APEXPULSEQUALIFYING INTELLIGENCE
Enter the pit wall ↗Exported data available

MACHINE LEARNING. RACING INSTINCT.

READ THE PACE.
ANTICIPATE
THE GRID.

The story of qualifying starts in practice.
We turn the laps into a forecast. You see the bigger picture.

Explore the forecast ↗How it works ↓

✓ Public data / Preserved predictions / Open methodology

THE NEXT RACING LINEROUND — / —

The next race awaits

Checking the public schedule

Loading the next circuit…

CIRCUIT OUTLINE / ILLUSTRATIVE LAPMOVE TO EXPLORE ↔

UP NEXT

Awaiting confirmed schedule

—DAYS
—HOURS
—MINUTES
—SECONDS
CLOSER TO THE START. ONE LIGHT AT A TIME.
Checking calendar… ↗

Lights: 7d → 24h → 6h → 1h → 5m. All out at the scheduled start.
Last verified: — · refresh every 5 minutes.

Monitored weekend

Las Vegas Grand Prix 🇺🇸

2026 · Round 21

Conventional weekend

Forecast monitoring active

Apex Pulse is monitoring the conventional practice-to-qualifying schedule for the next guarded prediction window.

01 / THE PIT WALL

Less noise.
More racing intelligence.

One clear view of the model’s call.
Explore a driver. Check the outcome.
See where the prediction stands.

Last completed prediction

Bahrain Grand Prix 🇧🇭

2026 · Kuala Lumpur · Kuala Lumpur

Qualifying completedPrediction after FP3
Settled · Updated 40 days ago

Weekend progression

Practice to qualifying

  1. FP1

    Completed

    Oct 2, 04:30 AM

  2. FP2

    Completed

    Oct 2, 08:00 AM

  3. FP3

    Completed

    Oct 3, 04:30 AM

  4. Qualifying

    Completed

    Oct 3, 08:00 AM

Status reflects exported artifact availability, not lap-by-lap live coverage.

QUALIFYING FORECAST

In predicted order. ↑ Higher ↓ Lower = MatchedArrows compare the official qualifying position with the prediction.

  1. P1+3.080sP4↓ 3Lower
  2. P2+3.089sP1↑ 1Higher
  3. P3+3.191sP3=Matched
  4. P4+3.328sP8↓ 4Lower
  5. P5+3.467sP5=Matched
  6. P6+3.615sP2↑ 4Higher
  7. P7+4.002sP6↑ 1Higher
7 / 22 DRIVERS

Model gap = estimated seconds to the pole benchmark. Arrows compare the official position with the prediction; rows stay in predicted order.

02 /THE CHEQUERED FLAG

The forecast meets
the finish line.

A prediction is only half the story.
Here is how this forecast
compared with qualifying.

MEAN GAP ERROR
2.439s

Average absolute gap error

POSITION ERROR
5.45

Average positions missed per driver

TOP 5 AGREEMENT
80%

Top-five overlap, regardless of order

TOP 10 AGREEMENT
70%

Top-ten overlap, regardless of order

This weekend only · 22/22 entrants covered · Not an estimate of overall model performance. Decorative graphics are not historical trends.

Explore the track record ↗
Artifact details
Artifact lifecycle
settled
Export timestamp
2026-10-03T09:46:51.977838+00:00
Protocol version
1.0
Checkpoint identifier
after_fp3
Forecast training policy
uniform
Registry lineage
valid_registry_lineage
Preflight status
ready_to_forecast
Protocol fingerprint
9b622b043c073d2a
Settlement valid
Yes
Weekend & data notes4 weekends without a forecast

Recent weekends without a forecast

  • São Paulo Grand PrixPractice timing data could not be loaded before qualifying. No retrospective prediction was added.
  • Mexico City Grand PrixPractice timing data could not be loaded before qualifying. No retrospective prediction was added.
  • United States Grand PrixPractice timing data could not be loaded before qualifying. No retrospective prediction was added.
  • Singapore Grand PrixThis weekend format is not supported by the FP3 model.

Prediction policy: season-aware selection is enabled after 5 eligible, settled events in the current season. Earlier forecasts use Uniform.

Data sources & methodology ↗

03 / UNDER THE HOOD

Built on data.
Driven by curiosity.

No crystal ball. No private team data.
Just a transparent machine learning project
exploring the pace before qualifying.

01↗

Listen to the laps.

Practice timing, representative laps, relative pace and session context become interpretable features.

PRACTICE → FEATURES

02↗

Find the signal.

A tabular model estimates qualifying gaps. Historical features use only information available before the weekend.

FEATURES → FORECAST

03⚑

Face the stopwatch.

The preserved forecast is compared with the qualifying result. Errors, missing drivers and limitations stay visible.

FORECAST → EVIDENCE
Data, limitations & methodology ↗

AN INDEPENDENT PROJECT BY MATTEO SGOBBA

A little obsession.
A lot of race data.

Exploring what machine learning can read between the laps.
Built in the open, for the love of the sport.

GitHub LinkedIn Email
Explore the project ↗