By the end of the year it is normally pretty reliable. This early in the season there's still too much noise/not enough sample for it to be the best indicator of future results. Before he got hired by the Rays, Jeff Sullivan wrote an article at FanGraphs examining which of actual 1st half results, pythag (run differential) 1st half results or projected 2nd half results did the best job of predicting actual 2nd half results. https://blogs.fangraphs.com/how-best-to-predict-the-seasons-second-half/ Even after 80 games (we're still only at 53) the projected 2nd half results still did a better job of predicting the actual 2nd half results than either of 1st half actual or pythag results. How are the projected 2nd half results calculated? I understand the other variables, but am a bit confused by how that calculation is done. Is it projected 2nd half results from the preseason?