Playing the Percentages in Baseball and other 'cap ramblings

[EXTRAPOLATER]

Reference is to the The Book-Plaaying the Percentages in Baseball,
which can easily be found at Amazon and elsewhere.

I'm taking a look. Thought, originally, it was from the past year or two but apparently it was
published in 2007; no big deal, really. They do some heavy statistical analysis from 2000-2003,
which is a reasonable sample, regardless of 'pitchers years' vs 'hitters years' arguments which might
be made (e.g. I have heard that offense has been up the past couple of seasons, but umpire analysis
over the same period shows a much higher strike%--more important for me in totals as opposed to
ump ov/un record--so that might be an anomaly). BTW, runs appear to have been up between '06
and '12 though I haven't looked at OV/UN marks (runs down some for '13).

I'm through chapter 2 and will comment what I think of their analysis and from my own attempts.

For starters, I find it interesting that they comment on how both on-base percentage and slugging
percentage are very relevant factors to consider, but provide a rather unconvincing argument that the
combined statistic--OPS (on-base + slugging)--is lacking, and try to replace it with their wOBA or
'weighted on-base percentage.' I look at many factors while assessing a game and while I have
been trying to follow somewhat of a K.I.S.S. rule (keep it simple, stupid), I can understand where
certain trends may be relevant and others not so much so; nobody else is using wOBA while
OPS is readily available, especially team OPS, and the team aspect I will comment on while looking
at chapter 2.

As an aside, perhaps, I find it a waste that they analyze expected results based on the state of a
game prior to the end of an inning. i.e. if the first runner reaches base, etc. It may be mildly
interesting but does me no good for my purposes. Their 'win expectancy' examination by 'game
state' has the same problem.

Chapter 2 gets into some interesting things about hot-and-cold streaks, which I have heard discussed
elsewhere on general probability theory, and would like to hear comments on. I have my own
beliefs and how many are based on real experience vs an unwanted bias is something that I am
still trying to work through.

I'm hoping to get some discussion going and figure that this forum tops the general discussion option
as I'll be sticking mostly to baseball, which the book focuses on and which I'm most proficient on
speculating.

More to come barring death or something worse.

 

[EXTRAPOLATER]

On hot and cold streaks

On hot and cold streaks

(for the athletes, not us)
((or maybe us, if it seems relevant (((it should)))))

The Book goes through a statistical breakdown to show to show no significant advantage for
a single offensive player after a hot streak, which they define in a couple of ways.
When they do a similar analysis of pitcher's hot streaks they DO find a significant advantage
for the player following the hot streak.

For starters, in several other places I've heard probability discussions that speak of a witness's
bias for something such as a coin-flip or the red-black appearance on roulette. They talk about
how one can believe that a certain flip or colour is due to an overall statistical likelihood (e.g
if heads or red has come up x times in row then tails or black is due to hit). Obviously, at least
for everybody using this forum--I hope--we all realize that past history in such a case is irrelevant
to the future, as in a coin-flip where each one is merely a 50%-50% call, each time, as long as
the coin is legit.

I have often heard the following analogy with athletes. Take a basketball player, for instance.
If the player has made their last 3 or 4 shots, for example, then our bias (and often the coach's,
and fellow teammates) that the player is hot and is more likely to make their next shot then their
overall shooting % dictates is wrong. I hope that one example will suffice but I, for one, have been
following sports long enough to believe that this analogy is dogshit; comparing the odds on a coin-flip
to a player's immediate performance--with whatever good sleep, good frame of mind, and whatever
other psychological and physical factors might be favoring (or disfavoring) him--seems to be a
total category mistake or some other kind of logical fallacy. Whether that can carry on from one
game to the next is another question, but anybody here who has had in-game action probably
knows that a coherent viewer has a serious edge save for the fact that the oddsmakers are not
as daft as we would like.

In-game is one thing and it's difficult to make money with that, especially on baseball.
What about games following a hot or cold streak?

The Book's analysis of individual player's records--hitters--following a streak is mildly interesting,
but I have not yet seen (through ch.2) a similar breakdown for team's performance. I'll mention
my approach to the same and take it from there.

I, for one, use team OPS quite a bit. I believe that it is the single, most important offensive statistic,
for anyone wanting to keep things as simple as possible. Things such as run differential (as Joe Peta
admires) can be crucial but, of course, such figures in pitching--starting and bullpen--significantly.
For OPS, obvious home-away and right-left matchups are important, so I can't look at the Rockies
overall OPS compared to, say, the Padres overall OPS and just leave it at that; adjustments have to
be made depending on a number of factors which, I confess, can sometimes be a pain in the ass.
When coming up with a number for offensive strength, however, as long as the sample size is
reasonable (as it should be now--post-May), I use OPS numbers (again, doing some weighting for
home-away, said park for match, vs L or R splits) but also factor in current form to the same. 7-day
numbers are good and readily available, as are month-to-month numbers and other breakdowns.
The weighting is tricky and mostly--unfortunately--subjective, but I'll generally weight about 25%
(+ or -) to recent form when calculating an offensive power number.

Just a quick word on pitchers. Certain starters appear to be streakier than others but I still haven't
figured out if that is an illusion or not. And, we basically all know who the stud starters are and who
the dogs are (as do the linesmakers), and the current form of those pricey studs seems crucial
before backing them, whatever the price. Any info we can obtain on outside distractions--contract
negotiations, family trouble, whatever) can help but are really inconclusive overall (for me, just giving
a reason for a no-bet, usually). Bullpen status can be important, too; an overuse of the BP in a
previous series, for example, can lower a team's expected win% but with a starter expected to go
7 or 8 and a superior offense then any such modification might be negligible.

I'm going to leave it there.
Ramblings, as mentioned.

Maybe some over/under thoughts next, as I'm trying to make sense of something, there.
Back to The Book after chapter 3 or 4 is consumed.

 

[The Mover]

Extra... the reviews on this book are good & I love stats but it seems the math part would be overwhelming . One of the reviewers quoted this & sums it up for me" they are aware of an important bit of knowledge: (a) that not everything is measurable, and (b) that some aspects of the not measurable are important"
Thanks for sharing, will check back on your additional thoughts.