Mark Buehrle?

[KeepTheFaith1229]

He didn't reject it. He was putting it on hold to see that he couldn't get Joba or someone from the yanks etc.

 

I have trouble believing it too. It probably wasn't the entire story.

 

The run statistics thing sounds interesting though, can you elaborate or get this guy to elaborate?

 

I'll ask him next time he's on. He said its brand new and a few teams including the sox, mets and angels are using it this year.

[schillingouttheks]

He didn't reject it. He was putting it on hold to see that he couldn't get Joba or someone from the yanks etc.

 

Right...like getting Joba in a package from the Yanks could possibly make a better package than one composed of Ellsbury, Bowden and Buchholz. I'll re-word my statement then:

 

The fact that Kenny Williams put this offer on hold in the first place is what kills the credibility.

[KeepTheFaith1229]

I don't know why either. Some of the baseball front offices have much different perceptions on prospects though.

[schillingouttheks]

I don't know why either. Some of the baseball front offices have much different perceptions on prospects though.

 

I guess. But it's just foolish for Williams to not have accepted any offer that included Buchholz. It's pretty much known that Buchholz is one of, if not the one, top pitching prospects in all of baseball.

[KeepTheFaith1229]

The run statistics thing sounds interesting though' date=' can you elaborate or get this guy to elaborate?[/quote']

 

Ok, he can't tell me everything, but he could tell me this much:

 

each hit is worth a point...double 2, single 1 etc

homer is worth a certain amount extra per base

SB is worth a certain amount of a base

a walk is worth a percent of a single

CS is worth a total out, as if you were never on at all

all these with some linear weights gives it

 

the highest in the mlb this season is barry bonds...juan pierre is one of the worst...JD drew is actually better than plenty of people you might not have thought he was

[schillingouttheks]

the highest in the mlb this season is barry bonds...juan pierre is one of the worst...JD drew is actually better than plenty of people you might not have thought he was

 

This is because of his above-average OBP. He hasn't been a solid player in some peoples' eyes because of his misleading average early on. He's actually gotten on base a lot more than people would think. Granted, he hasn't done his job as a 5-hole hitter because Fenway has sapped some of his power out to right-centerfield, but I believe that he's been doing a lot better than some people would believe.

[ksushi]

He didn't reject it. He was putting it on hold to see that he couldn't get Joba or someone from the yanks etc.

 

I have trouble believing it too. It probably wasn't the entire story.

 

 

 

I'll ask him next time he's on. He said its brand new and a few teams including the sox, mets and angels are using it this year.

 

Bucholz, Ellsbury and Bowden is better than any three prospects the yanks could muster up. Guarenteed. Unless you're talking about Hughes but thats a different story. The yankees wouldn't offer hughes in the same way the Sox wouldn't offer anyone Papelbon.

[jacksonianmarch]

Bucholz' date=' Ellsbury and Bowden is better than any three prospects the yanks could muster up. Guarenteed. Unless you're talking about Hughes but thats a different story. The yankees wouldn't offer hughes in the same way the Sox wouldn't offer anyone Papelbon.[/quote']

 

We can beat Bowden soundly, but Buchholz and Ells are further advanced than the guys on our farm right now. You are correct though, Hughes surpasses them all.

[ORS]

The run system isn't new, it's called linear weights. Each possible event at the plate has an impact on runs scored and the weighting of each event is determined from the team stats for the league each year.

 

The equation for each team looks like this:

 

1B + 2B + 3B + HR + BB + HBP + SB + SF + SH - OUTS - HIDP = Runs

 

That's 11 independent variables going into 1 dependent variable. You have enough teams in each league (AL and NL - since the rules are different) to determine the weights each year for the league by setting up a system of equations solving for each event's weighting. The concept is about 2 or 3 years old, and it would shock me if every team's statistical department wasn't using it.

[KeepTheFaith1229]

The run system isn't new, it's called linear weights. Each possible event at the plate has an impact on runs scored and the weighting of each event is determined from the team stats for the league each year.

 

The equation for each team looks like this:

 

1B + 2B + 3B + HR + BB + HBP + SB + SF + SH - OUTS - HIDP = Runs

 

That's 11 independent variables going into 1 dependent variable. You have enough teams in each league (AL and NL - since the rules are different) to determine the weights each year for the league by setting up a system of equations solving for each event's weighting. The concept is about 2 or 3 years old, and it would shock me if every team's statistical department wasn't using it.

 

He said that's not the same forumula. The formula takes much more things into account, like for example, a walk is worth like 73% of a single because a walk can't advance anyone two bases or score anyone unless the bases are loaded.

[ORS]

He said that's not the same forumula. The formula takes much more things into account' date=' like for example, a walk is worth like 73% of a single because a walk can't advance anyone two bases or score anyone unless the bases are loaded.[/quote']

I know. The equation I show is simplified to show what the variables are. Mathematically, each event is an X. Since there are 11, there would be X1 through X11 (I would subscript those numbers if I could). By solving the system of equations, you get the relative weight of each X. So, in solving it, you would get singles with more weight than walks, and so on. This is nothing new, and I think you are having the wool pulled over your eyes.

[example1]

I know. The equation I show is simplified to show what the variables are. Mathematically' date=' each event is an X. Since there are 11, there would be X1 through X11 (I would subscript those numbers if I could). By solving the system of equations, you get the relative weight of each X. So, in solving it, you would get singles with more weight than walks, and so on. This is nothing new, and I think you are having the wool pulled over your eyes.[/quote']

 

So it is basically a linear regression with runs as the DV and any factor that could possibly lead to a run as an IV? They could probably take a few years' averages of those IVs and get a pretty good idea of the basic value of a walk or a single, or double, etc., and then rank players' values based on those findings. That is a basic statistical process that they use in social sciences all the time.

[ORS]

Absolutely. It's a linear regression, and typically they will use 5-year weighted values to reduce the impact of variance.

[KeepTheFaith1229]

ok, heres what he said..dont kill the messenger

 

"the point is there is no such thing as a double or a single or a triple, because those are circumstancial..you cant place a weight on a double when it is circumstancial..if you hit a ground ball to LF, and he slips, it would be a 'double'..like mark beurhle, who is succeeding so much more on circumstance this year.."

[ORS]

That's not what was posted above. Singles get one point, doubles get two. And the notion that every double is just circumstantial over a single is crazy. Sure, there are plenty that fall into that category, but for a whole team over a whole season (ie a large sample), I'll bet money there's significant correlation to quality of contact in batted ball types to XBH's.

[example1]

Absolutely. It's a linear regression' date=' and typically they will use 5-year weighted values to reduce the impact of variance.[/quote']

 

that's a big sample size. My guess is that the specific weights on the IVs are pretty accurate.

[example1]

ok, heres what he said..dont kill the messenger

 

"the point is there is no such thing as a double or a single or a triple, because those are circumstancial..you cant place a weight on a double when it is circumstancial..if you hit a ground ball to LF, and he slips, it would be a 'double'..like mark beurhle, who is succeeding so much more on circumstance this year.."

 

Your friend is right in that there are a whole lot of circumstancial aspects to the difference between a single and a double, or a double and a triple. There are a large number of "outlier" doubles that should have been singles, or outs, or triples. But--as ORS pointed out--over a season with thousands and thousands and thousands of ABs throughout baseball, I imagine the majority of doubles are balls hit to the gap, or down the line. They aren't flukes.

[ORS]

Your friend is right in that there are a whole lot of circumstancial aspects to the difference between a single and a double' date=' or a double and a triple. There are a large number of "outlier" doubles that should have been singles, or outs, or triples. But--as ORS pointed out--over a season with thousands and thousands and thousands of ABs throughout baseball, I imagine the majority of doubles are balls hit to the gap, or down the line. They aren't flukes.[/quote']

Exactly. Over a whole season, the amount of hits that don't go for the amount of bases one would typically expect tends to cancel out. In other words, while there are some doubles that aren't a result of good contact, there are also some smashed singles where the OF made a good play (or it was at someone but would have gotten through if hit at a gap).

[ORS]

I will note this. I have seen it suggested that triples should just be counted as doubles as they are a rare event for the typical player and only frequently happen for the speedy. I can accept that. However, speed is a tool, and it certainly leads to runs, so if the goal is to find the most realistic model possible - which I think Lwts does - then 3Bs should be a separate IV.

[KeepTheFaith1229]

eh, i really dont know. you'd probably have to ask him for a complete explanation of it but he said it just came out a few months ago.

[example1]

eh' date=' i really dont know. you'd probably have to ask him for a complete explanation of it but he said it just came out a few months ago.[/quote']

 

I wouldn't be suprised if the Sox feel that they have the most accurate linear regression going, and went so far as to "patent" it, or at least give it a label other than linear weights. There really isn't a statistical tool other than linear regression when trying to figure out the quantitative value of doing something (i.e., applying an independent variable) to the results of something else (dependent variable).

 

One can measure the impact of college education (independent variable) on income (dependent variable), by setting up situations where a huge simple random sample of people with a degree against a huge simple random sample of people without one in yearly monetary terms. It isn't perfect in any particular case (there are many people without college degrees who are successful), but when you do it time and time again, thousands of times, all things being equal one is better off with a degree than not. It doesn't say that a degree causes financial success, it says that a degree correlates with success. By either controlling for other factors (making sure that all 10,000 participants are from the same relative socio-economic group, to control for background as a factor) or increasing the sample size the outliers can be removed or largely disregarded.

 

Similarly, baseball provides so many examples it is the perfect place for such statistical analysis and the Sox have some of the best. It wouldn't shock me at all if the brains in the Sox possess the best regression formula available, have a special name for it, and are improving on it regularly. I just don't see any way they could apply a point value for various game outcomes unless it is some form of a linear regression. The complexity of that regression could be (and likely is) much larger than most of us could imagine. There are a lot of factors that could potentially influence a game, the only problem is measuring them.

[example1]

I will note this. I have seen it suggested that triples should just be counted as doubles as they are a rare event for the typical player and only frequently happen for the speedy. I can accept that. However' date=' speed is a tool, and it certainly leads to runs, so if the goal is to find the most realistic model possible - which I think Lwts does - then 3Bs should be a separate IV.[/quote']

 

If it is a possible outcome in the game of baseball when the ball is in play, it should be an IV.