Randomness. Just like I can replay a quarter of a game 10 times and sometimes getting 5 seconds to throw and sometimes running for my life from the start.
Randomness. Granted, randomness will be there even if ratings have superb differences in effectiveness, but if that randomness doesn't change based on ratings, then it doesn't matter.
The thing is two games, or replaying the same game two times doesn't indicate much. Now if testing those two games with all CPU and HUM sliders at 50 consistently showed a difference between those two games, then that would show more of a ratings influence (sample size large enough to decrease the chance that things are just the result of randomness, but instead are the presence of other factors).
My view is it's mostly randomness. Just like if you played a baseball game and all hitters had the same attributes. Some of them will end up with different stats than others just from randomness.
OL seems similar. Perhaps ratings matter a little, but overall it just feels the same exact way.