Steamer makes many of the same types of adjustments as other “fancy” projection systems: adjusting for park effects, league effects, aging and starting v. relieving.
We have “backtested” a large number different possible forecasts systems, in order to determine the proper ratio of weights for data from recent seasons and the correct amount of regression for each component of performance.
You can observe a lot just by watching.
– Yogi Berra
The best possible forecasts would incorporate all of the information available: play-by-play data, pitch-f/x, minor league scouting reports, injuries histories and, hopefully someday, hit-f/x and field-f/x. We have a long way to go but, bit by bit, we’re using more and more information in our forecasts and the early returns are promising (credit richter). In the two years since we starting using pitcher’s fastball velocities (a crude measure of “stuff”), our pitching projections have paced the field (see here and here).
If you’re interested in finding out a little more, here’s Dash’s presentation from the 2012 Saber Seminar.