Also known as the Pythagorean Expectation, a formula concocted by Bill James, to estimate how many games a team should win based on the number of runs the team scores versus the number of runs the team allows over the course of a season.

How good is this formula, really?

Recently, I had way too much time on my hands for a couple of days, so I compiled some stats from one of my worlds for the total number of runs scored and total number of runs allowed by each franchise over the history of the world.  Probably not very meaningful, but I was looking for a way to pass some time.  After I compiled this information, I then decided to calculate the expected win percentage for each franchise over the history of the world and compare it to actual wins.  Was looking to see who the over-achievers were, and who the under-achievers were.

The results were kind of surprising.

Most surprising was the fact that the one team that had (a) won the most games over the history of the world, (b) scored the most runs, and (c) allowed the fewest runs, came in as the biggest under-achiever of all 32 franchises, clocking in at (-67) wins below expectations.

In fact, of the five winningest franchises in world history, four of them came in among the five biggest under-achievers.  The fifth one came in as a slight over-achiever, at +9 wins above expectations.

So this makes be wonder: is my approach flawed in my looking at too large of a sample size (19 seasons) with teams that may have gone through multiple ownership changes and many cycles of up and down seasons?  Or is there some sort of inherent flaw in the theory behind the formula itself that means that any correlation between what the formula "predicts" and what happens in reality is little more than coincidence?

Any model that uses a curve predicts the "middle" better than the "ends"

Apply the real life example I know - insurance.

More specifically, home insurance. Your suburban home is going to be picked up by any insurance company, but the ultra rich dude probably is going to need a specialty insurer because he gives the actuaries in your garden variety insurance company heart attacks.

ie. his or her home is at the end of the curve.

67 wins low out of 19 seasons at an assumed average .550 winning percentage (1693 total wins) is only 4% off.  I would say that is pretty close...

I'm guessing his formula ends up over-valuing good teams and under-valuing bad teams (good teams always fall short and bad teams always win more) but for a majority of teams it's pretty close.

pretty much what death said

The good teams are bitchslapping the bad teams by 6 and cranking up their EXP%.   The bad teams are sneaking by the good teams  by 1 and winning more than their EXP%.

I don't think that's very surprising.

All good points.  I guess I expected to see something flatter with what I thought was significantly large sample size.  Instead I did indeed get something more like a bell curve, with a very noticeable pattern at both ends (good teams under-achieving, poor teams over-achieving).  Wasn't really expecting that.

Well you should have asked us first.

Nothing like spending a couple of days figuring out what everyone else sees as obvoius.

Yea, dumbass.

Sorry, but I've got to go. I'm doing some research on the best way to get gasoline from the nozzle into my car.

They were REALLY slow days.

I'm guessing teams abandoned mid-season and playing tired players will alter the results a bit.   

With the new abandonment replacement rules, are we expecting more threads like this from tec?  Lots of time on his hands now...

I noticed that my 9 starters are getting more at bats than my bench players. Tec, can you do some analysis on the reason why?

I'll get on that as soon as I figure out why there's such a strong affinity between wins and scoring more runs than your opponent. I find that very odd.

This sounds like typical commentator "insights"..."let me tell you, everytime this team scores more runs than the opponent, they will win games!"