This test is OK but I leaves too many variables up in the air. As nice as a track test it, too many thing can be atributed to some other thing (driver, motor, tires, setup, etc...).
To truly get an accurate resluts you would need to setup a test that removes variables like the driver and the car but still keeps and accurate "on track" performance. If I had the time (and money) I would setup a test like this:
1. Get battery packs, all new from one matcher, SMC or Orion, or whomever. 3 packs of one type, 1 packs low numbers (pic one stat to go off of, say voltage), 1 mid-range, 1 high numbers.
2. Charge them up at the same rate (say 6A, with 25mV peak)
3. Using a setup that could dischage the packs at a variable rate (100% for 3s (down the straight), 75% for 2s (throught the sweeper), 50% for 2s (hair pin), 50% for 5s (Chicane) , 100% for 2s (sweeper), 50% for 3s (on-to straight), then repeat). Do this discharge procedure 10min after full change.
4. Record the results and plot using excel (do three or Four cycles) and compare
If I was back in school this would be an excellent engineering undergrad project. Using a motorola micro controller this could be done. But I'm a woking stiff now, no time for this.