A data-driven tool that analyzes your historical telemetry to predict optimal cold tire starting pressures. Input your target hot pressures, the track and the number of laps, the predictor calculates the exact cold values needed to reach those targets during the session.
Traditional cold pressure selection relies on experience and general guidelines. A racing tires cold pressure predictor takes a different approach: it learns from your actual data. By processing historical sessions where you recorded both starting (cold) and ending (hot) pressures, the system builds a statistical model specific to your car, tires, and driving style. This data-driven method eliminates guesswork and provides predictions grounded in real-world measurements from your own racing history.
The predictor doesn't use generic lookup tables. Instead, it captures the unique pressure build-up characteristics of your specific setup. Whether you're running Pirelli DHE in Ferrari Challenge or Michelin slicks in GT racing, the model adapts to the actual behavior observed in your telemetry.
Every racing tire has an optimal operating window, a hot pressure range where grip and consistency are maximized. The challenge is determining the optimal starting pressure (cold) that will land you in that window once the tires heat up. Set too low and you'll be under-inflated mid-stint; set too high and you'll overshoot the target, losing grip and increasing wear.
The cold pressure predictor solves this by reversing the relationship: you specify your target hot pressures (based on tire manufacturer recommendations or your own testing), and the system calculates the corresponding cold values. This hot-to-cold calculation accounts for thermal expansion, carcass heating, and track-specific factors derived from your historical data.
For race stints (long runs), the predictor suggests relatively low cold starting pressures. However, starting the formation lap with such low pressures creates two risks: bead unseating (tire coming off the rim) due to under-inflation during aggressive weaving, and a softer carcass structure that makes it harder to generate heat in the tire. Engineers address this by intentionally setting higher pressures for the formation lap.
Once the car reaches the grid after the formation lap, the engineer performs a pressure bleed to bring the tires down to the optimal values suggested by the predictor. The bleed calculation uses the thermal gain coefficient (k = ΔP/ΔT), extracted specifically from formation lap telemetry uploads. This data is processed separately and not used for ML model training, since warm-up lap thermal behavior differs from racing conditions.
Not every team has full telemetry logging. Vehicles without TPMS (Tire Pressure Monitoring System) or teams using tire heaters/blankets cannot rely on automated pressure build-up logging. The cold pressure predictor from Excel file containing input/output user-inserted points supports this workflow. Using the provided template (Pressures_Database_Template.xlsx), you record cold pressures, hot pressures, track name, ambient temperature, tire temperatures during pressurization, and number of laps.
The system imports this Excel data and builds the same statistical model used for telemetry-based predictions. Required columns include cold/hot pressures for all four corners, track name, ambient temperature, and lap count. Suggested columns like individual lap times improve analysis accuracy. This flexibility means you can start using the predictor immediately with historical notes from your pit board.
Racing at a new circuit presents a challenge: you have no historical data for that track. How can you predict cold pressure for a new racetrack you've never visited? The track similarity matching feature addresses this by comparing the new circuit's characteristics against all tracks in your database.
The matching algorithm considers factors like corner count, lap length and average speed. It identifies the most similar tracks and uses weighted data from those circuits to generate baseline predictions. This gives you a reliable starting point for your first practice session, which you can then refine as you collect track-specific data.
Race car pressure build-up differs substantially between session types.
The pressure prediction for long run vs short runs accounts for these differences. You can select the Number of Laps to account for qualifying, race, or practice, the model applies the appropriate pressure build-up curves. This ensures your cold pressure recommendations are calibrated for whether you're chasing a single flying lap or managing tire performance over a race stint.
Understanding this distinction is critical: setting qualifying pressures for a race will leave you over-inflated by mid-stint, while race pressures in qualifying will compromise single-lap grip because of the under-inflation.
The predictor's accuracy improves with every session you upload. A statistical model trained on 5 sessions will provide reasonable estimates, but one trained on 50+ sessions across multiple tracks and conditions becomes highly reliable. Regular usage transforms the tool from a useful guide into a precision instrument calibrated specifically for your car and team.
Team-wide adoption amplifies this effect. When all engineers contribute data to a shared database, the model learns from collective experience rather than individual knowledge. A new engineer joining the team immediately benefits from years of accumulated data, avoiding the trial-and-error phase that typically accompanies onboarding. This centralization also prevents inconsistent practices: everyone works from the same data-driven baseline rather than personal preferences or outdated rules of thumb.
Managing multiple tire compounds across a race weekend requires more than just tracking pressures. Race Tires Management introduces compound-level organization: assign a compound to each run via a configurable dropdown (with your own custom compound list), and filter all processed data by compound to isolate behavior differences between tire types.
Interactive charts visualize pressure build-up patterns. The ΔP vs ΔT chart shows how pressure changes relate to temperature changes, while the ΔP vs Laps chart reveals how pressures evolve over the stint length. Both charts support per-wheel toggle, letting you isolate individual corner behavior or compare all four wheels simultaneously.
K factors (thermal gain coefficients, k = ΔP/ΔT) are automatically calculated via linear regression across your processed runs, with R² goodness-of-fit displayed so you can assess reliability. These K factors feed directly into the integrated Pressure Calculator, which uses the verified OptimumG ideal gas law algorithm to convert between hot and cold pressures — accounting for tire temperature, atmospheric pressure, and per-wheel averaging with quadratic-weighted combined temperature.
Different telemetry software packages export data in different CSV formats — column names, separators, header structures, and file organization all vary. Rather than requiring you to manually select the correct format, the app automatically identifies which telemetry software produced your data.
When you upload a ZIP archive, the system samples multiple files inside and analyzes their content patterns. It recognizes WinTax4, WinDarab, and MoTeC formats by examining column headers, separator characters, and file structure. A majority vote across sampled files determines the format, with a confidence percentage displayed so you know how certain the detection is.
If the auto-detection doesn't match your setup (for example, with a custom export template), you can override it manually. The interface provides visual feedback showing both the detected format and your selected format, so you always know what parser will be used for processing.
Every prediction you calculate can be saved to a persistent history with rich metadata: driver name, event, session type, tire condition (new/used), and free-text notes. This creates a searchable log of every setup decision your team has made, organized per vehicle/tire pair.
The history is filterable by track, driver, date range, and event type, making it easy to find previous setups for a specific circuit or compare how different drivers' predictions performed. Entries are color-coded by status: amber for unverified predictions, green for verified ones, purple for manually entered data, and emerald for plateau auto-saved baselines.
After a session, you can verify predictions by adding the actual hot pressures measured on track. The system compares your predicted cold pressures and expected hot pressures against real-world results, giving you immediate feedback on prediction accuracy. Over time, this verification loop helps you understand how reliable your data is and where to focus on improving data quality.
During a run, tire pressures build up from cold to hot and eventually stabilize — they reach a plateau. The app automatically detects when this stabilization has occurred, specifically when pressures have reached 97% of the total expected build-up. Runs that reach this plateau are flagged as "Plateau Reached" in a dedicated section of the Data tab.
Stabilized runs are particularly valuable because they represent the actual steady-state hot pressures for a given set of conditions — no extrapolation needed. The app auto-saves these runs to your Prediction History as verified baselines, tagged with an emerald badge to distinguish them from manual predictions and user-entered data.
This feature is especially useful for building a reliable database of pressure baselines without any manual measurement or entry. Just process your telemetry data, and the system automatically identifies and preserves the most informative runs. The plateau detection uses selective propagation: bleed events invalidate the plateau status for all subsequent laps, while abort events only affect the specific aborted lap.
When you have data from multiple circuits, the Track Relations Matrix reveals how pressure build-up patterns relate between them. Rather than treating each track independently, this feature generates a matrix of cross-track coefficients that quantify the relationship between any two circuits in your database.
This is more powerful than simple track similarity matching. Instead of comparing track characteristics (corner count, length), the matrix is built from actual pressure data — it captures how tires behave differently at each circuit based on your real-world measurements. The interpolation uses bilinear interpolation within the data range for smooth transitions, and linear extrapolation outside it for reasonable estimates at the boundaries.
Matrices can be exported and imported as JSON files, making it easy to share cross-track knowledge within your team or transfer it between seasons. This is particularly valuable when arriving at a track where you have limited data — the matrix can leverage rich data from other circuits to provide a much better starting point than a cold guess.
Common questions about race car tire pressure prediction
See how data-driven pressure predictions can improve your racing consistency
Request a Demo