While looking toward the future with our comprehensive slate of current content, we'd also like to recognize our rich past by drawing upon our extensive (and mostly free) online archive of work dating back to 1997. In an effort to highlight the best of what's gone before, we'll be bringing you a weekly blast from BP's past, introducing or re-introducing you to some of the most informative and entertaining authors who have passed through our virtual halls. If you have fond recollections of a BP piece that you'd like to nominate for re-exposure to a wider audience, send us your suggestion.

Ten years ago, Nate Silver looked five years ahead in the piece reprinted below, which was originally published on April 2, 2003.

I knew it was bad when, on opening night, I found myself rooting not for my favorite team, or my favorite players, or even the talent on my fantasy rosters, but for the PECOTA forecasts.

David Eckstein can't get a hit! He's a collapse candidate! That little tow-headed twerp! Michael Young? Michael Young?!? You're killing me, man, killing me.

Between watching my boys go down in flames in the NCAA tournament, and trying to get anything at to run over a modem connection without crashing my computer, it had been a long couple of days. I don't believe that a projection system is useless once the season begins–especially something like PECOTA, which should be helpful in analyzing fast starts and April slumps once the numbers begin to accumulate.

But some of the more unlikely events of the first two days–Doug Glanville's season-opening walk, Corey Patterson's Tuffy Special, Pedro Martinez's failure to no-hit the Devil Rays–served as a reminder that there's no substitute for actually playing the games.

With that in mind, I'll run one more projections article this week, this time focusing on PECOTA's five-year forecasts, while promising to shift my focus to other topics for the remainder of the month.

As far as I am aware, no other forecasting system has published a complete set of projections that run more than one year into the future. Apart from the mathematical limitations of competing models, there isn't much good reason for this: There are many situations in which it's markedly more helpful to look several years into the future than to take a piecemeal approach, whether it's determining which player to lock up for his arbitration years, or deciding who to pick up in a keeper league.

Certain problems tend to compound over time. Nagging injuries become chronic, and then debilitating; eroding foot speed produces ever-lower batting averages; the stress on a pitcher's arm accumulates without warning, until, like an earthquake fault, it snaps. The further one projects into the future, the more uncertainty there is.

Just as it 'solves' the problem of the relative unpredictability of pitching statistics by building this uncertainty into the forecast itself, PECOTA can explicitly model these sorts of risks when projecting more than one year going forward. Even for good, young players, it is common to see their projected wins above replacement values decline with surprising briskness over time. Baseball, like any other sport, is a game of attrition, and for a player who is an established star, there is more room to fall downward than to make incremental gains upward. With a few notable exceptions, a Hall of Fame player isn't someone whose value increases over the course of his career, but rather someone who establishes a high baseline at a young age and is able to sustain it over a long period of time.

But enough with theory. Following are the top 20 players in the National League over the course of the next five seasons, as rated by the PECOTA wins above replacement projection. Without getting into too much discussion, five years is an appropriate time frame for a number of reasons. It jibes with the number of years that a team can expect to keep a young player cheaply, the average lifespan of a high-level executive, as well as the term of the depreciation tax shelter that savvy owners use to keep more money in their pockets as they flip their franchise to the next horse in the carousel.

                2003    2004    2005    2006    2007    Total
1  Pujols       +6.8    +6.9    +6.5    +5.8    +5.6    +31.6
2  Bonds        +12.3   +9.6    +4.3    +3.8    +0.5    +30.5
3  Berkman      +6.9    +6.4    +5.5    +5.0    +4.7    +28.5
4  Guerrero     +6.1    +5.5    +5.7    +5.0    +4.8    +27.1
5  Chipper      +6.4    +5.6    +4.6    +4.4    +2.8    +23.8
6  Thome        +6.8    +5.5    +4.5    +4.2    +2.4    +23.4
7  Helton       +5.8    +5.8    +4.5    +3.7    +3.1    +22.9
8  Schilling    +7.1    +5.0    +5.2    +3.0    +2.3    +22.6
9  Giles        +6.1    +5.0    +4.4    +3.3    +3.7    +22.5
9  Rolen        +5.3    +5.1    +4.5    +4.1    +3.5    +22.5
11 Abreu        +5.2    +5.1    +4.7    +3.7    +2.9    +21.6
12 Dunn         +4.2    +4.0    +4.1    +4.5    +4.6    +21.4
13 Vidro        +5.0    +4.8    +4.3    +3.6    +3.1    +20.8
14 Andruw       +4.9    +4.3    +4.0    +4.0    +3.5    +20.7
15 Oswalt       +4.8    +4.2    +4.4    +3.7    +3.2    +20.3
16 Kent         +6.5    +5.2    +4.0    +2.3    +1.6    +19.6
17 Ra. Johnson  +6.6    +4.7    +2.6    +3.3    +2.1    +19.3
17 Green        +5.3    +4.7    +3.8    +3.0    +2.5    +19.3
19 Sosa         +5.9    +5.0    +3.1    +3.3    +1.9    +19.2
20 Burrell      +4.2    +4.2    +3.8    +3.3    +2.7    +18.2
20 Ja. Vazquez  +4.4    +4.6    +3.5    +3.1    +2.6    +18.2

The best player in baseball debates are usually fought between A-Rod, Bonds, and Vladimir Guerrero, with more enterprising analysts daring to throw the names Pedro and Randy into the mix. Very rarely is Albert Pujols' name mentioned–and yet, he may well be the most valuable player (lower case) in the National League over the course of the next five seasons. Pujols isn't flashy and doesn't play a glamour position; he just hits the snot out of the ball, and should continue to do so for the next decade. It is fitting that his comparables list includes Cal Ripken and Eddie Murray, two of the most boring great players in memory.

Some of the reluctance to place Pujols among the best of the best stems from concerns over whether he's really as young as his birth certificate suggests. While there's no smoking gun on the matter, there's enough circumstantial evidence that it's worthwhile to play what-if, and PECOTA is up for the task. Let's move Pujols' listed birth date back in time by exactly three years, making him 26 this year, instead of 23, and re-run the projection.

                   2003    2004    2005    2006    2007    Total
Pujols (1/16/80)   +6.8    +6.9    +6.5    +5.8    +5.6    +31.6
Pujols (1/16/77)   +5.2    +5.0    +4.5    +4.4    +3.8    +22.9

The picture changes rather dramatically; Pujols is no longer being compared to inner circle Hall-of-Famers, but to merely good players like Richie Zisk and Wally Joyner, who came up relatively late and hit well for a few years, but weren't tremendous athletes and had their best seasons before the age of 27. Those three years make a lot of difference, even in the short run, and it may be no coincidence that the Cardinals appear content to tie Pujols up for a year at a time, rather than lock him down through his arbitration years.

There are some interesting names further down the list, too. Among serious analysts, Todd Helton might be the most underrated player in baseball. Sure, he's a first baseman in Coors Field, a position for which we ought to have higher standards than the Michelin guide. But he's also posted EQAs ranging between .320 and .345 for the past three years, played good defense, and his back problems notwithstanding, has been very durable. He's a comparable player to Will Clark or Keith Hernandez in their primes.

Curt Schilling beats out Randy Johnson and Roy Oswalt as the top pitcher in the NL. While it's true that age doesn't matter so much as strikeout rate in projecting pitching performance, Johnson's slim edge in strikeouts isn't enough to overcome Schilling's three years in the age column. Where's Mark Prior?

        2003    2004    2005    2006    2007    Total
Prior   +3.0    +4.1    +3.7    +3.7    +3.0    +17.5

He's the very next name on the list, behind Pat Burrell and Javier Vazquez. With all the praise that has been heaped on Prior since his debut, that projection seems a little stingy, and if you want to give him extra credit for poise and mechanics, it probably is. But considering the risk inherent to any young pitcher, and that Prior has accumulated all of six big league wins, it's actually a remarkably good score. Compare Prior to some other highly regarded youngsters, and it's clear just how far he is ahead of the curve:

         2003    2004    2005    2006    2007    Total
Sabathia +2.5    +2.6    +2.0    +2.3    +1.7    +11.1
Beckett  +2.2    +2.1    +2.3    +2.3    +2.1    +11.0
Sheets   +2.5    +2.1    +2.2    +2.1    +1.6    +10.5

Now turning to the American League, where there isn't quite as much depth.

1  A-Rod    +9.0    +8.6    +7.4    +7.0    +6.5    +38.5
2  Pedro    +8.4    +7.1    +5.9    +6.3    +5.8    +33.5
3  Matsui   +7.1    +6.4    +6.4    +5.6    +4.5    +30.0
4  Giambi   +7.9    +6.6    +5.8    +5.0    +4.6    +29.9
5  Nomar    +7.2    +6.3    +5.6    +4.7    +3.9    +27.7
6  Chavez   +5.9    +5.3    +5.4    +4.7    +4.4    +25.7
7  Glaus    +5.7    +5.4    +5.4    +4.2    +4.3    +25.0
8  Ramirez  +5.9    +5.8    +4.3    +4.1    +3.2    +23.3
9  Soriano  +5.0    +4.8    +4.2    +3.9    +3.9    +21.8
10 Tejada   +5.0    +4.5    +4.4    +4.1    +3.5    +21.5
11 Jeter    +5.0    +4.4    +4.1    +3.9    +3.0    +20.4
12 Hinske   +4.6    +4.1    +4.2    +3.4    +3.1    +19.4
13 Magglio  +4.9    +4.2    +3.9    +3.2    +2.5    +18.7
14 Zito     +4.3    +4.3    +3.7    +3.3    +2.9    +18.5
15 Delgado  +5.6    +4.5    +3.3    +2.9    +1.9    +18.2
16 Beltran  +4.0    +3.9    +3.5    +3.2    +3.2    +17.8
17 Williams +5.7    +4.2    +3.2    +2.1    +1.3    +16.5
18 Sweeney  +4.6    +3.9    +3.3    +2.4    +2.1    +16.3
19 Hudson   +4.2    +3.5    +3.0    +2.8    +2.4    +15.9
20 Mussina  +4.5    +4.6    +3.0    +2.0    +1.6    +15.7

Well, that's about what you'd expect. A-Rod and Pedro are truly in a class by themselves. The top 12 is rounded out by the three next-best shortstops, the three good young third basemen, and a bunch of Yankees. Hideki Matsui might be a notch too high since PECOTA is giving him credit for being a center fielder, as he was for most of his Japanese career, but even so, the Yankees have acquired a superstar talent in mid-career, and if he plays up to even 80 percent of that projection, he'll be a bargain.

Each of the third basemen can make a case for being underrated. Troy Glaus, even after a down season, still has Mike Schmidt and Harmon Killebrew high up on his comparables list, Eric Chavez will be an MVP candidate with a 10 percent improvement in his plate discipline, and Eric Hinske has a robust set of offensive skills that should ensure a high return on the Jays' investment. Speaking of third basemen, the top rookie-eligible player on the list will surprise absolutely nobody:

         2003    2004    2005    2006    2007    Total
Teixeira +2.0    +2.6    +3.0    +3.2    +3.6    +14.4

That's T-E-I-X-E-I-R-A, and if he hits the ball anything like he did in spring training, he'll eliminate the Travis Lees from his comparables list, and be well on his way to becoming the best spelling error since Carl Yastrzemski.

The five-year forecasts can also be used as a barometer for organizational health. For each organization, I've summed up the five-year wins above replacement totals for all players aged 25 and under–these are the players who have the best chance to remain with their organizations in the medium term. It isn't a perfectly objective standard since we haven't run a PECOTA for every last player; still, with nearly 1900 players covered, these should serve as a mighty interesting leading indicator. (Note that I'm only counting seasons of positive value over replacement in generating the totals; there's no need to punish an organization for developing a player who would be unlikely to be in the majors to begin with).

Starting from the top of the list:

              Hitting  Pitching  Total
Toronto       118.5     18.2     136.7
Cleveland     93.5      35.0     128.5
Chicago (N)   73.6      26.1     99.7
Oakland       49.3      47.9     97.2
Minnesota     71.1      23.9     95.0
Atlanta       74.8      18.6     93.4

There are some organizations that stand head and shoulders above the rest in terms of the quality of young talent they have in their systems, and the Blue Jays and Indians are at the head of the class. Oakland has lapped the field in terms of pitching, and that's before counting Tim Hudson, who was a year too old to make the study but is signed cheaply through 2005. Atlanta's farm system sometimes gets overlooked, but it has populated a dynasty that has lasted for more than a decade, and the Braves continue to be the best in the game at leveraging a traditional scouting approach into tangible results.

                Hitting  Pitching  Total
Detroit         76.4      13.5     89.9
Colorado        58.3      31.5     89.8
New York (A)    73.0      15.8     88.8
Texas           78.2      8.7      86.9
Chicago (A)     56.5      29.9     86.4
Florida         54.5      31.8     86.3
Cincinnati      68.8      12.3     81.1

It's good to see the Tigers fare well by at least one metric. Sure, having a lot of young talent is more a necessity than a luxury in their case, but the team has drafted well in the past few seasons, and it's showing up in terms of players who are performing slightly ahead of their age cohorts at each level. At the moment, the names of most of their youngsters aren't widely known even to prospect hounds, but there's a deep enough talent base that some, like Cody Ross and Rob Henkel, may well be recognized a year from now.

Colorado's showing will surprise some, but they have a pitching-heavy system that paid a big dividend last year in the form of Jason Jennings, as well as a reasonably deep contingent in the minors led by Chin-Hui Tsao. They also have the best prospect that you've never heard of, first baseman Ryan Shealy.

               Hitting  Pitching  Total
Tampa Bay       53.0     22.3     75.3
Pittsburgh      61.6     12.5     74.1
San Diego       54.6     16.9     71.5
Milwaukee       51.1     19.3     70.4
Philadelphia    46.2     24.1     70.3
Anaheim         49.6     19.6     69.2
Baltimore       47.4     20.4     67.8
Kansas City     41.5     24.9     66.4

Placing in the middle of the pack is a perfectly acceptable result for a team like the Phillies, which uses a combination of homegrown and imported talent, but more disturbing for the likes of the D-Rays and the Royals. A tools-heavy approach is supposed to produce a few big hits amid a lot of misses, but PECOTA remains skeptical about the development patterns of prospects that fail to exhibit adequate plate discipline, such as this Tampa triumvirate:

            2003    2004    2005    2006    2007    Total
Hamilton    +0.4    +0.9    +1.6    +2.0    +1.8    +6.7
Crawford    +0.2    +0.9    +1.4    +0.8    +1.4    +4.7
Baldelli    +0.3    +0.8    +0.9    +1.1    +1.5    +4.6

The bottom of the list is populated by big-budget veteran teams that have been able to achieve success in spite of mixed results with in-house player development.

               Hitting  Pitching  Total
Seattle         38.6     25.3     63.9
Los Angeles     55.1     8.3      63.4
New York (N)    44.0     16.7     60.7
Houston         27.5     32.9     60.4
Boston          42.2     14.6     56.8
St. Louis       52.8     1.8      54.6
Arizona         33.7     20.5     54.2
Montreal        17.6     25.9     43.5
San Francisco   25.8     13.5     39.3

Yep, the Red Sox, the Diamondbacks, and…the Expos? As we've pointed out before, a Motel 6 scouting budget and Omar Minaya's itchy trigger finger have rendered the Expos virtually bereft of minor-league talent, especially at the plate. The problem is serious enough that it ought to be considered by whichever kind soul eventually buys the club, as the franchise will require serious investment above and beyond the purchase price to remain even nominally competitive after Vladimir's inevitable departure.

It's a testament to PECOTA's distrust of pitching prospects that the Giants place as low as they do. With apologies to Mr. Foppert, Mr. Ainsworth, and Mr. Williams, the Giants' situation somewhat resembles that of the early 90s Athletics, who relied on a pitching-first draft strategy–including the infamous four aces–to cover up for an aging, declining core. An elbow twinge here, a torn labrum there, and the Giants will be ill-equipped to compete once Barry takes up golf, especially since it's a stretch to say that they have a single hitter under the age of 25 who projects to be more than a league-average performer.

A couple of other observations from that list. First, while there are good arguments that the American League has been the shallower of the two circuits for the past few seasons, it is poised to do bit of catching up. American League teams average around 86 wins above replacement over the five-year period, compared to just 70 for the National League clubs, with the gap widening each successive year. With a disproportionate amount of young talent under its grasp, the American League is entirely deserving of the nickname 'junior circuit'.

Second, as I've already alluded to, these results illustrate the divergence between scouting-based and statistically based approaches to player development. When I compared the PECOTA scores to the organizational grades assigned by Baseball America (warning: non-BP premium content), which tend to favor a more traditional approach, the correlation was very weak (+.19). You can argue that there's a chicken-and-egg problem at work: Analysis-friendly organizations acquire players based on their favorable statistical profiles and, lo and behold, those players continue to put up favorable statistics with their new organizations. Still, the Blue Jays and Indians can take comfort in the fact that they're on track to accomplish what they've set out to do, which is a tougher case to make in Tampa, Pittsburgh, or Kansas City.

Thank you for reading

This is a free article. If you enjoyed it, consider subscribing to Baseball Prospectus. Subscriptions support ongoing public baseball research and analysis in an increasingly proprietary environment.

Subscribe now
You need to be logged in to comment. Login or Subscribe
In that AL list, what Williams is that? Can't think of one.
Oh, the memories!

Jose Vidro, Troy Glaus.. They just dropped off the radar at some point.

Would also be cool to see a similar 5-year forecast article written 5 years ago, so we can see how those things look today.
Bernie Williams. My all-time favorite player. Once I even rear-ended his car on the Merritt Parkway.
It's always nice to read old Nate again, but what good is an old article on projections if it doesn't actually compare those projections with reality?

It would be nice if a link to the actual WARs (for these players in this period) were provided.
Yes, please spend at least some time analizing what ended up right and wrong and why.
... and please account for the changes in how WARP is calculated since the article was written. That's the only thing holding me back.
This is a lot of work for the players, but for the teams, it's fairly straightforward, and leads to the following table (I wish I could present this in graph form, but oh well):

Team, predicted WAR, actual WAR
ARI, 54.2, 50
ATL, 93.4, 122
BAL, 67.8, 34
BOS, 56.8, 142
CHA, 86.4, 102
CHN, 99.7, 79
CIN, 81.1, 42
CLE, 128.5, 87
COL, 89.8, 47
DET, 89.9, 41
FLA, 86.3, 78
HOU, 60.4, 95
KCR, 66.4, 0
LAA, 69.2, 119
LAD, 63.4, 91
MIL, 70.4, 46
MIN, 95.0, 112
MON, 43.5, 47
NYM, 60.7, 77
NYY, 88.8, 160
OAK, 97.2, 116
PHI, 70.3, 106
PIT, 74.1, 21
SDP, 71.5, 82
SEA, 63.9, 63
SFG, 39.3, 85
STL, 54.6, 123
TBY, 75.3, -1
TEX, 86.9, 66
TOR, 136.7, 75

The correlation between predicted and real WAR is positive but weak (correlation coefficient = 0.102), with some remarkably accurate hits (Arizona, Montreal/Washington, Seattle) but also some very severe misses. In general I think PECOTA did pretty well in predicting player performance, but this prediction of team performance is nothing to brag about. One could do just as well -- no, not "just as well," rather "a great deal better" -- by just "predicting" that good teams will continue to be good, and bad teams will continue to be bad; the correlation between 2002 WAR and summed 2003-2007 WAR is vastly stronger (correlation coefficient = 0.739).
Super interesting. Seeing Pujols up there seemed natural. Hard to realize that it was un-intuitive. I've still never heard of Ryan Shealy, and the bit about Prior makes me think that the computers are winning.