and speed.

That derivation is where the variables enter. The two broad families behave differently and it matters which one you were on:

Inertia dynos accelerate a heavy drum and calculate power from how fast the drum's speed increases. They are simple, repeatable, and cannot easily hold the car at a steady point.

Load-bearing dynos apply controllable resistance, which means the car can be held at a fixed speed and load while the tuner works. This is far more useful for actual calibration because it lets a tuner sit in one cell of a table and study it, which is how VE table correction is done properly.

Either way, what comes out of the machine is a wheel figure. Everything after that is arithmetic and assumptions.

The three settings that move the number

Correction factor

Air density changes what an engine can produce. A cold, dry, high-pressure morning makes more power than a hot, humid afternoon, with nothing about the car changed — the same physics discussed in heat soak and density altitude.

A correction factor mathematically adjusts the measured result toward a defined reference condition so results taken in different weather can be compared. Several standards exist and they use different reference conditions, so the same raw measurement corrected under two standards produces two different numbers.

The practical rule: a sheet without a stated correction standard and the actual air conditions is not a sheet you can compare to anything. Any reputable operator prints both.

Smoothing

Raw dyno data is noisy. Smoothing averages neighbouring points to produce a readable line, and the amount is a setting.

Light smoothing leaves a curve with texture, including small dips and steps. Heavy smoothing produces a clean arc that looks professional and quietly erases exactly the features a tuner needs to see. A dip at a particular engine speed might be a fuelling hole, a resonance, a cam timing issue or a boost control wobble — all worth knowing, all invisible under heavy smoothing.

Ask for the setting. Prefer less. A slightly ragged curve from a competent shop is worth more than a glossy one.

What is being reported

Figure on the sheet	What it is	How much to trust it
Wheel horsepower / torque	Measured at the tyre, corrected	The measured result — the honest number
Engine or crank horsepower	Wheel figure plus an assumed drivetrain loss	Depends entirely on the assumed loss percentage
Peak power	The single highest point of the curve	Real, but says little about how the car drives
Area under the curve	Not usually printed; read it visually	What you actually feel on the road
Air-fuel ratio trace	Measured by the dyno's own analyser	Often the most valuable line on the page
Timing / knock trace	From the vehicle, if logged	Tells you what the peak cost in safety margin

Drivetrain loss deserves a note. It is not a fixed percentage — it varies with drivetrain type, gear selection, tyre pressure, temperature and how the car is strapped down. Any crank figure on a chassis dyno sheet is an estimate built on a chosen number, which is why the same car appears at several different crank ratings in different advertisements.

Reading the shape, not the summit

Peak power is one point on a line the car spends almost no time at. What you feel driving is the whole curve, especially the torque curve in the range where you actually operate.

Three things to look for:

Where torque arrives and how flat it stays. A curve that reaches a high plateau early and holds it will feel dramatically stronger than a peakier curve with the same headline number. This is why a modification that "only added twelve horsepower" can transform a car — it filled in the middle.

Dips and steps. A clean curve rises and falls smoothly. A localised dip is information. Common causes include a fuelling error at that load point, a resonance in the intake or exhaust, cam phasing transitioning awkwardly on a variable valve timing engine, or boost control overshooting and settling.

Where the curve stops. A pull that ends abruptly rather than falling away naturally usually means the operator lifted, the car hit a limiter, or something intervened. Ask which.

The traces that matter more than the peak

If the sheet includes an air-fuel ratio trace, read it before the power line.

Under full load the engine is in power enrichment and running open loop, which means fuelling comes from a table rather than from sensor feedback. The mixture that trace shows is what the calibration actually delivered, and a mixture that wanders or leans out as the pull progresses matters far more to whether the engine survives than any power figure. The dyno's own analyser is doing this measurement, which is separate from whether your car carries its own sensor — a distinction covered in do you need a wideband before tuning.

If the sheet includes a timing or knock trace, read that second. A peak number achieved while the engine was pulling timing to protect itself is a number produced under duress. It will not repeat on a hot day, and the calibration behind it is asking for more than the fuel is giving. What counts as normal there is set out in how much knock retard is normal.

A sheet with only a power curve on it is the least informative version of the document. It is also, not coincidentally, the version most often used for advertising.

Comparing two sheets honestly

The only genuinely valid comparison is the one taken on the same dyno, on the same day, with the same settings, same fuel, same gear and same tyre pressures, with only the calibration changed between runs. That is exactly how a before-and-after tuning result should be produced, and it is why a good shop will insist on a baseline pull before touching anything.

Everything else degrades from there. Same dyno, different day: usable, with correction applied and conditions noted. Different dyno, same car: indicative at best. A sheet from the internet against your sheet: not a comparison at all, however satisfying it is to make.

Three specific traps:

• Gear selection. Pulls done in different gears load the engine for different durations and heat-soak it differently. Same gear or no comparison.
• Tyre pressure and strap tension. Both change how much energy is lost between crankshaft and drum, and neither is usually printed.
• Successive-pull fade. If pull three is lower than pull one, the car is getting hot. That is real behaviour worth knowing, and it is also why comparing your third pull to someone else's first is meaningless.

Why a datalog often beats a dyno sheet

A dyno sheet is a summary. A datalog is the evidence.

The sheet tells you the car made a number. The log tells you what the air model reported, what the fuel trims were doing, how much timing was actually commanded versus requested, what the intake air temperature had climbed to, and what the mixture was doing at every point rather than at peak. Almost every question that starts "why did it only make…" is answered in the log rather than on the sheet, and that is true whether the pull happened on rollers or on a road — the reading method is the same and is laid out in reading your first wide-open-throttle pull.

The practical consequence for an owner: if you are paying for dyno time, ask for the datalogs as well as the graph. The graph is what you show people. The log is what lets anyone — including you, later, or a different tuner — understand what actually happened.

That is the analysis TuneVault is built to do. It reads your HP Tuners VCM Editor tables and audits them against your logs, so the question stops being "is this number good" and becomes "is this number safe, and where is the rest of it". A guided tune with a datalog verification pass is $39, which is a small fraction of a dyno session and answers a different question than the rollers do — the comparison is in what the copilot adds on top of HP Tuners alone, and the plans are on the pricing page.

The one-paragraph summary

Read the conditions, the correction standard and the smoothing setting before the peak. Read the torque curve's shape before its summit. Read the mixture and timing traces before you celebrate. And compare only against sheets produced under the same rules on the same equipment. A dyno is an excellent instrument used carefully and an excellent marketing device used carelessly, and the difference is entirely in the margins of the page.

Frequently asked questions

Why do two dynos give different numbers for the same car?

Because a chassis dyno measures force at the tyre and then applies a set of assumptions to turn that into a horsepower figure. Different dyno types load the car differently, correction factors adjust for air conditions using different standards, and smoothing settings change the shape of the curve. Two honest measurements of the same car on the same day can differ by a noticeable margin without either being wrong.

What is a dyno correction factor?

A mathematical adjustment that estimates what the engine would have produced under a defined reference set of air conditions, so results from a cold morning and a hot afternoon can be compared. It works from air temperature, pressure and humidity. Different standards use different reference conditions, which is why the correction standard used should appear on any sheet you are asked to take seriously.

Does smoothing change the horsepower number?

It changes the shape considerably and the peak somewhat. Smoothing averages adjacent data points to remove noise, and heavier smoothing rounds off spikes and dips. A heavily smoothed curve looks beautifully clean and hides exactly the small dips that tell a tuner something is wrong. Ask for the smoothing setting, and prefer sheets that show less of it.

Is wheel horsepower or crank horsepower the real number?

Wheel horsepower is measured; crank horsepower on a chassis dyno sheet is calculated by adding an assumed drivetrain loss. The measured number is the honest one. Any crank figure quoted from a chassis dyno depends entirely on the loss percentage someone chose, which is why the same car can be advertised at several different crank numbers.

What should I look at on a dyno sheet besides peak power?

The shape of the torque curve, the air-fuel ratio trace and any knock or timing trace, if the sheet includes them. A curve with a dip or a sudden flat spot is telling you something mechanical or calibration-related. A mixture that wanders across the pull matters more for engine health than the peak figure does, and a peak with retarded timing behind it is not a peak you should be pleased with.

How many pulls should a dyno session include?

Enough to show repeatability, which usually means at least three comparable pulls with cooling between them. A single pull is a data point, not a result. If successive pulls fall off noticeably, the car is heat-soaking rather than making less power, and that itself is useful information about how the car will behave on a hot day.

Can I compare my dyno sheet to numbers I saw online?

Only very loosely. Comparing across dyno types, correction standards, smoothing settings, fuel, altitude, gear used and tyre pressure is comparing measurements taken under different rules. Comparisons are meaningful on the same dyno, the same day, with the same settings — which is exactly why before-and-after tuning is done that way.