No, that doesn't work. See https://cs.stackexchange.com/q/602/86141 for an explanation of why it's impossible. If you tried your method with concrete numbers, you'd see that it would always say the delay is perfectly symmetric, even when it actually isn't.