I've been playing around with Seong's Post-Earnings Drift Trading Strategy with Estimize (PEAD) algo and looked at incorporating the Wall Street EPS and Revenue numbers (reported by Estimize) as additional data points, but got strange results. I then started to compare the numbers that were being output in the log with what was on the Estimize site and discovered that the vast majority of the time the numbers were incorrect. The Estimize ESP, actual EPS, and even the Estimize revenue and actual revenue numbers all matched up perfectly, but the Wall Street EPS and Wall Stree revenue were wrong more often than not. A little more digging and it appears that the data shown is often from other quarters. Here is a sample log dated 1-24-2014 (I formatted it for easy reading and just made a screenshot of the table):

Screen Shot of Log Output (formatted)

Columns I and J are what's shown in the log and columns K and L are what the numbers should be. I then included notes about where I actually found numbers that correlated to the numbers reported in the log.

I've done a lot of testing and reviewing across many different dates and it's always the same; most of the data is incorrect (although not all). Thinking maybe it was a problem with the free data, I subscribed to the full Estimize data set and it's incorrect even up to the present. I've attached a short backtest of what I ran. I just took Seong's original algo, included the "consensus_wallstreet_eps_free" import and created two additional factors to show the Wall Street EPS and revenue. Everything else is the same.

Any idea what might be going on?

Scott