I tried something similar and was astonished at the amount of variance caused by stoplights and weather. Because of that, it's hard to get a good dataset for co

I tried something similar and was astonished at the amount of variance caused by stoplights and weather. Because of that, it's hard to get a good dataset for comparison. In the end, because of the variance caused by external factors, there was no clear winner in my case, so I now use an app from random.org to flip a coin and follow the results accordingly. That's not because either way is better, but this way if I have a long delay for some reason, I can blame it on a bad coin flip.