As I've mentioned in previous posts, I'm less than enamored about a lot of the products that are available to help get a picture of one's opponents in a less-than-noble way, that being by direct play at the tables. Sharkscope is one of those products, though it's probably at the minor end of the scale, certainly not as offensive to a purist as a PokerHUD or a Pokerbility or other products which try to create an artificial 'live time' advantage for the player.
I made a deep run in the $20K on Poker.com / Carbon Poker a week ago, and in one of those weird moments of satori, I realized something about Sharkscope: Unless the product has a hidden way of accounting for it --- and I don't believe it does --- the ROI numbers it generates for its customers carries an unknown skew. Sharkscope does not, to the best of my knowledge, account for tournaments where an overlay is in effect, therefore making ROI percentages a bit higher over the long run than they otherwise might be.
In the tournament above, I finished fifth, in itself my best cash of the year. But this tournament had a chunk of overlay, somewhere around 25%, and it occurred to me that a player who did nothing but hunt down events with sizable overlays would generate a much better-looking profile on Sharkscope than a similar player who played a lot of non-overlay events. Same player skills would yield different results. Sharkscope allows five searches a day, so I searched on the five highest finishers here besides myself.
You can see that all of them look to be really tough players, but is that the whole truth? It's possible that one or two of them just might be middlin' players who focus only on big-overlay specials --- I'm not saying that it's so, but that it could be. As for me, thinking back to my own tourneys played, I'd guess that my ROI is 2-3% inflated over what it should really be , due to unaccounted overlays. It's not a big deal, but it's a hidden factor nonetheless.
Lesson: If you use a Sharkscope or something like it, use it only in a general sense. Specifics can't be trusted.