Founding season 2026–27

Was that me, or was that the car?

Every lap you run, read at its own sample rate and kept on one permanent record — with every figure carrying how it was arrived at, and a blank wherever nothing was established.

Open to any team running a Ferrari 296 Challenge. Reads the Marelli WinTAX export your engineer already produces — no new hardware, no new wiring.

10Tenths Analytics
Brake demand > 5 bar Throttle > 80% Neither Yas Marina · 2,511 GPS samples · 20 Hz
Speed272km/h
Gear6
Throttle100%
Brake0bar

Drag along the lap — or tab to it and use the arrow keys

Laps on the record 761

Every WinTAX export the team has produced, parsed and kept.

Timed line to line 462

Laps where the window demonstrably held two crossings of the start/finish line.

Shown as blank 299

The rest. No estimate, no time borrowed from a neighbouring lap. A dash, and the word unverified.

Circuits 5

Yas Marina, Dubai Autodrome, Bahrain, Jeddah Corniche, Losail — each length measured from its own exports, never looked up.

Why 299

Most software would have shown you 761 lap times.

It would have taken the logger's own lap-time channel at face value, or the difference between two timestamps, and printed a number for every lap. For two in five of them there was no second line crossing to measure against — so whatever it printed, nothing on the screen would have told you which two in five to distrust.

  1. 01

    Professional software is expert-bound

    MoTeC i2, AiM, WinTAX. Channel-by-channel trace on a desktop, and the answer is only as available as the person who can read it.

  2. 02

    The engineer's memory walks out of the door

    One memory, one laptop, one weekend at a time. What he worked out in Bahrain in March is not on the record in Jeddah in November.

  3. 03

    Sector times do not tell you where

    “Slow in sector two” describes the timing screen. It is not information about the corner, the metre, or the tenth.

  4. 04

    Silent failures are never silent

    Bleeding tyres, a decaying brake line, a setup drifting off its baseline. All of it is already in channels nobody is watching.

The platform

Six products. One is the record.

The record is the asset, and it is the first of the six. The other five are layers on it, and every layer inherits the same approval architecture — an engineer signs off anything consequential, against an append-only log.

  1. FOUNDATION

    10T Unified Record

    Founding season

    Laps, setup, driver, tyres and conditions filed together for every session, each figure linked to whatever it was derived from. It keeps growing, and it does not leave when a person does.

  2. FOUNDATION

    10T Team OS

    Founding season

    Every session, setup, decision and signature, versioned and permanent. The run plan your engineer writes on Friday and the setup he signs on Sunday are the same object a year later.

  3. ATTRIBUTION

    10T Performance Intelligence

    Founding season

    Splits the tenths you lost into the ones a driving action explains and the ones nothing yet explains — and prints the second number rather than absorbing it into the first.

  4. INTERFACE

    10T Engineer

    Founding season

    Turns the record into a briefing in plain English. The numbers are computed before it is asked to say anything; it phrases them, and it is not permitted to introduce one that is not in the payload.

  5. NETWORK

    10T Fleet Data Network

    Opens with a fleet

    Identical-car benchmarking. Your laps stay yours: nothing crosses between teams, and no other team's answers are sharpened by your data without a signed agreement that says so.

  6. VERIFICATION

    10T Validation Lab

    Season two

    Pre-registered, paired-car A-B-A-B testing for suppliers. The fee buys the test; the result is published either way.

How it handles a number

A value you did not measure is worse than no value.

A wrong lap time does not sit there being wrong. It becomes the reference for the next comparison, and the setup change after that. So every figure carries how it was arrived at, and anything unestablished is shown absent rather than filled in.

  1. Measured

    Read directly off the car. A lap time is measured when the window demonstrably held two crossings of the start/finish line — the elapsed time between them, and nothing else.

  2. Calculated

    Derived from measurements by a stated method. A time loss accumulates over distance, so it can be calculated for a lap that could never be timed — which is why one figure can be present while the one above it is not.

  3. Inferred

    Worked out rather than recorded — which circuit a lap belongs to, from the distance covered before it wrapped. Shown, and labelled, so you know which kind of thing you are trusting.

  4. Entered

    A person typed it. That is a perfectly good basis, and it is not the same as a measurement, so it does not get to look like one.

  5. Unverified

    Nothing was established, so nothing is shown. A dash, and the word. Not an estimate, not the previous lap’s time, not a plausible number that would have made the screen look finished. Two in five of the laps on the record are in this state, and the product says so on every one of them.

Four fields on a lap record — driver, circuit, car and lap time — each labelled with how it was established: inferred, inferred, entered, and unverified.
One lap, four values, four different bases. Driver and circuit inferred from the data, car entered by a person, lap time unverified — so it is a dash.
A lap header showing lap time 1:59.485 marked measured, time lost 4.275 seconds marked calculated, and the split between driver-attributed and unattributed time.
A lap it could time. Two crossings of the line in one window, so the elapsed time between them is the lap time — measured, no qualifier needed.
The same lap header for a different lap, where lap time shows a dash marked unverified, while time lost and the attribution split still carry calculated values.
A lap it could not. A dash and the word unverified — not an estimate, and not a time quietly borrowed from the lap before. The loss is still calculated, because a delta accumulates over distance and needs no line crossing to exist.
How it works

From the export he already makes.

Nothing to fit to the car, nothing to change about the weekend, nothing new for your engineer to learn on a Saturday morning.

  1. 01

    He exports the session, as usual

    A Marelli WinTAX .mat. If your second car runs a different logger, tell us before you reserve — today this reads WinTAX and nothing else.

  2. 02

    The record parses it

    Every channel at its own native rate, taken from the manifest rather than guessed from array length. Versioned, queryable, immutable. Once it is on the record it does not move.

  3. 03

    Answers, before the helmet is off

    The corners that cost the lap, one change for the next run, and the confidence attached to it. Your engineer approves it, and the next run starts from the new baseline.

Founding season

Reserve a founding seat.

Open to any team running a 296 — no waiting list, no invitation. Teams that register during the founding season keep founding terms, in writing, for as long as they stay.

Founding seats taken

Leave an address and a person sets your team up; there is no self-serve sign-up yet, so credentials follow separately rather than instantly. Your laps are yours, and that matters more on an open platform: the team on your token is the only team any request can touch, including requests from teams you race against. Nothing crosses between teams. You can export the record in full at any time, and deletion on exit means deleted, not deactivated.