SPLap · engineering notes

Reading room

The short list of work that shaped how SPLap thinks about race telemetry, and where the next release is headed. One honest line each on why it earned a spot. Nothing here is ours; all of it made ours better.

Predictive analytics in racing

Tulabandhula & Rudin, Tire Changes, Fresh Air, and Yellow Flags: Challenges in Predictive Analytics for Professional Racing · Big Data, 2014 (open access via MIT DSpace).
The clearest statement we know of why racing is a hard prediction problem: cautions, tyre changes and traffic break the clean time-series assumptions most models lean on. SPLap's answer to that hardness is humility, which is why our Race Engineer emits deterministic findings with confidence and evidence instead of forecasts. Their yellow-flag analysis is the intellectual ancestor of our Incident Review evidence windows.

Race strategy simulation

Heilmeier et al., Virtual Strategy Engineer: Using Artificial Neural Networks for Making Race Strategy Decisions in Circuit Motorsport · Applied Sciences, 2020 (open access).
The TU Munich group taught neural networks to make pit-stop decisions. We cite it partly as the strongest version of the approach we deliberately did not take: SPLap keeps every finding reproducible in SPL with no learned model in the loop. Read both and you understand our design space.

Heilmeier et al., Application of Monte Carlo Methods to Consider Probabilistic Effects in a Race Simulation for Circuit Motorsport · Applied Sciences, 2020 (open access).
How to simulate a whole race when safety cars and lap-one chaos are random variables. This is the shape of what SPLap's Strategy Lab wants to grow into, with the simulation transparent and the parameters on screen.

TU Munich FTM, race-simulation · open source (Python).
The companion code to the papers above: a working, inspectable race simulator for pit-stop strategy. If you want to go from watching SPLap replay Monza to simulating your own strategy calls, start here.

Tyre degradation

A State-Space Approach to Modeling Tire Degradation in Formula 1 Racing · arXiv, 2025.
Recent work treating tyre degradation as a state estimation problem over lap times, which is very close to how we intend to model it in SPL: degradation as a slope you can watch drift, not a curve you take on faith. This is the homework for our Phase 2 physics.

Data and tooling we build on

OpenF1 · community API.
Every observed event in SPLap comes from here, under its non-commercial terms: laps, sectors, stints, pit stops, positions, intervals, race control, weather and 3.7 Hz car telemetry. The project that makes this whole category of hobby work possible.

FastF1 · open source (Python).
The standard Python toolkit for F1 timing analysis. We took a different route (SPL and lookups instead of pandas), but if your happy place is a notebook, this is the canonical door in.

Spotted something we should read, or think we read one of these wrong? Tell us. The best kind of correction comes with a citation.