Baudib
04-19-2002, 05:35 PM
Originally posted by SmedIndy
I'm using whatever Lee has in his database. He doesn't use the adjusted formula, which he says James changed for no good reason.
I'm backing up and walking away before I angry up the blood. I suggest you do the same.
My blood never gets angry.
Anyway, Bill James changed the formula because it was wrong and inaccurate.
It was based on a seemingly logical assumption, but if you work through it, you'll see it doesn't work.
That assumption is that if you can accurately predict a team's runs by virtue of their batting stats, then you can apply this formula to a player's stats.
This sounds great, except that runs created is a non-linear formula.
Linear weights, by Pete Palmer, assumes that a homer is a homer, that it's going to be the same value no matter what. Runs created says that the value of a homer is dependent on how many runners a team has on base.
If you have a team that has a lot of baserunners, but little power, adding home runs to the mix is very powerful. Each homer could be worth, say, 1.45 runs.
But if you have a team with a low OBP and lots of power, each added homer has diminished value. The homers for this team might be worth 1.3 runs.
While this is fine for a team, it causes horrendous distortions in trying to calculate individual player's stats.
This is because it assumes that each player is a team unto himself.
The formula says, "Wow, this team called Barry Bonds has a .515 OBP, which interacts with an .863 slugging percentage, which yields a huge number of runs."
The reality is that Bonds' OBP and SLG NEVER interact. Bonds' OBP interacts with Jeff Kent's SLG, and his SLG percentage interacts with Rich Aurilia and Marvin Benard's OBP.
Thus, under the old runs created formula, hitters with both high OBP and SLG would overshoot their real totals, the better the hitter, the worse the calculation becomes.
The formula also suffers in reliability when calculating totals for teams that score many runs...the likely cause is that it overestimates the impact of home runs.
I'm using whatever Lee has in his database. He doesn't use the adjusted formula, which he says James changed for no good reason.
I'm backing up and walking away before I angry up the blood. I suggest you do the same.
My blood never gets angry.
Anyway, Bill James changed the formula because it was wrong and inaccurate.
It was based on a seemingly logical assumption, but if you work through it, you'll see it doesn't work.
That assumption is that if you can accurately predict a team's runs by virtue of their batting stats, then you can apply this formula to a player's stats.
This sounds great, except that runs created is a non-linear formula.
Linear weights, by Pete Palmer, assumes that a homer is a homer, that it's going to be the same value no matter what. Runs created says that the value of a homer is dependent on how many runners a team has on base.
If you have a team that has a lot of baserunners, but little power, adding home runs to the mix is very powerful. Each homer could be worth, say, 1.45 runs.
But if you have a team with a low OBP and lots of power, each added homer has diminished value. The homers for this team might be worth 1.3 runs.
While this is fine for a team, it causes horrendous distortions in trying to calculate individual player's stats.
This is because it assumes that each player is a team unto himself.
The formula says, "Wow, this team called Barry Bonds has a .515 OBP, which interacts with an .863 slugging percentage, which yields a huge number of runs."
The reality is that Bonds' OBP and SLG NEVER interact. Bonds' OBP interacts with Jeff Kent's SLG, and his SLG percentage interacts with Rich Aurilia and Marvin Benard's OBP.
Thus, under the old runs created formula, hitters with both high OBP and SLG would overshoot their real totals, the better the hitter, the worse the calculation becomes.
The formula also suffers in reliability when calculating totals for teams that score many runs...the likely cause is that it overestimates the impact of home runs.