Provide the past predictions over a series of times at fixed intervals. Provide an objective standard for saying that your algorithm works.
That doesn't prove anything, I could claim my "algorithm" "predicted" 240 on the 9th, 120 on the 10th, and 60 on the the 12th.
I say provide us with price predictions for the next 7 days. Yesterday's "prediction" with the crude growth line drawn over the increase from 60 to 120 transposed sideways a little has so far failed to pan out.
Agreed that future predictions means he can't simply lie. That wasn't my presumption. It seems plausible he has something he calls an algorithm. I suspect, like some technical analysis people I've met. He's just fooling himself. However the problem with future predictions is that they themselves are information and could affect the outcome (e.g. knowing that $200 is going to be hit tomorrow could change what a large group of people do today)
You could mitigate this buy using "sealed envelop" predictions.