We turn racing data into predictions we grade in public.

What we do

One engine, pointed at every session. We read the raw timing and telemetry and turn it into analysis you can use: how the tyres are degrading, where the real race pace sits, how a strategy is likely to unfold, and what the weekend ahead looks like. The output is the charts, the session reads and race debriefs, and forward-looking predictions that we put on the record and grade against the result.

The machinery is the craft: an automated pipeline that ingests every session the same way, prediction models, and quality checks that catch a bad number before it reaches a chart. That is what we are really building. Formula 1 is what we point it at first.

Why Formula 1 first

Formula 1 is an unusually honest place to build an analytics process. The timing and telemetry are public, a fresh result lands roughly every other weekend, and the audience checks the maths. A read made on Friday is right or wrong by Sunday, in front of everyone.

That fast, public scoreboard is exactly what an automated process needs to get good. It forces us to check the data before we trust it, set aside the laps that are not comparable, and grade our own calls in the open. We started here because the sport hands us a demanding test we cannot quietly fail.

Where this is going

Formula 1 is the first domain, not the last. We chose it because it lets us prove the method in public and sharpen it every week. In time we intend to bring the same automated analytics to other fields. The racing work is where we earn the right to say it.

How we work

• Straight from the source. Every chart is built from the raw timing and telemetry of the session itself, never from broadcast footage or someone else's graphics. The data is the sport's, and it is public; what is ours is the pipeline that reads it, the models, and the analysis we publish.
• Calibrated, not loud. Numbers carry their uncertainty with them, and we say plainly when we do not know yet. We are after precision, not volume.
• Fast, because it is automated. The pipeline turns a session around in hours and holds the same standard every time, with checks it cannot skip. Quick and right, not one traded for the other.
• Graded in public. We make reads you can check, then grade them on a fixed schedule where you can see how they landed, whether they went our way or not. Every number we publish traces back to the session data behind it.
• Independent. Not associated with Formula 1, any team, or any sanctioning body.

The name

Tempering is a real process. Steel is hardened, then reheated under controlled heat to draw out the brittleness. Done well, it trades a little hardness for toughness that lasts.

That is the promise in the name. Analysis that is calibrated rather than loud, built to hold up after the chequered flag, not just to win the afternoon. A claim has to survive the next race weekend before we stand behind it.

The mark

Our mark is a sector gauge: three arcs and a heat point at the centre, like the dial of an instrument reading off a lap. The colours are not decoration. When steel is heated, it runs through a real oxide spectrum as it warms, from straw to bronze to purple and finally to a cool steel blue. That spectrum is our palette, and it ends on a cool steel blue that sits beside the blue our charts use, so the brand and the analysis are of a piece.

At small sizes the gauge stands alone, built to stay recognisable at a glance on a phone screen.

Who's behind it

A small, independent operation that builds automated analytics. One editor has the final say on everything we publish, and our automated engine does the heavy lifting, drawing the charts and running the models. Over all of it sits one rule: nothing goes out until the numbers check out.