Stolen Goods: Comparing The Forecasting Systems (2007-2010)

Tom Tango, courtesy of Brian Cartwright's tedious annual projection cataloging, just analyzed the various popular projections systems out there for accuracy in various capacities (best at forecasting overall, best at forecasting rookies, etc.). The results are quite interesting. CHONE, my forecaster of choice over the past few seasons, faired the best overall, while Oliver came in a close second, while also being the best system for forecasting rookie hitters. PECOTA was the weakest overall, thanks to its horrible 2009 showing, but was on par with ZiPS otherwise (yet another great reason not to pay for PECOTA -- ZiPS is free).

With CHONE defunct now (Sean Smith, the creator, works for a major league front office these days), Oliver would seem to reign supreme as the best available forecaster on the market. Though all of these systems are generally interchangeable overall, Oliver does the best at forecasting young players based on minor league production, which gives it a clear advantage over the rest of the systems. As Tango notes, players with higher sample sizes of major league production tend to have the most similar projections between systems.

You can purchase access to Oliver via THT Forecasts for $14.95 for the entire 2011 season. I personally use Oliver and can attest to its awesomeness, but if you do not believe me, just click the link to Tom Tango's blog (above) and read about its accuracy for yourself.