Nick claims he’ll wright up a race summary and he has all the pretty photographs so if you don’t want spoilers then wait for his post. Also, if you don’t like numbers then this is probably a good post to skip or jump to the tl;dr section.
Last weekend’s race at Thunderhill Raceway Park was a great race with a very competitive field. Sorted by fastest lap were were the 24th fastest car of the 44 that started the race on Friday. In spite of that we came in third place for Friday’s race. A great result! So what contributed to our success?
I don’t have any racecapture data yet (leo has to pick up the race car and pull the SD card before I can get my hands on it) so I looked at our published laptimes (available on google drive).
Above is a scatter plot of every one of our lap times (car 4, green) vs every one of the laps by the car that placed right behind us (car 201, red). Assuming a fast in-lap is 140 seconds (2:20) then any lap shown that takes longer than 440 seconds is likely a fuel stop. Immediately you can see that our fuel stops were much closer to the minimum. [The last one even came in 3 seconds faster than minimum due to the fact that our timer fell off so we got released a couple slightly early.] How much faster were our stops? The difference comes out to a total of roughly 368 seconds. That’s over two and a half laps at this track. Add to that car 201′s fourth stop, which could have been either a driver change or most likely a penalty stop, we gained 4-5 laps on stops (or lack there of) alone.
So what about our pace? I started out saying that our pace as measured by fastest lap of the day was 24th fastest and when compared to the the team with the next best result (car 201, as above) we were nearly 5.5 seconds slower. Thats huge! carried over every lap of a 160 lap (~7hr) race that comes out to over a 6 lap deficit. Since we came out ahead of them clearly that difference over represents team’s overall speed difference.
It’s hard to measure a teams actual pace from the lap times since they include flags, stops, and traffic but I have a couple candidates metrics that I believe are closer than fastest lap.
The first is the average lap time of all laps faster than 2:40, which would exclude most serious incidents and stops. By this metric we averaged 2:45.8 laps while car 201 averaged 2:24.3, only a second and a half difference or over an entire race a lap and a half (I should also point out that this means you should only add $10 past $500 to your value if you know it’ll get you a full second per lap). Below is the same data as above with the y axis zoomed to only include laps counted in the average. You can see we were slower but not by 5 seconds.
The second metric is the 75 percentile lap for each team. This one has the built in assumption that roughly half of laps are in traffic or otherwise impaired. If that’s true then the 75th percentile gives the median clean lap time. It can also just be taken as a purely synthetic metric and I’m inclined to believe that it’s a better metric for the car’s performance rather than the team’s. Either way, our 75 percentile lap was 2:23.2 while car 201′s was 2:20.3–three seconds (or about 3 laps) different.
The fact that these two metrics disagree about the speed gap between teams implies that we may have been more consistent. Even know our car was most likely slower, the fact that we stayed closer to our car’s best speed meant that rather than being on average five seconds slower per lap we were only slower by 1.5 second.
Visualizing the lap data as a histogram we can see that our lap times (green) were more closely clustered around our best time, while team 201 had a more. Doing the math our first moment was in fact less than theirs (i.e. the standard deviation for our laps was 3.6 seconds while team 201 had a standard deviation of 5.4 seconds.).
tl;dr version & conclusion
We did well because we had awesome pit stops, no penalties or extra driving stints, and consistent laps.
When Leo pulls the SD card and I get telemetry data I’ll be sure to make a follow up post with at least some fun stats like highest entry speed into turn 1 and, if I have time, some analysis.
While writing this I hacked together some python code to analyze the data for me. I used some sciency libraries but all are included in the free anaconda python distribution. I’ve posted the code as a github gist and it can be modified fairly easily to view any two cars from any race. I give no guarantees that the code is easy to read, elegant, will work, or won’t damage your computer or brain, so take it as-is and if it helps you then great, maybe give it a star on github so I know it’s worthwhile to publish these code snippets.