How to analyze telemetry data from a track day.
Most drivers open a session, look at the lap times, and close it again. The gap between that and a real debrief is not more data. It is an order of questions. This is the order, in five steps, with the numbers from one track day at Autodrom Most.
Nothing here is specific to us. Any competent engineer works through the same five questions. Where Vetkuro answers one of them, it says so; where you are on your own, it says that too.
Mobile
Web
Step 1
Can I trust this recording?
Every hour wasted in telemetry starts here. A driver stares at a strange wobble in the line through T3 for twenty minutes before realizing the satellite fix dropped. Check the recording before you read anything into it.
Two things to look at. First, resolution: how far did the car travel between two recorded points? Second, integrity: were there moments where the receiver was guessing?
The session carries both. Accuracy, FixStatus, SatelliteCount and SatelliteHDOP are recorded channels, not hidden diagnostics, so a bad corner and a bad fix are two different things you can tell apart.
| Sample rate | Distance | What you can read |
|---|---|---|
| 1 Hz | 50 m | A corner is one point |
| 10 Hz | 5.0 m | You see that you were slower |
| 25 Hz | 2.0 m | You see where you turned in |
Work out your own numbers on supported devices.
Two habits that keep this step honest
Where the data is missing, Vetkuro leaves it missing. A dropped fix shows as a gap rather than a smooth, confident curve through nothing.
Values worked out from motion rather than read from the car are named for it: ThrottleEstimate, BrakeEstimate, GearEstimate.
Step 2
Which lap am I measuring against?
A comparison is exactly as useful as its reference. Measure a clean lap against one that caught traffic and you get a dramatic graph with nothing in it. The shape is someone else's Porsche, not your driving.
Coaches tend to say the same thing about which lap to take: the one you could go out and drive again. So read the times before you choose. Laps sitting within a few tenths of each other were driven much the same way, and any of them is a fair reference. A single lap half a second clear of everything else was a one-off – a gap in traffic, a tow, a kerb that happened to pay – and measuring against it asks you to repeat something you cannot.
Keep the conditions close too. Same layout, same tires, roughly the same weather. A lap from a cold morning in April is a legitimate target, but it answers a different question from the one you are asking ten minutes after getting out of the car.
Then pick according to what you are asking. Your best clean lap answers what did I do differently that time. Your theoretical best answers what is already in my hands. Your previous personal best answers am I getting better.
Set a reference lap in Vetkuro →Three legitimate references, one session
Best clean lap
Lap 3, flagged Ref in the list below. The default, and the right one to start with.
1:58.905
Theoretical best
Your fastest zone times from this session, joined up. How it works.
1:58.327
Previous personal best
Earlier the same day, same layout and car. The lap this run is chasing, and the one that says whether the day is going forward.
1:57.838
Not a reference
The out lap and the in lap. Vetkuro leaves them unticked, and so should you.
3:21.85
Step 3
Where did the time actually go?
Zone splits narrow a lap down to one corner at a time. A delta curve narrows it to a meter. But both depend on one decision that most people never notice they are making: what goes on the horizontal axis.
Line two laps up by the clock and the faster one arrives everywhere earlier. By the back straight you are comparing its braking zone against the other lap's full throttle. Line them up by distance travelled and 2 100 m is the same tarmac on every lap you ever recorded there. Everything downstream is then a statement about a place, which is the only kind you can act on next time round.
In the product
Delta along the lap, zone by zone
The delta view plots one number against lap distance and bands the lap into its zones, each named by the turns it covers. Clicking a point on the curve moves the track map and the onboard video to the same place on the lap, so the corner where the line turns is one click away.
Read the slope, not the height
Lap 9 against lap 4 of the morning run: 0.2 s down leaving the T4-T5-T6 complex, 0.495 s up at the line. A flat curve at +0.300 s means you are behind but driving that section just as well. Chase the parts where the line is steep.
After T4-T5-T6
+0.208 s
At the line
−0.495 s
Where it turns is where to look
The curve turns on the way out of T6 and falls from there to the line, steepest in T15: three tenths in one corner. That is the corner to open first, because it is the one you already know how to drive. And time gained or lost on a corner exit shows up down the following straight, so read slightly earlier than the turn in the curve.
Read a delta curve properly → Open the delta view in Vetkuro →Sources that never agree on when to speak
A position stream, an OBD adapter and a heart-rate strap sample at different rates and share no timestamps. They are resampled onto that same distance axis, each by a rule that suits it.
linear
Speed, throttle, temperatures, G.
circular
Heading. Halfway between 359° and 1° is 0°.
hold
Gear and zone number. There is no gear 3.4.
Step 4
What happened in that one corner?
This is where analysis stops being arithmetic. You have a corner and two laps. Now you narrow the distance range to the braking zone and stack the traces that explain it: speed into the corner, the point where the brake comes on, steering angle, lateral G through the middle, throttle on the way out.
Every trace shares the distance axis, a lap keeps one color across all of them, and moving the cursor moves the map and the onboard video with it. With 2.0 m between samples you are looking at a real difference in turn-in point, not at a rounding error.
-
Entry speed
Two laps a few km/h apart at the same point mean the difference started before the corner.
-
Braking point
Where Brake leaves zero, in meters, so "twelve meters later" is a measurement.
-
The middle
LateralG and SteeringWheelAngle together show a corner taken in one arc or in two.
-
The exit
Where Throttle gets back to full. This is usually the tenth, and it is paid out down the straight.
Step 5
Was it me or the day, and did my change work?
The last step is two halves of the same discipline. Before you credit yourself with an improvement, rule out the things that improved on their own. And having decided what to change, change exactly one thing, then come back and check.
Each session keeps its own conditions and vehicle, so a comparison across two days stays a comparison. Air temperature, humidity, pressure and wind travel with the recording, and if the last laps of a stint keep losing tenths, HeartRate is often a better explanation than technique.
In the product
Two sessions from different days, side by side
Each session carries its own weather, vehicle and lap list. Comparing across days lines the laps up on the layout distance, so the same corner matches even though the recordings are years apart, and the conditions sit next to the times rather than being forgotten.
15 April 2023
1:58.201
S1 71.966S2 46.235
9 October 2026
1:57.838
S1 71.953S2 45.885
Sector 1
−0.013 s
Sector 2
−0.350 s
Over the lap
−0.363 s
This is the tell. Three and a half years apart and 0.363 s faster, almost all of it in sector 2. A gain spread evenly across the lap is mostly the car and the track. A gain concentrated in one sector is far more likely to be you, and that is the one worth repeating deliberately next time.
Compare two of your own sessions in Vetkuro →Three things this does not do for you.
Any tool that claims to do the whole debrief is selling you the part it does badly. Three limits worth knowing before you sit down with a session.
-
Sectors come from the layout
Autodrom Most has one sector line, at 2 472 m, and it is the same for everybody, which is what makes sector times comparable at all. It also means you cannot draw a split either side of the one corner you are working on. Use the delta curve and the distance range for that.
-
There is no verdict button
Vetkuro will show you that lap 9 found three tenths in T15. It will not tell you whether that was the braking point or the exit. Reading the traces is the skill, and it is yours. This page exists because that skill is learnable.
-
What you changed on the car is your note
Tire pressures, damper clicks, fuel load and which set of tires went on are not in the session. Step 5 depends on knowing them, so keep them somewhere. A notebook in the trunk beats a perfect memory.
