Although I didn't look really carefully, it looks like the data only took a single lane change into account.
What about weaving into the lane that seems faster, passing several cars, then switching back? It seems like the snaking effect would work to your advantage if you always get in the lane where there is more distance between cars. Then again, maybe it's perception.