From: tom dunne on
On Mar 3, 1:15 pm, John Kasupski <w2...(a)spamfilter.verizon.net> wrote:

> >They don't affect future events, but they certainly play a hand in
> >predicting future events quite accurately.  Albert Pujols is going to
> >have a great year, yes?
>
> That remains to be seen. He could get drilled in the head with the first pitch
> on Opening Day and miss the rest of the year. He could...well, any number of
> things could happen.
>
> >Why do we think that and why will he continue
> >to bat 3rd for the Cardinals?  
>
> "We" think that (and it goes without saying that I don't, necessarily) because
> it's natural to expect players who have performed at a high level to continue to
> do so. He will continue to bat 3rd for the Cardinals for as long as LaRussa
> feels he deserves to hit there.
>
> But again...players have outlier years, they get hurt (and Pujols is if I'm not
> mistaken coming off of offseason surgery), etc. Nothing is certain.

I don't think I follow what you're saying here. Statistical analysis
lends itself to discussing future probabilities, not certainties. I
don't think any branch of stats claims that, baseball or otherwise.
Regardless, we all use our past experiences and our future
expectations to make decisions. We'd be paralyzed with options if we
didn't.

You say that you don't necessarily think that Albert Pujols will
continue to have a great year and bat 3rd for the Cardinals. That's
not certain, but surely that's your expectation, right? If some of
your life decisions required you to base them upon how you expected
Pujols to perform, wouldn't you base those decisions on his past
performance and future forecast?

I see it like this: Let's say you decide to go outside for a walk.
It's summer, 90+ degrees, in the midst of a drought, and all weather
forecasts you have at your disposal say there's not a drop of rain
coming as far as they can see. There's no guarantee that a freak
thunderstorm won't come out of nowhere and soak everything, but it's
exceptionally unlikely. Are you going to wear your raincoat and
galoshes out for you walk because you can't be 100% certain it won't
rain, or are you going to dress as your prior experiences and future
expectations suggest is appropriate for a sweltering summer day?

Pujols putting up MVP-caliber numbers is the weather staying hot and
dry. Pujols totally tanking the season, for any reason, is the
unpredictable thunderstorm. We recognize that it's a remote
possibility, but we don't build our expectations upon it. That's how
I approach it, anyway.
From: John Kasupski on
On Wed, 3 Mar 2010 10:48:21 -0800 (PST), HTP <tmbowman25(a)yahoo.com> wrote:

>and how do we know that a player has "performed at a high level"?
>
>
> Further up the thread you committed several collosal errors of logic,
>and rather than answer those individually i decided its best to tie it
>all up with a bow and ask you the above question.

I thought that was obvious - we know that a player has performed at a high level
because we have had the opportunity to observe his performance.

I think it's equally obvious that such observations guarantee nothing about his
future performance. Players get hurt. Players get older. Players have
uncharacteristically bad seasons in the middle of HOF careers. Players
experience uncharacteristically good years in the middle of otherwise
unremarkable careers. Players go on, or off, the juice - or get caught and
suspended for 50 games thus depriving their teams of their services.

And in tonight's game...did that player have an argument with his wife or
girfriend before he left to go to the ballpark? Did he tie one on last night and
show up at the ballpark today feeling like death warmed over? Did he eat
something for lunch that didn't agree with him? After all, these guys are human
beings. I don't always feel 100% every single day, and I wouldn't expect anyone
else to, either...even pro athletes.

I gotta go run some errands before the bank closes. Have a good afternoon, guys,
and I'll check in again later.

JK

From: Ron Johnson on
On Mar 2, 3:31 pm, John Kasupski <w2...(a)spamfilter.verizon.net> wrote:

>
> My point here is that, given that saber is looking at long-term results while
> traditionalist managers are looking at a shorter time frame, maybe the
> difference isn't as significant as some might think. I recall reading online
> somewhere that Hits actually tracks wins pretty well at 71%, that BA is a better
> stat than it's given credit for all things considered because it tracks wins at
> a 79% clip, that OBP tracks wins only 1.2% better than BA and that the
> difference between BA and OPS is another 5% of improvement.

You read wrong or somebody was writing BS.

First of all any correlation based argument will give nutty
results. The correlation between doubles and team runs scored
is much higher than the correlation between triples and
runs scored.

There's a positive correlation between runners left on
base and runs scored. Does that mean it's good to leave
runners on base? Hardly.

The proper way to approach (say) the BA/OBP/SLG issue
is to run a multiple regression against team runs scored.

And when you do this it turns out that OBP is worth
~1.7 times what SLG is and that BA is insigificant.

Likewise the double/triple issue. If you run a
multiple regression (including the other counter
stats) you'll get a proper weight.

Now as to the specific claims about correlations,
OPS has about a 94% correlation with team
runs scored, batting average 83% with OBP
and SLG at just under 90%.

And the standard error in modeling team runs scored
using batting average is ~48 runs. Better metrics
like linear weights are in the 15 run range. Runs
created is about 25 -- about where weighted OPS is.

OPS itself is around 28.
From: HTP on
On Mar 3, 11:32 am, John Kasupski <w2...(a)spamfilter.verizon.net>
wrote:
> On Wed, 3 Mar 2010 10:48:21 -0800 (PST), HTP <tmbowma...(a)yahoo.com> wrote:
> >and how do we know that a player has "performed at a high level"?
>
> > Further up the thread you committed several collosal errors of logic,
> >and rather than answer those individually i decided its best to tie it
> >all up with a bow and ask you the above question.
>
> I thought that was obvious - we know that a player has performed at a high level
> because we have had the opportunity to observe his performance.

and how is that observation of a players performance measured and
expressed? I'm trying to get you to say it here.

>
> I think it's equally obvious that such observations guarantee nothing about his
> future performance. Players get hurt. Players get older. Players have
> uncharacteristically bad seasons in the middle of HOF careers. Players
> experience uncharacteristically good years in the middle of otherwise
> unremarkable careers. Players go on, or off, the juice - or get caught and
> suspended for 50 games thus depriving their teams of their services.
>
> And in tonight's game...did that player have an argument with his wife or
> girfriend before he left to go to the ballpark? Did he tie one on last night and
> show up at the ballpark today feeling like death warmed over? Did he eat
> something for lunch that didn't agree with him? After all, these guys are human
> beings. I don't always feel 100% every single day, and I wouldn't expect anyone
> else to, either...even pro athletes.

You've been steadily obscuring the difference between absolute
certainty and probability, either intentionally or unintentionally.
Tom Dunne expressed it well in the previous post, that stats, as a
means of predicting and planning for the future, arent about
guarantees but about probabilities, and that human beings would be
unable to plan for the future or even to act at all if we were unable
to utilize information from pertinent past events.

Getting back to the specific issue at hand.

Willie Taveras has never been a guy that gets on base. He hits
leadoff ahead of good hitters, runs fast, yet doesnt score as often as
most other leadoff hitters. He has made no change in his approach
coming into 2009 and its reasonable to think that his main assett -
his speed - will or would diminish because of his age. Therefore its
more likely than not that his offensive capabilites would be similar
or worse to his previous season, and that once again he will be a
below average run rpoducer in the leadoff slot. Its not guaranteed but
its likely. Given that information, does it make sense to have Willy
set the table? Not if theres a better option. Well, there werent many
if any intially, but after some time passed, and more information was
gathered, it became more and more likely that Dickerson was more
capable that Taveras of getting on base. Yet Dusty kept Willie in that
spot. That deserves criticism.

The fact that Walt signed taveras, and to the contract he did, also
deserves a fair amount of just criticism, all based on Willies past
performances.

>
> I gotta go run some errands before the bank closes. Have a good afternoon, guys,
> and I'll check in again later.
>


yeah, lets call it for the day. later

From: Ron Johnson on
On Mar 2, 6:57 pm, tom dunne <dunn...(a)gmail.com> wrote:
> This is a bit off-topic, but I'm wondering about the benefit of stat-
> oriented managers versus the 'traditional' types.  As John mentioned,
> there are plenty of recent examples of successful field managers who
> aren't really noted for particularly forward thinking about baseball
> theory.  Joe Torre, Bobby Cox, Jim Leyland, Lou Pinella and plenty of
> others have been very successful despite starting their careers prior
> to the SABR era and continue to lead winning teams today.  Maybe a
> statistcially minded manager isn't that import.

A manager can bring other things to the table. Torre's track
record was nothing special before (like Stengel before him)
he ended up in a situation he was well suited for.

Don't screw it up is harder than it seems. And not everybody
can work for Steinbrenner.

Cox has a great record in three important areas. Turning
talent into players, judging who is actually good enough
to play and buiding (and rebuilding) bullpens (a specific skill
Sparky Anderson doesn't get enough credit for)

> There is something to
> be said for intelligent bullpen management and sensible lineup
> construction, but making those decisions properly really won't make a
> bad team good.

Yup. Sorting out the talent is the primary job.

Bill James wrote an interesting article (the Ken Phelps
All-Stars) which touches on manager preconceptions.

Specifically how guys like Henry Cotto tend to get
more chances than guys like Ken Phelps (and to see how
this is still at least partly true consider how
long it took for Jack Cust to establish himself)

Phelps could only do two things well (walk and hit home
runs) in spite of which he was a useful regular. Cotto
was a good fielder with good speed who managers thought
was one adjustment away from hitting .300 (there are
any number of quotes out there about how he was going
to change some flaw that was holding him back) but
was in reality nothing more than a good 4th outfielder.