Best lefty off all time? My vote is Koufax!

[Snowman]

Originally Posted by BobC

Don't disagree with you at all, just throwing out different ways to maybe look at and interpret things out there. And I just wanted to make sure you didn't take me the wrong way in making friendly banter and conversation. LOL

And I get the thinking about how the 5 inning games nowadays change the overall perspective of WINs. But, would you agree or disagree that even if a starting pitcher only goes 5 - 6 innings anymore, how well they pitched and the situation when they left will generally still have a dramatic impact on the outcome of that game, and the decisions and choices of their manager, coaches, and teammates in finally deciding who wins? I'm wondering if the impact of shortened appearances by starting pitchers in the modern game on the final outcomes of their games started isn't being discounted too greatly? Problem is, this is one of those types of questions that there are no statistics for.

Ignoring your personal attacks... I'd be remiss if I didn't point out that there ARE statistics that answer this question though. This is precisely what the entire field of statistics was developed for. The entire point of the mathematical discipline of statistics is to be able to make probabilistic estimates about the outcomes of future events using whatever data we have available. The more skilled you are as a statistician, the more accurate your predictions are. As far as your question goes about how much impact a SP has over final outcomes depending on how many innings he pitches, I assure you this exact problem is well understood. In fact it is extremely well understood. It is by far, the single most important factor in the models I build for betting on baseball games. It is also the single most important factor that the sports book handicappers use in their models when they set the betting lines. There are many other factors at play, but at the end of the day that entire industry is about predicting the outcomes of future events with data and statistical theory. And the casinos are pretty damn good at making predictions.

This is also how and why the entire field of sabermetrics was developed. People wanted to bet on baseball games but they quickly realized that the standard statistics that have been used for decades were not very useful for making predictions with because many of those stats are highly subject to luck. So they engineered new statistics that account for factors outside of an athlete's control and that focus in on what they actually have power over. The aspects of their game that are within an athlete's control are the only factors that have predictive power with respect to how well (or how poorly) they will perform in the future. Any statistic that cannot accurately predict future performance is a poor choice for evaluating one's skill level. Knowing that someone is hitting 0.375 at the all-star break tells us very little about how well he will hit for the rest of the season despite it being a seemingly large sample size of 350 at bats. A deceiving statistic like batting average is another great candidate for paving the way for another heated debate between a regular baseball fan and a statistician. One could ask "who is the best hitter this season?" and the casual fan will point to the guy with the 0.375 AVG, but the statistician looks deeper and points out that he benefited from having a 0.430 BABIP while player B is hitting 0.369 with a 0.300 BABIP. In this case, player B would be the clearly better hitter despite having the lower batting average since BABIP is useful for understanding how much of a role luck played in their performances.

People keep talking about wins here as ultimately being the only thing that matters. I agree. Winning games is what matters most. That's why we statisticians use Wins as the dependent (or target) variable in our predictive models. But the difference is that you guys seem to be conflating the "wins" statistic that is awarded to a pitcher with the actual wins and losses which can only be attributed to the teams. These are not the same thing. A pitcher cannot win a game. Assigning them "wins" and "losses" has always been a bad measure of performance. Not just in the modern era. And it turns out that a pitcher's win-loss record is actually an extremely poor predictor of a team's likelihood of winning a game. And furthermore that in the presence of other statistics, it is in fact not predictive at all of their likelihood of winning a game. This is why it is a poor measure of performance. It tells you nothing at all about how well they pitched or are likely to pitch in the next game. It only tells you what the outcome was of a set of prior games. If you want to know how "good" a pitcher (or hitter) is, then you have to look at statistics that only they can control. Otherwise, you're looking at how lucky or unlucky they got rather than how well they performed. This is the job of the statistician. To find the signal in the noise. To control for factors outside of their control. To remove elements of luck.

I find it humorous that when I posted in the thread about the role of artificial intelligence in grading cards that everyone praised and valued my inputs when it seemed to reinforce their views about grading. But when my views are shared here, where they are in conflict with the majority opinion, everyone shits on me.