I have a lot of experience with gambling and I do this reguarly. I can verify that in my experience makes you better calibrated. What I've had success with is to generate a probability range before I incorporate the market opinion, then use it to generate another. I find the key in practice is not to define a mathematical distribution but to give a mean prediction and a range that you find plausable which should have a probabiility of around 95%. Often the mean is not centered.