by Nathan Ensley, graduate student
The green flag waves. The engines roar. And the chase begins. The difference between victory and defeat is measured in thousandths of a second, and teams do everything to close that gap.
In the motorsports world, innovation is key. The innovators in this field have long been the engineers. Engineers design, build, and test these race cars, seeking optimizations that can save fractions of a second. While the technical battle persists, a new battlefront has developed over the last decade: the data analytics battle.
Data has always played a crucial role in modern motorsports. Over one hundred sensors collect data on all aspects of car performance from throttle and braking inputs to fuel load and tire pressure. Engineers use these sensors in real-time to evaluate car performance and make strategy decisions. With so much available data, there is immense potential for data scientists to pioneer new advanced analytics that transform the sport.
Background
One of these data-driven pioneers is Connor Bryson, a recent IU graduate and data scientist for Andretti Global’s INDYCAR team. In the heart of INDYCAR’s busy summer season, Bryson tasked me with analyzing past pit stop data. The goal – find where Andretti ranked against other teams and why some crews performed better than others. While most teams prioritize on-track performance, Andretti could gain a competitive advantage by maximizing pit road efficiency using analytics.
A pit stop is split into two parts: the tires and the fuel. The fuel time remains mostly consistent, with minor fluctuations depending on stint length. The tire stage is much more variable, with most time gained and lost in this stage. Before the tires can be removed, the Air Jack needs to use its wand to lift the car. Once the car is lifted, the wheel changers remove the old tire, add a new tire to the axle, and use their wheel gun to tighten that new tire. Once all four tires are complete, the air jack can drop the car.
For this project, the Andretti pit stop coach provided an anonymized dataset of over 300 pit stops from six different crews. For each stop, the times at which each crew member performed critical actions are listed, as collected from frame-by-frame video analysis. Additionally, I also collected roughly 1,400 pit stop times (Pit In to Pit Out) from all 26 INDYCAR full-time entries. To account for varying pit road lengths, an adjusted pit stop time – excluding drive-through time – was used for consistent track-to-track comparison.
Insights & Key Takeaways
Using percentile ranks, I evaluated Andretti’s pit crew relative to their competition. Their 10th percentile pit stop time (reflecting pure speed) is half a second behind top teams. However, their consistency is one of the best, with just 1.25 seconds separating their 25th and 75th percentile times. Overall, these results indicate that Andretti has the 4th best INDYCAR pit team, behind Team Penske, Ganassi, and McLaren, who have faster times across all percentile ranks and represent the series standard.
The air jack needs to wait until all four tires are completed, meaning the slowest tire position generally dictates the pit stop time. Across all stops in the Andretti dataset, the Outside Rear tire changer is the final tire 45.7% of the time. This occurs because the Outside Rear guy has to run around the car to his position, while the other tire changers are already in place and have about a 0.4-second head start.
With there being differences between the wheel changing positions, I created a new key metric to assess their overall performance called Time Above Replacement Wheel Changer, or TARWC. TARWC measures the difference between a wheel changer’s median time for each tire changing phase compared to the expected median time for that position. A positive TARWC indicates that a wheel changer is faster than a replacement-level wheel changer.
Bryson prioritized minimizing mistakes over outright speed. Thus, I identified when major mistakes occurred and who made them. On a pit stop under caution – when on-track racing stops – it takes a pit stop difference of 0.6 seconds for a trailing car to pass a leading car on pit road. Thus, a 2-second time loss would result in three lost positions, a major loss in track position, if all had comparable average pit stop times. For that reason, I classified a segment time two seconds longer than expected as a major mistake. 72% of major mistakes for Andretti occurred during the tire tightening phase, indicating wheel changers with good wheel gun skills reduce major time loss.
The TARWC effectively captured which tire changers were more likely to make a major mistake. Those who had a positive TARWC averaged 0.33 major mistakes, while those with a negative TARWC averaged 2 major mistakes.
The Air Jack is the most critical position on a pit crew. Slow plug-ins or early drops – releasing the car before all tires are complete – can cost valuable seconds. Conversely, delayed reaction after tire completion can also cost time, with one Andretti car losing 0.4 seconds on average to other cars from a slow Air Jack response.
To identify areas of improvement for each car, I developed a tire changer dashboard. Primarily, this dashboard displays which phases each changer is gaining or losing time and indicates which tire is most frequently the final tire completed. For the 82 car, shown in the dashboard below, the Inside Rear was the weakest tire, with four of the five largest final tire differentials. This time loss was primarily in the wheel gun phases of removing the tire and tightening the new tire, resulting in the Inside Rear being the final tire 49% of the time.
Conclusion
This Andretti pit stop project revealed opportunities to close the performance gap to top teams. Retaining top wheel changers (who excel in tire tightening) reduces mistakes, while the time above replacement wheel changer (TARWC) metric provides an objective comparison across different positions. The developed dashboard quickly displays improvement areas, giving the coach a real-time tool to optimize pit crew performance, where every tenth of a second counts.
Leave a Reply