Twins Tidbit: How Predictive is a Pitcher's First Start?

3 hours ago, Otaknam said:

Obviously there is no correlation between the first start and career success. Not sure why you raised the question. 

Obvious or not, this study seems skewed by looking only at, as the writer stated it, "notable Twins pitchers."  Do the study on first MLB career starts, period, and the data might show a stronger correlation.  Maybe low quality first career starts are made mostly by pitchers who ultimately don't pan out?  In that case Berrios might look like more of an outlier, while luminaries like JC Romero would cluster at the lower left. 

That "other" Romero, with game score 34 in his first major league start, was pretty terrible for two dreary years until he was converted to full-time reliever (which could be another variable to try to account for).  Does anyone remember Brad Thomas?  Me neither, but his 2001 debut led to a 30 game score and presaged a lifetime WAR in the negative numbers.  Leaving guys like them off the chart might lead one to think those 30ish first games are the rarity.

I think it's also a mistake to limit the Y-axis to be the portion of a pitcher's career spent with the Twins.  Kyle Lohse's value is a good deal higher than the 6.5 WAR he earned as a Twin.  Although, really, achieving even 5 WAR is a pretty good accomplishment by itself, as would be demonstrated better if the chart showed every Twin who was allowed to try.  The "notable" pitchers would stand out without having to be sifted through beforehand.

I'm just not much in favor of throwing out data that doesn't happen to fit a narrative.  (Except perhaps whenever it's my narrative. )