THE BOOK--Playing The Percentages In Baseball

The two things I tend to look at for hitters when it comes to basic counting stats are walks and doubles. You can get a decent sense of a player, and if you’re already looking at the established triple crown numbers they’re a great sanity check; everybody tends to underrate walks and doubles in traditional analysis so they’re a counterweight.

Runs Scored for hitters and WHIP for pitchers.

How about the TTT for pitchers?

Ks, BB (fewest), and Homers allowed (fewest)

Would obviously need an IP requirement for the latter two.

I like the 2 counting/1 rate stat split.

So maybe…
OBP, HR, XBH
ERA, K, QS

Though I’m not terribly impressed with my choices.

Pitchers I would use ERA, innings pitched, and K/BB ratio.

For hitters, OBP, Total Bases, and I’m not sure what beyond that.

Constrained to basic stats....

Batters:
Use the two rate stats as suggested, and then also use RBI. Even though it is situation dependent, it’s the only common stat that can capture any hint of “clutch”, if that even exists.

Pitchers:
The two givens from Tangotiger, and WHIP. I consider the latter common enough now to be called basic, thanks to its use in Roto leagues.

Total bases, total “reaching first base without recording a hit” (i.e. IBB, UIBB, HBP, RBOE, FC, etc.), and ISO?

I’m not as hard core anti-batting average as some of you.  For me the triple crown is the 3 rate stats BA/OBA/SLG, and that’s how I present player stats when writing about them in my blog.

Maybe it doesn’t tell you much value wise, whether Nick Swisher hits 240/370/450, or Ichiro hits 330/370/450, but it adds a lot of descriptive value.  If you knew baseball numbers inside and out but were not familiar with current players, and I told you those two player lines, you’d know:

Who has more power?
Who’s more likely to swing at the first pitch?
Who makes better contact?
Which one is probably faster?
Which one do you need to play in at the corners on?

You could even make a good guess on which one is more likely to spray the ball and which one you might want to use a shift on.

Hmmm… since OBP is H+BB+HBP per PA-SH, then maybe the rule about “mixing” doesn’t make much sense.

So, BB+H per IP (WHIP) should count, so I’ll go with WHIP, ERA, K. 

For batters, I like total bases, so OBP and TB would be nice.

OPS is I think definitely mixing.

Since any rate stat is by definition a combination, I think a simple combo counting stat also ought be ok.  On that basis, I nominate:

For batters: OBA, SLG, R + RBI
For pitchers: WHIP, Opp Slg, RA (not ERA)

(I know you prefer R + RBI - HR; I went with the other primarily because of simplicity.)

I hate the whole concept of “earned” and “unearned” runs; hence, RA for pitchers.

Note both sets are pretty symetrical.

If you want a really simple set (more in keeping with the stated intention), I’d nominate for batters: OBA, HR and hits.  For pitchers, probably WHIP, RA and K’s.

I think OBP/ISO/R+RBI would be nice I guess, the last one kinda feels dirty. WPA/LI might be cool but it’s obviously too confusing.

For Pitchers I agree with the guy’s simple set above me, WHIP/RA/K’s

How about:

--OBP, XBH, R+RBI
--HR/9, K, BB/9 (Essentially FIP)

Side note: Fangraphs just added wOBA to the player pages. I asked Dave to add some kind of rate stat last year (I think I said EqA at the time, although I was more naive then), and I guess someone asked him again more recently. Very cool.

I know these are lame choices.

Hitters:

1. wOBA
2. Runs Above Average (traditional Linear Weights)
3. Runs Created w/PAR if the LW are reconciled to runs scored using R/O. Simply Runs Created if the LW are reconciled using R/PA.

Pitchers:

1. FIP or DIPS, or tRA (I don’t know which one is the best.)
2. Pitching Runs Above Average (using Baseruns)
3. Pitching Runs Created (Something like David Gassko’s stat, but using Baseruns)

Batters and pitchers each have one rate stat, one value stat, and one counting stat.

I was thinking a rate stat, a context-neutral counting stat, and a context-sensitive counting stat.

Some examples:

OBP, Total Bases, Runs + RBI (- HR?)
OBP, XBH, WPA
[got on, got ‘em over, got them in]

WHIP, Ks, Wins - Losses
component-type ERA, IP, Wins
[pitched well, pitched lots, team won]

hitters:

ABSO, GWRBI, Productive Outs

pitchers:

Wins, Pick-offs, WHIP

What useful information does WHIP give us about a pitcher that isn’t encapsulated in his walk rate and strikeout rate?

hitters:
BA, BB/K, ISO

pitchers:
BB/9, K/9, GB%

I like the discussion, but using all rate stats allow for a player that didn’t play the entire season to win the title.  If only rate stats are used they should be adjusted for plate appearances and innings pitched.

For example, settings innings at 200.  If a pitcher goes 220 innings or 10% more, he would get to adjust his ERA down by 10%.

Hitters
BA, XBH, R+RBI-HR

Pitchers
RA, K, WHIP

Hitters:

OBA, TB-H, RP (R+RBI-HR)

Starting Pitchers:

ERA, Fibonacci Pts., K

Hitters:
WOBA* / +/- / PA

Pitchers:
FIP* / xFIP* / BFP

* park adjusted line

OK, that’ doesn’t quite meet your criteria, but one could dream.

or probably better for triple crown

Pitchers:
FIP / K-BB / BFP

Actually scratch those.  PA doesn’t make sense, BFP only a little.  IP would probably be better for pitchers.

It’s hard to keep the “feel” of the traditional triple crown while at the same time coming up with stats that are actually meaningful.

I guess it would be OBP, HR, TB.

For pitchers:  ERA, K, K/BB seems right.  The importance of strikeouts is intentional, of course.

I’ve actually thought about this months before this had been brought up… and here’s my answer:

For hitters—GPA, ISO, and RC.
For pitchers—WHIP, ERA, FIP (If it was possible, I’d prefer to switch out ERA with WPA—except WPA only goes back so far...)

The only issue with this would be that, well, maybe too many Triple Crowns would be awarded.

Is anyone checking their suggestions to see who would have won any of them?

Jon #24: Your hitting Triple Crown looks just right: OBP, TB, HR.

With this formula in the post-1900 NL, Hornsby would keep his 2 Crowns, Klein would keep his 1933 Crown, Medwick would lose his but new winners would be Gavvy Cravath (1915), Mays (1965), Schmidt (1981), pre-PED Bonds (1993) and Larry Walker (1997).

In the AL, Lajoie and Cobb, Gehrig, Frank Robinson and Yaz all keep their Crowns.  Ted Williams keeps his two Crowns plus adds a third (1949). Foxx and Mantle lose their Crowns, but most interestingly, Babe Ruth now wins 5, count ‘em, 5, Triple Crowns: 1919, 1921, 1923, 1924 and 1926.

Colin/#16,

As far as information about the pitcher and his skills, WHIP probably doesn’t add a lot to the discussion, though it does allow for more noise since hits are tied to defensive performance. But, in the context of having a Triple Crown, WHIP is a good way of showing which pitchers (and their defenses) allowed more baserunners. There may be more noise this way, but I don’t think the point here should be to determine which pitcher has the best skills.

#25,
How much should the frequency be a factor? We’d still want it to be a special event, not an annual one, right?

It’s clear that we need at least one counting and one rate stat.

To that end, my favorites from what has been discussed is:
Pitchers: ERA, WHIP, K-BB
Hitters: OBP, SLG, R+RBI-HR

I could also go for “All Purpose Bases”, meaning BB+TB+SB-CS.  Not sure which of the other 3 to drop.

I think you still need to keep 2 counting and one rate stat.

Bryan/#32 - Why?

It just seems like it feels right for the Triple Crown.  I think it is important to make sure the guy who wins the Triple Crown plays alot and is also able to maintain a high rate stat.  For example, Chipper Jones had a great year this past year.  If it wasn’t for the god that is Pujols, he would have almost won all of the common rate stats.  The reason why he (or a player like him) didn’t win the triple crown is that he just didn’t play in enough games.  Having more rate stats rewards a guy who can play 162 games which helps his team win more games.

I think maintaining the rate/counting stat ratio maintains the essence of the Triple Crown.

For Hitters: OBA,XBH,RC all easily accesible in
Baseball Reference

Has happened 21x in the AL, 23x in the NL since 1901

AL winners
2001 GIAMBI
1994&1992 THOMAS
1972 ALLEN
1967 YAZ
1966 ROBBIE
1955 MICKEY
1949,47,46,42 THUMPER
1938 FOXX
1924,23,21,20,19 BABE
1917 COBB
1912 SPEAKER
1904&01 LAJOIE

NL WINNERS:

2001,1993,92 BONDS
2000 HELTON
1998 MCGWIRE
1997 WALKER
1981 SCHMIDT
1956 THE DUKE
1948,44,43 THE MAN
1933 KLEIN
1925,24,22,21,20 RAJAH
1915 CRAVATH
1910 MAGEE
1909,08,07,04 HANS

Suppose we were to say that before divisional play started the winner had to lead the Major Leagues not just his league. Then it drops to 12x in NL and 16x NL. The breakdown becomes 11x since ‘69 and 16x prior.

Rempart #35: I think the high frequency of occurrence under your proposed combination (44 winners since 1901, up from 13 under the current formula) suggests that there may be too much overlap across the three categories.  XBH flattens doubles, triples and homers in the same way OBA does, and RC is mostly just OBA times TB.  The OBA/TB/HR combination, as suggested in comment #24, measures three relatively distinct hitter qualities and thus will only be won by a hitter who demonstrates an unusually wide and versatile range of league-leading batting skills, which seems to me the primary point of the Triple Crown.  OBA/TB/HR would produce 8 NL winners since 1900 and 8 AL winners not including Babe Ruth who would himself have five.  10 of thirteen post-1900 traditional Triple Crown winners would still hold their Crowns, and there would be 11 new winners (Babe’s 5 plus 6 others), for a total of 21, compared to the current 13, or to put it another way, 16 non-Babe winners compared to the current 13, plus 5 Ruth seasons.  Note that one of the Ruth seasons would be when he was still with the Red Sox, in 1919, when Cobb won the AL BA title by a wide margin but Ruth had the AL’s highest OBA (Ruth had 101 BBs compared to Cobb’s 38).