Pondering Francoeur

What if I told you about a young player, a 22-year old rookie with the following
numbers in his first full year:

G     AB    R    H     2B   3B   HR   RBI
157   618   84   171   32   4    29   105 

If I then told you he was a quality corner outfielder with a

cannon
for an arm (19 outfield assists), you’d say, “Wow! A sure-fire Rookie
of the Year candidate. And he’s only 22? The sky’s the limit for this
guy!” There is a key number that I’ve left out of the above line,
though. Maybe 20 or 30 years ago you wouldn’t have noticed,
but nowadays everybody is aware of the importance of the walk, and indeed I
haven’t told you how many walks our rookie drew in his first
year. Nineteen (ouch). But it’s even worse than that, because six of
those walks were intentional. His rate of unintentional walks (UIBB)
is only 2.1%.

Only four qualifying batters in the last 10 years (out
of more than 1400 total) have
had a lower UIBB rate. Of course, the near-total absence of walks
leads to a very low OBP. Here is our batter’s rate stat triplet for
his first full season: .277/.305/.482. That .305 is disturbing, but
the overall performance is still good for a 22-year old rookie, isn’t
it?

The Rise and Fall

Jeff Francoeur made his major league debut on July 7, 2005. He went
1-4, with an 8th-inning 3-run homer off Glendon Rusch to ice a Braves
victory. His line for the day, .250/.250/1.000, would be an omen of
things to come. But not right away. Frenchy hit his next home run
three days later against Milwaukee, and on July 20 he hit another
against San Francisco. He kept on hitting like a madman for about two
months, and at the end of August 2005, his batting line stood at
.342/.378/.632. He’d hit 10 home runs and driven in 31 runs in 43
games. Nobody cared a whole lot that he’d only walked six times (four
unintentional) during that period. Francoeur was riding high.

Since Sept. 1, 2005 (through the one-year mark, July 7, 2006)
Francoeur has batted .255/.282/.444. What are we to think of that
line? The power is nice, but the OBP is terrible. How does Francoeur’s overall
offensive performance compare to other right fielders? Well, there are
10 right fielders in the National League who qualify for the 2006 batting
title (as of August 25). Francoeur ranks 9th in Runs
Created per Game and 8th in OPS.
If we want to factor in his defense, we can consider his Win Shares total: He currently
ranks 30th
in Win Shares among all NL outfielders (not just right fielders).

So, to answer my own question posed above, despite the impressive home
run and RBI totals, Francoeur is truly among the very worst regular
outfielders in the NL right now. And it is very clear what the problem
is: The almost complete lack of walks is keeping Francoeur’s OBP at
sub-replacement level. Of course, he’s only 22 years old and there is
a lot of time to improve.

But can plate discipline be learned or is it an innate skill? Many
think it’s more of an innate skill than something that can be learned,
and I’m sure there is some truth in that. I expect that pitch
recognition, which is surely a significant part of plate discipline,
depends on innate physical attributes, such as good vision and quick
reflexes. On the other hand, Tom Tango has shown that a player’s walk rate tends to improve with
age, essentially until he retires. At some level, batters can, and do,
learn to take a walk.

One way to approach the question is to see if there have been players
in the past who are similar to Jeff Francoeur at age 22 and, if so,
see what they did in the rest of their careers. Let’s see if this
kind of investigation can teach us anything.

Everybody wants you to be just like them …

I believe Bill James was the first person to come up with the idea of
“Similarity Scores”. The idea is to compare two players based on their
stats. Did they have a similar number of home runs? Did they walk the
same amount? Did they play the same position? James devised a
numerical method for evaluating how similar two players are, and he
used these Similarity Scores extensively in his book Whatever
Happened to the Hall of Fame? You can see James’ Similarity
Scores in action at Baseball Reference, where the top
matches are shown for each player.

Similarities among players are also used by analysts doing
projections. To project how a young player might do in the future,
similar players in the past are identified to see what kind of careers
they ended up having. You can find a discussion of sim scores for
projection systems in Part
4 of our Projection Roundtable. Here I’m going to be doing
something similar in looking for comparables to Jeff Francoeur. I’m
not going to compare body types (height and weight) or handedness or
any of that stuff. I’m just going to look for young players who had
about the same number of plate appearances, similar batting average
and power numbers, and similar walk and strikeout rates.

Here’s what I did: I first selected a sample of players who have accrued
roughly the same number of plate appearances at the same age as
Francoeur. Anybody who by the end of their age-22 or age-23
seasons who had made between 500 and 1000 plate appearances makes it
into my sample (Francoeur will have around 950 plate appearances by the end of this
season). Intentional bases on balls were not officially recorded
before 1955, and they are needed in this study, so my sample contains players who debuted after that
year. I’ll refer to these players as “rookies” below, although I
haven’t applied the usual rules for rookie-eligibilty.

The first thing you can do with this sample is see if there are any
players with a similar walk rate to Francoeur’s. Here are the 10
closest matches in UIBB rate:

Low Walk-Rate Rookies
+-----------------+-----------+----------+
| name            | years     | uibbrate |
+-----------------+-----------+----------+
| Reitz_Ken       | 1972-1973 |    0.018 |
| Fuentes_Tito    | 1965-1966 |    0.019 |
| Francoeur_Jeff  | 2005-2006 |    0.019 |
| Oliver_Al       | 1968-1969 |    0.021 |
| Lanier_Hal      | 1964-1965 |    0.024 |
| Nixon_Russ      | 1957-1958 |    0.024 |
| Baines_Harold   | 1980-1981 |    0.025 |
| Alou_Jesus      | 1963-1965 |    0.025 |
| Santiago_Benito | 1986-1987 |    0.026 |
| Izturis_Cesar   | 2001-2002 |    0.026 |
| Lee_Carlos      | 1999-1999 |    0.026 |
+-----------------+-----------+----------+

Actually, every one of these players had a significant major league
career, something that I wasn’t expecting. Several of them were
primarily glove men: Ken Reitz, Tito Fuentes and Hal Lanier (the double play combo on the
Giants of the late 1960s), and Cesar Izturis. There are a couple of catchers
(Benito Santiago, Russ Nixon), and three guys who turned out to be very good
hitters: Al Oliver, Harold Baines and Carlos Lee.

But did these guys ever learn to take a walk? The next table shows
their career UIBB rate:

Low Walk-Rate Rookies: Before and After
+-----------------+----------+--------+
| name            | < 24 yrs | Career |
+-----------------+----------+--------+
| Reitz_Ken       |    0.018 |  0.028 |
| Fuentes_Tito    |    0.019 |  0.047 |
| Oliver_Al       |    0.021 |  0.044 |
| Lanier_Hal      |    0.024 |  0.029 |
| Nixon_Russ      |    0.024 |  0.044 |
| Baines_Harold   |    0.025 |  0.081 |
| Alou_Jesus      |    0.025 |  0.024 |
| Santiago_Benito |    0.026 |  0.052 |
| Izturis_Cesar   |    0.026 |  0.041 |
| Lee_Carlos      |    0.026 |  0.069 |
+-----------------+----------+--------+

Well, this looks like some good news for Frenchy. Nine out of these
10 players improved their walk rate after putting up poor rookie
numbers. On the other hand, only Baines brought his walk rate up to
the league average, and only two other players managed to surpass the
5% threshold. If Francoeur could get his walk rate to 5%, how
would that affect his overall performance? Well, it depends on where
the walks come from: the outs or the hits. Let's be optimistic and
assume that he turns enough outs into walks to give him a 5% UIBB
rate. Doing a quick-and-dirty calculation, his modified lifetime batting
line comes out .281/.322/.499 (compared to his actual numbers
.272/.300/.483). That's a lot better, of course, and it's about average
for a corner outfielder these days.

Frenchy's Doppelgangers

The thing is, though, most of these players are not really comparable
to Francoeur. Reitz, Lanier, Fuentes, Nixon, Jesus Alou and Izturis all hit
for much less power than Frenchy. And it turns out that the two best power
bats in the list, Baines and Lee, showed the biggest improvement in
walk rate. So, maybe Francoeur has a better shot at improvement than
this list would have you believe. We should be looking at players who
are similar to Francoeur in various ways, not just walk rate.

So, in an attempt to find
players that were generally similar to Francoeur, I
also computed for each player the following stats for those
plate appearances: batting average, isolated power and strikeout
rate. I then formed something called a chi-square to combine these
stats (together with UIBB rate) to give me a single number that tells
me how similar any given player is to Jeff Francoeur. (If you are
interested in the details of the chi-square calculation, you should
get a life. Uh, I mean, I give them
below in the "Resources" section). The following table gives the top
20 matches for Jeff Francoeur, based on our matching stats.

Jeff Francoeur's Most Similar Players
+--------------------+-----------+------+-------+----------+-------+-------+------+
| name               | years     | pa   | ba    | uibbrate | iso   | krate | chi2 |
+--------------------+-----------+------+-------+----------+-------+-------+------+
| Francoeur_Jeff     | 2005-2006 |  746 | 0.272 |    0.019 | 0.211 | 0.197 |  --- |
+--------------------+-----------+------+-------+----------+-------+-------+------+
| Encarnacion_Juan   | 1997-1999 |  742 | 0.271 |    0.032 | 0.203 | 0.214 |  0.3 |
| Samuel_Juan        | 1983-1984 |  806 | 0.273 |    0.036 | 0.170 | 0.231 |  1.2 |
| Presley_Jim        | 1984-1985 |  873 | 0.261 |    0.046 | 0.199 | 0.187 |  1.3 |
| Cantu_Jorge        | 2004-2005 |  807 | 0.289 |    0.034 | 0.200 | 0.159 |  1.5 |
| Santiago_Benito    | 1986-1987 |  631 | 0.299 |    0.026 | 0.168 | 0.198 |  1.9 |
| Gonzalez_Alex      | 1998-1999 |  683 | 0.260 |    0.036 | 0.150 | 0.213 |  1.9 |
| Mondesi_Raul       | 1993-1994 |  542 | 0.304 |    0.028 | 0.208 | 0.174 |  2.0 |
| Munoz_Pedro        | 1990-1992 |  671 | 0.273 |    0.040 | 0.147 | 0.205 |  2.0 |
| Dye_Jermaine       | 1996-1997 |  584 | 0.259 |    0.043 | 0.157 | 0.203 |  2.1 |
| Dunston_Shawon     | 1985-1986 |  874 | 0.253 |    0.037 | 0.152 | 0.179 |  2.4 |
| Baines_Harold      | 1980-1981 |  805 | 0.266 |    0.025 | 0.167 | 0.132 |  2.5 |
| Dawson_Andre       | 1976-1977 |  651 | 0.275 |    0.053 | 0.175 | 0.163 |  2.6 |
| Lee_Carlos         | 1999-1999 |  509 | 0.293 |    0.026 | 0.171 | 0.143 |  2.6 |
| Uribe_Juan         | 2001-2002 |  888 | 0.260 |    0.046 | 0.141 | 0.199 |  2.9 |
| Roberts_Dave       | 1972-1973 |  933 | 0.266 |    0.033 | 0.135 | 0.158 |  3.0 |
| Rice_Jim           | 1974-1975 |  676 | 0.304 |    0.050 | 0.174 | 0.200 |  3.4 |
| Pepitone_Joe       | 1962-1963 |  751 | 0.265 |    0.032 | 0.182 | 0.113 |  3.4 |
| Gant_Ron           | 1987-1988 |  696 | 0.260 |    0.062 | 0.172 | 0.186 |  3.4 |
| Durham_Leon        | 1980-1981 |  677 | 0.281 |    0.057 | 0.163 | 0.160 |  3.5 |
| Howard_Frank       | 1958-1960 |  536 | 0.261 |    0.064 | 0.195 | 0.240 |  3.5 |
+--------------------+-----------+------+-------+----------+-------+-------+------+
chi2 - the lower the number, the more similar the player

The overall best match is Juan Encarnacion, the current right fielder
for the St. Louis Cardinals. Encarnacion matches almost perfectly in
all categories, except for walk rate. In fact, as you can see from the
table, none of
these players had walk rates as low as Francoeur's.
Again, the majority of the people on
this list had solid major league careers and several of them were
stars with borderline Hall of Fame credentials. The two players who
never did much were Pedro Munoz and Dave Roberts, who fashioned 600-game
careers as part-time or semi-regular players.

And how did this group fare in career walk rate, after the slow
starts? The following table shows the rookie and
career walk-rates for these players (I've removed Jorge Cantu and Juan Uribe, who haven't really
had careers yet).

Walk Rates for Most Similar Players
+------------------+----------+--------+-------------+
| Name             | < 24 yrs | Career | Improvement |
+------------------+----------+--------+-------------+
| Encarnacion_Juan |    0.032 |  0.057 |       0.025 |
| Samuel_Juan      |    0.036 |  0.062 |       0.026 |
| Presley_Jim      |    0.046 |  0.051 |       0.005 |
| Santiago_Benito  |    0.026 |  0.052 |       0.026 |
| Gonzalez_Alex    |    0.036 |  0.039 |       0.003 |
| Mondesi_Raul     |    0.028 |  0.068 |       0.040 |
| Munoz_Pedro      |    0.040 |  0.053 |       0.013 |
| Dye_Jermaine     |    0.043 |  0.076 |       0.033 |
| Dunston_Shawon   |    0.037 |  0.026 |      -0.011 |
| Baines_Harold    |    0.025 |  0.081 |       0.056 |
| Dawson_Andre     |    0.053 |  0.043 |      -0.010 |
| Lee_Carlos       |    0.026 |  0.069 |       0.043 |
| Roberts_Dave     |    0.033 |  0.050 |       0.017 |
| Rice_Jim         |    0.050 |  0.067 |       0.017 |
| Gant_Ron         |    0.062 |  0.102 |       0.040 |
| Pepitone_Joe     |    0.032 |  0.043 |       0.011 |
| Durham_Leon      |    0.057 |  0.088 |       0.031 |
| Howard_Frank     |    0.064 |  0.091 |       0.027 |
+------------------+----------+--------+-------------+
| Average          |    0.040 |  0.062 |       0.022 |
+------------------+----------+--------+-------------+

Again, almost everybody improved their walk rate, with the notable
exceptions of Andre Dawson and Shawon Dunston. The average walk rate of these 18
players went from 4% to 6.2%.

What does it mean?

So, what does all this mean for Jeff Francoeur? Well, let me tell you
a secret: I just don't know what it means. To be perfectly
honest, I don't know if the "similar players" method actually
works. To my knowledge, no one has ever demonstrated that it helps in
predicting players' futures. Oh, I agree that
it seems logical, but

many things that seem very logical at the time turn out to be quite
wrong.

Still, I think we can draw some conclusions about
Francoeur. I think there is a good chance he will improve his plate
discipline significantly. Many others have done so. However, when we
look at how many players make a
dramatic improvement in walk rate, say, more than
doubling it, there are not very many. In the list above, only Baines,
Raul Mondesi and Carlos Lee have accomplished the feat. My guess is that
Francoeur's upside is "average corner outfielder".

References & Resources

• What Ever Happened to the Hall of Fame? is an excellent book by Bill James. It's more than just an analysis of who should or should not go into the Hall, it's a history of the institution itself and it also contains lots of sabermetric stuff. And, as usual, the writing is entertaining. Good stuff.
• The chi2 variable that I used to find similar players to Francoeur is the sum of four different terms, each one having the same general form: [(v - jf)/s]^2. The four terms correspond to the 4 matching variables that we use: UIBB rate, strikeout rate, batting average and isolated power. "v" is the value for the player being compared, "jf" is Jeff Francoeur's value for that variable. "s" represents the RMS of the variable in question, for our sample of rookies.
  A perfect match would give a chi2 value of zero and the worse the match, the higher the chi2 value.