YEAR-END REVIEW: PART 2
published Tuesday, December 9, 2008
by David Luciani

In the first part of this series, we looked at a few of my least favorite forecasts from the pre-2008 set.  In the same way that we explained for that first piece that there was not necessarily a statistical basis for the ones we chose to look at, we'll do the same with some of the projections that I'm most fond of from last season's set.  That is, sometimes, the most statistically correct projections are not the ones that help us most.  For example, take these three actual forecasts from our Opening Day set and compare them to the eventual outcome:

BARTOLO COLON

JUSTIN HUBER

KEVIN CASH

The only thing helpful about the Colon forecast was that his health and status were a complete unknown and we forecasted him to have little value, even though other than the earned runs allowed column, this forecast proved to be very close.  The Justin Huber forecast may very well be the most statistically accurate forecast in the set (not that we're sure of that but it would be pretty hard to beat considering how close every column is) as might the Kevin Cash one but neither forecast really mattered in the context of what it added to our knowledge of the season that hadn't yet been played.  This can even be true of players who got significant playing time.  Take Gary Glover's forecast and actual results for example:

GARY GLOVER

Again, he played a lot but in the categories most people care about, this may be a forecast with a relatively close outcome but one that doesn't really help us.

So, let's look at some forecasts that may have been more useful than anything that was already available to us in the form of a recent performance or multi-year average.  In that respect, one of my favorite forecasts from 2008 was our Kevin Slowey one:

KEVIN SLOWEY

The reason I like this one is that Slowey had a very limited track record to work with before last year, at least in the majors.  He had fewer than 70 big league innings (and had a 4.73 ERA over that span) and had only made it above Double-A ball for the first time in 2007.  I really don't think we could have published a better forecast for him and he performed very much like the pitcher projected and he was projected to be valuable to a modest degree to those who have interests such as for a fantasy team.

Another forecast that would have likely helped a fantasy leaguer value better than recent historical data available before last year was this one:

BRONSON ARROYO

Like Slowey, the above forecast may have underestimated Arroyo's wins but what I particularly like about it was that Arroyo's recent track record before 2008 implied a completely different pitcher than the one we were forecasting.  He had not had an ERA above 4.25 since 2005 and our pre-season comments warned: "Arroyo actually was much worse in 2007 than his ERA reveals, largely because a pitcher who allows as many hits as he did over the same amount of playing time should have an ERA closer to 4.60 or 4.70 than the 4.23 he did have."

While this next one is definitely not the best forecast in the set, some of the concerns we raised in it did begin to come true:

ROY OSWALT

Specifically, we had forecasted Oswalt to start allowing more home runs than his recent career implied he would.  In fact, Oswalt had never allowed more than 18 home runs in a season and was coming off a 14 home run allowed performance in 2007 when we published the above projection and the following comment:

"Important to understanding the forecast is that we believe he got away with a lot of high pitches last year, ones that used to strike hitters out in the 2004-2005 years but which will be hammered if he keeps pitching that way in the long run. Our projection is rooted not only in those concerns but just as much, if not more, because of the way our age model projects the natural progression of his career. Thus, we're projecting a career-worst home runs allowed total which, in turn, leads to the higher-than-usual projected ERA. If we turn out to be wrong about the home run projection, then we'll be wrong about the rest of the forecast."

While we didn't nail the forecast, Oswalt did start slow and ended up with the worst ERA of his career and a new career-high in home runs allowed.

A younger pitcher for whom we actually received some negative feedback for our forecast before the season started was Justin Verlander.  We emphasized that we just didn't think he would pitch as well as he had previously when we published the following projection:

JUSTIN VERLANDER

While it may seem that the ERA is way off, we must point out that his ERAs for the previous two seasons were 3.66 and 3.63 respectively and many readers actually thought our ERA forecast was pessimistic when in fact, it turns out we were still overestimating how he would perform.  Here was the comment published for Verlander before the 2008 season:

"We continue to have concerns about his control that the walk totals haven't yet revealed, because hitters keep swinging at balls and he gets credit for a strike rather than a ball, which will eventually change. This concern has been hinted at in the number of wild pitches and hit batters he's had - he led the league in both columns in 2007. Also, we believe that he should be allowing a slightly higher rate of hits per ball in play as he was within decimal places in this category last year of the rate of Johan Santana and Carlos Zambrano and we don't rate his pure stuff as high as those two pitchers... We could lie and tell readers what they'd expect to see but we'd rather make the best forecast we can and based on our observation of his stuff and control, we actually think he's overachieved a bit in his career so far."

While we said earlier that sometimes the best forecasts from a statistical forecast are not necessarily the most useful, sometimes even ordinary forecasts for average players who are full-timers can be close enough to at least make us pretty happy while still being useful, like these two:

YUNIESKY BETANCOURT

ADAM KENNEDY

Betancourt had hit .289 each of the previous two seasons and Kennedy was coming off a year that saw him play only 87 games while hitting just .219 so both batting averages were a bit tougher to project.

Here's another one that we really can't complain about:

VLADIMIR GUERRERO

While we overestimated the average, we actually projected Guerrero to drop compared to the .324 and .329 averages he had posted the previous two years and our forecast was for his lowest batting average since 2001 and lowest RBI total since 2003.

These next two players are easily two of the most asked-about players year after year and we're pretty happy with the way they performed relative to the forecasts:

ALBERT PUJOLS

ALEX RODRIGUEZ

Pujols' health status was in doubt for the entire pre-season of 2008 and we came within a game of projecting his actual outcome.  He exceeded our batting average forecast but we had projected an increase in power back to more typical territory.  Our A-Rod forecast was actually a popular topic before 2008 because we had projected only 138 games played even though he had never missed that many games since 1999.  Our comment last winter explained:

"A-Rod's forecast was a tough one to make because he's starting to approach that age (he's now thirty-two) where the injury risk begins to crawl up, enough so that it's no longer a high percentage play to forecast so many games played even though he's cleared the 150 game mark for seven straight seasons. Miguel Tejada [in 2007] was in exactly the same boat and actually fell short of our then apparently-low projected Opening Day total of 144 games. It's inevitable that eventually, players will begin breaking down more and it seems to happen to third basemen and shortstops more than other positions. In any case, he's still projected to be extremely valuable and if he exceeds the projected games total and gets as many at bats as he had last year, as many might think he will, then this is basically a forecasted repeat of his 2005 season."

While we overestimated his power some, A-Rod's forecast was exact on both the eventual number of games played and at bats and was close in most other categories, including the stolen base, runs and RBI totals.  Here's a player whose minor league career prior to 2008 helped us project that he would return to Earth a bit compared to his breakout 2007 season:

JACK CUST

Cust was better power-wise than we thought he would be but he was coming off a .256 batting average and had 82 RBI in just 395 at bats in 2007 so we were projecting a relative drop in his performance, which did happen, even with the power performance.

One name who seemed to disappoint readers in 2008 actually performed almost exactly as we had projected with just a bit lower batting average:

HUNTER PENCE

Pence was coming off a .322 season and we didn't expect him to maintain that rate for the long run.  There aren't many things to complain about in terms of how the forecast went there.

Though we underestimated Geovany Soto, this next one was still a tough forecast to create considering he had never had more than 54 at bats in a season in the majors:

GEOVANY SOTO

One player who kept showing up low on our ranking sheets because of our projected batting average was Michael Bourn.  Though we were projecting good speed, we just couldn't justify ranking him well and while we don't consider this a superb forecast, there are elements here that would have adequately steered our readers clear of him, especially since he had hit .277 in limited time a year earlier:

MICHAEL BOURN

He played a lot more than we thought he would but our comments had explained our forecast:

"...that [the projected speed and role] doesn't necessarily make him a candidate to rush out and pick up in fantasy leagues. He's never hit for power in the minors and never cleared the .285 mark above Double-A ball, probably because he has trouble making contact... he's a safe bet for almost 30 steals and his forecast reflects that but he'll have trouble keeping his average up and hates facing left-handed pitching."

Bourn ended up getting even more at bats against lefties than we expected and continued to have the same problem, hitting .190 in 116 at bats in that respect.

Perhaps the best way to finish up this segment of the series is to focus on probably my two favorite forecasts of 2008, even if technically they're not the ones that turned out closest to the eventual outcome.  Let's start with this one:

JOSH HAMILTON

Hamilton's a player I formed an entirely new opinion about in recent years.  Up until about a year and a half ago, I never did see him as the top prospect that everyone else did, especially when he was first drafted, but around mid-2007, I could see that he had really developed into an excellent player.  Though our forecast may seem a bit light, we were projecting a monster player here and the main difference is that he got a lot more playing time than we thought (>130 at bats) which means we slightly overestimated his power.  Considering he had just one big league season in the books, a 19 home run performance in 298 at bats with Cincinnati in 2007, we're pretty pleased with this one and we had tagged Hamilton as a key sleeper last spring.

This next and final forecast we'll focus on has a story to go with it:

MILTON BRADLEY

We had forecasted Bradley to set a new career high in home runs (he had never hit more than 19 and hadn't done that since 2004) and new highs in runs scored and RBI with a batting average well above his pre-2008 career mark of .273.  A long-time reader of ours had signed up in December or January before last season and actually sent us an email with the subject "Milton Bradley" and went on to ask for a refund of his 2008 membership based on this forecast alone, deciding we were just too far out on a limb here to take any of our forecasts seriously.  We granted the refund and he went on his way never to be heard from again.  While not all risky forecasts turn out so well, it was satisfying to be lucky on this one, even if we did overestimate his steals.

The final part of this series (to be published late next week) will focus on aggregate results.  In truth, that's really the best way to look at a forecast set, even if taking a handful of the best and worst can be a fun exercise.