Monday, February 8, 2010

Rox Talk - Runs Score Part I

Few years back I was wondering with all of these new stats whether one, specifically On Base Percentage, was worthy of runs score. I mean baseball is about scoring runs. If a master stat doesn't relate to runs score what does it really mean? So I started easily and plotted our Rox runs scored versus batting average, on base percentage (OBP), and on base plus slugging (OPS).

Guess you'd figure that on base plus slugging would correlate the best but OBP works pretty good too. Since OPS seemed to correlate the best with scoring runs, what about the rest of the league (see below)? Well there you have it...on the league level it seems that OPS is the big winner. Funny thing to me is the data below it makes the OBP seems a whole lot more correlated than OPS...hmmmm the miracles of excel!

OK so OPS correlates to runs scored. Well another thing that makes me wonder. Runs scored versus runs allowed...which one matters more? I have always been fascinated by the simple concept behind the Pythagorean run theory (read about it here). Thus if you plot runs scored divided by runs allow against wins you get the graph below.

The graph shows data from 2000 - 2009 seasons. Again this is probably an obvious point but when you score more runs than your opponent you tend to win more games ;-). OK so back to my second thought, is there any way to determine which is more important; a runs scored or a run prevented? Think about it...if a team wins 1 - 0 what is more important the one run scored or the nine innings of preventing a run? I have no idea just something to ponder but I went ahead and tried in my own way to see what plotting the data might suggest. Based on ten years of data the average runs scored was 771 runs. I then removed this number of runs from every team's total and then plotted this against number of wins. I also did this for runs allowed. The two graphs produced looked like this:

Pretty much two scatter shotgun pattern of data points. Both suggest a week correlation (0 being no correlation and 1 showing a correlation). The runs scored had a 0.35 and the runs against 0.45. On this basis could you suggest that the higher correlation in runs against might suggest wins are more depended on allowing your opponent less runs? I am not a stat guy, don't pretend to be, and probably the power of excel makes real stat guys cringe when people like me play with numbers looking for patterns that truly don't exist but either way to me it is fun...

No comments: