If you want to check it on practice, try this: make a thousand rolls strictly following some sophisticated strategy, and then make a thousand completely random rolls after that, and compare the results. Only one condition: your bankroll must be large enough for making those rolls in both cases.
I don't think we need large bankroll for making random rolls because if you're rolling in random then you never need to double your bets which means you may not need big bankroll. ~
Well, it's true that for random rolls you will probably need a smaller bankroll than for martingale, but we are talking about thousand rolls here, and even if your bet size is $0.01 you still need a $10 bankroll for the experiment, right? Also by "random rolls " I meant random bet size too. So, if you don't want your experiment ruined by zero bankroll after 980 rolls, you better be prepared.
Actually, if your bankroll happened to be zero before the experiment ended, you could just deposit more and continue, but then, although valid, the experiment wouldn't be looking so pretty, aesthetically wise.
~ Still, I agree that results from strategy based rolls and random rolls may not show big differences. Because, randomness of gambling may or may not work in favor of us. Even it will work in favor of us, we cannot expect that to be happening all the days and moreover, due to the presence of house-edge, gamblers get lesser chances than houses to be favorable in the environment of randomness.
As I'm always saying, don't overestimate the impact of house-edge on your performance. It's not that big as people might think.