Richard Justice recently opined on his blog that the Houston Astros are going to take a 'wait-and-see' approach to top prospects Jason Castro and Chris Johnson. Neither will have a job locked up going into spring training and will need solid performances there in order to make the Opening Day rosters.
I like Castro quite a bit and there's been talk around here about whether the Astros should save some money and go with a youth movement in 2010. Castro, Johnson, Tommy Manzella and company would probably struggle at first, but if they're big league material, they will find their way eventually. That's not the point of Justice's blog post, though. The Astros are apparently basing that decision on spring training performance.
My question is this: why do teams base such crucial decision on such a small sample size? Do you really get a good enough look at a player in spring training to determine whether they'd make such an important decision with a player? I'm not talking about just prospects here, but with non-roster invitees and other players. How do you evaluate a player on such a small subset of play?
In the 2009 Grapefruit League, Michael Bourn led the Astros with 103 plate appearances. The rest of the regulars had around 60-70 plate appearances each, including guys like Jason Smith and Chris Johnson. Obviously, Bourn was the Astros' most improved player in the regular season, but had a .685 OPS in Spring Training. Smith had an .885 OPS in 65 plate appearances, which basically locked up a position for him on the opening day roster. Smith, of course, got no hits in 25 at-bats in the majors before being designated to Round Rock.
Of the regulars in the Houston lineup last season, only two had higher OPS numbers in Kissimmee than during the regular season (Lance Berkman and Miguel Tejada). The rest performed much better once the games started counting for something. The question then becomes, why do we use such a small sample size to determine if a player will break camp with the big league team?
I realize the team probably evaluates other things, like preparation, handling of the pitching staff, defense, and lots more off-the-field stuff that can't be quantified. Still, time and time again we see a guy have a huge spring numbers-wise and it changes how the club looks at him. It's a bit of a self-fulfilling prophesy. If the organization likes a player, they'll use his numbers as a justification for breaking camp with him. If they don't, the numbers are a way to make the cut, even though they don't tell us anything useful.
Houston hasn't always been run this way. Under Gerry Hunsicker, the team cut ties with Shane Reynolds, even though he was healthy enough to pitch 167 innings for Atlanta that season. Reynolds had won 103 games as an Astro and been the Opening Day starter. Yet, the team still made an objective evaluation of his talent level at the time and made the cut during spring training. I'm not convinced that this front office can do that yet, since I don't think they have the decision-making autonomy from owner Drayton McLane that Hunsicker did, but they could surprise me.
0 recs | 15 comments
Hunter Pence
Crushed the ball in spring training, but was sent down to the minors, and was called up again a couple months later and continued to hit the ball.
Castro if he’s ready or not, will not start the year in Houston (you can take that to the bank) he will go down for a couple months and be recalled in May or June as part of a money/arbitration move to keep him under the limited number of days to put off arbitration for another year. At that point if they like what he sees they’ll call him up.
Johnson though might be another store, and I’d think they’re more open to putting him into a platoon as opposed to last year. The word last year was that he would not be in a platoon and would be preparing to play every day.
Timothy De Block - November 24, 2009
With Castro, I agree that he’s likely to see Opening Day in Round Rock. My personal feeling is that he won’t see the majors until July 9 or later. That’s the first game of the first home series after he passes the service time mark that would allow him to be Super Two arbitration eligible after 2012. If they don’t call him up until after then, he won’t be arbitration eligible until 2013. It’s also the last series before the All-Star break, so it would seem to make sense.
Johnson in a platoon doesn’t feel like a good baseball decision to me. To put him in a platoon, I’d want to see strong left-right splits. In his minor league career, he shows a weak split in favor of hitting RHP. But in 2009, in Round Rock, he hit significantly better against lefties. Is that a sign that he needs even more seasoning? Also, his K% and BB% don’t seem to have actually improved. I guess if you think he’s a major league player, now is the best time to play him, given our budget woes and quasi-rebuilding status. But I’d rather just have Blum/Kepp holding down the fort over there.
AstroAndy - November 24, 2009
Also re: Castro
ESPN’s Jason Grey suggested that scouts are puzzled that Castro isn’t hitting for more power “given his frame, balance, and the loft in his swing”.
Part of this, as we’ve discussed here, is fatigue. I also seem to recall that Corpus suppresses home runs to right field, despite being above average for homers overall. Given that Castro is a left-handed hitter, this could be part of it, too. I can’t find the right set of park factors to back this assertion up, though.
But the interesting thing about the Grey blurb was that he suggested Castro has more room to add muscle. That could bode well for 2011 if Rex Jones (of the walrus mustache) and Nate Lucero get a hold of him.
AstroAndy - November 24, 2009
If you go to minorleaguesplits.com, look only at Castro’s AA season, and then neutralize his stats, the number of HRs stays the same (3). His OPS goes up slightly, from .750 to .758, indicating that Corpus Christi slightly depressed his offense. Castro does have a substantial left/right split, but surprisingly it is an “opposite split” (hits better vs. LHB). The split exists in both his cumulative and seasonal OPS, but the differential is higher in AA. At AA, the split is .852 (LHP) vs. .712 (RHP) for OPS, and .444 (LHP) vs. .366 (RHP) for slugging. When the park is neutralized in AA, Castro’s slugging rises slightly vs. RHP but stays the same vs. LHP.
clack - November 24, 2009
So maybe it’s an opposite field approach that’s costing him power?
OremLK - November 24, 2009
hard to say....
maybe, but it’s speculation without really seeing him hit.
clack - November 24, 2009
I don’t think it’s just the Astros who use spring training to make cuts based on “small sample size,” but really all baseball teams. It’s inevitable based on the length of spring training and the deadline for setting the 25 man roster. As a practical matter, even if you lengthened the “trial period” for two or three more months, the sample size probably is still inadequate from a statistical standpoint. That’s why the team also has to give a lot of consideration to previous seasons’ performance, whether minor or major league, as well as factors like age. Managers frequently say that they don’t pay attention to stats in spring training, but instead use it to gather coaching opinions (scouting approach) about players. (Was it Jeff Bagwell’s .150 batting average, which wasn’t unusual in spring training for him, or the coaches’ observation of his ability to swing the bat with a bad shoulder, that led both Bagwell and the team to end his last spring training effort?)
A NRI player knows that they have make a huge impression to make the roster; if you excluded the small sample size performance of spring training, then they would never have a chance of making a team. What else can you do? I sometimes think spring training stats are more important as a rationale for players to accept the results of spring decisions. Players are used to competing and accepting the results, no matter whether it is a playoff series or a whole season. So players who don’t have a position sewn up know that they have to compete with several other players, and the players accept the results as based on “merit.” A young position player who bats .100 in the spring, or commits 10 errors in the spring, understands why they didn’t make the team. They can be sent back to AAA, attempt to get back in a good groove, and hope to be called up to the big league team later in the season.
Wade has shown that he will make significant decisions on established players based on spring training. Wade released Woody Williams before spring 2008 was over, based on a terrible spring, and that was harder than Hunsicker’s Shane Reynolds decision, because the Astros had to eat $8 million when Williams was released. Wade released Fernando Nieve after only 3 or 4 spring game appearances last year, which shows that the roster cutdown deadlines force sample size decisions even smaller than a full spring training.
As for Castro, if I was in the Astros’ position, I would tell him that he shouldn’t expect to be on the opening day roster when he goes to spring training (which is what Pence was told in his spring test). Castro needs the AAA time, given his A/AA performance in 09. The small sample size issue comes to the fore with Chris Johnson. Suppose he hits very well in spring training. Do you accept the sample sample size results, or do you base your decision on the bigger sample size of his overall career? I think the decision maker has to place some sort of weighting on the two sets of results (say, 30% spring, 70% AAA results).
clack - November 24, 2009
By the way, with respect to the Jason Smith decision....
which you mentioned, I think that is also a case of a manager who had a preconception about the players too. My recollection is that Maysonet had good numbers in the spring too, and many of us felt that Maysonet rather than Smith should get the roster spot. But, for some reason, Cooper had it in his mind that Jason Smith is a good hitter. Early in spring training, Cooper characterized Smith as a “good offensive player who has to prove to me that he can play defense.” However, that is just the opposite of what Smith’s career statistics showed. He had fielding skills but his offense never was good enough to stay in the majors. Given how little Cooper played Maysonet after he was put on the roster later in the season, I think Cooper had some negative preconceptions about Maysonet.
clack - November 24, 2009
Bagwell
Was notorious for having bad springs.
Timothy De Block - November 24, 2009
I hate to hog the comments on this article....
but I also think the topic is interesting. It seems to me that spring training results will have more significance this year than normal, because the Astros have a new manager and a turnover in the coaching staff. For most of these players, spring training will be a “first impression” on the manager and coaches. (Arguably, this is one of the benefits of hiring an outside manager…getting a new set of eyes on the team’s players.) It can be good opportunity for guys who haven’t impressed in previous springs to turn over a new leaf and make a positive impression on new coaches.
clack - November 24, 2009
here is an article on spring stats which have predictive value....
John Dewan’s article says that a spring training with 100 point increase in slugging percent over the player’s career slugging percent has some predictive power for the upcoming season. This doesn’t apply to pure minor leaguers, though, since it is based on minimum 200 at bats in major leagues.
clack - November 24, 2009
It’s called steroids…
jmike - November 25, 2009
Small sample size objections are totally relevant for baseball decisions based on statistics, but I’d argue that spring training decisions are more likely based on scout’s eyes rather than the statistician’s tables.
You can get better information from a smaller sample size if you’re judging a player’s approach because you don’t have to wait for at-bats to accumulate and for luck to filter out. Fifty spring training at-bats might not give you enough statistical information to make a solid decision, but the batting coach should be able to get a good read on whether a guy is swinging at the right pitches, how good his bat speed is, etc.
Spring training performance is probably also better used to make cuts rather than make affirmative decisions about who should make the opening day roster. If a guy has a reputation for making lots of throwing errors from 3B to 1B, and he makes 5 of them in 5 spring training games, it’s probably safe to say that he hasn’t really fixed that issue…the decision to send him down to AAA is probably a pretty solid one.
This is all probably more true for position players than pitchers, and more for NRI guys with longer histories rather than young guys with only a few years in the org who are trying to crack the big league roster for the first time. With the younger guys, non-performance issues are going to play a larger role…arbitration status, available playing time, current mix of veteraniness/youth already on the squad, organizational depth at that position, etc.
AstroAndy - November 24, 2009
Nicely said
Joe in Birmingham - November 24, 2009
yes, it is.
clack - November 24, 2009
You must Login with your SB Nation account and be a member of The Crawfish Boxes to post a comment.