While our database has only been properly live for four months, we try to qualify the tendency of managers to report positive results early and worse results with a delay. While this is clearly a limited data sample and set, we can at least quantify this bias.
Throughout this period, it has been small, and the the effect has been 12 bps per month or approximately 1.4% per annum. Noteworthy is that the effect always points in the same direction, early performance is almost always better than later, fuller estimate. Over this sample, the backfill and reporting biases approximately cancels out. NilssonHedge will track this bias over time and report occasionally.
Sign up for the underlying database and a user account here (free).