Mastersball

One of the favorite discussion points at this time of the season is what players historically have strong second halves and are therefore acquisition targets and the flip-side, what players usually crash and burn so it is best to get rid of them.  This is a concept that has long bothered me as it is my belief that while examples of such players likely exist, the fact they have done it for a few years is not predictive of it continuing to happen.  That is, I do not feel it is salient analysis to label a player as “first half” or “second half” and strategize accordingly.

The example I have used in the last involves the flipping of a coin.  My argument is if 32 people flip a coin five times, probability dictates that one will flip five heads.  One person flipping five heads is therefore expected within the range of possibilities.  I use this metaphor for the fantasy baseball population.  There are 750 active players at a given time and well over 1200 that are active over the course of the season.  Based on sheer randomness, one of 32 could have a better first (or second) half five years in a row.  That is somewhere between 25 and 35 players which is a pretty decent amount.  And no doubt, the list fantasy enthusiasts come up with each season are replete with these 25 to 35 players.  If you want to extend the metaphor further, one of 64 players will repeat their first or second half performance six consecutive campaigns.  So, I would even contend that if a player displayed a trend for six or even seven years, it fell within the realm of statistically anticipated outcomes.

The thing is, this sort of analysis is done on players who fared better the second half of LAST season, let alone two or three seasons.  If the analysis is spotty over multiple years, it is certainly suspect based on three months of data.  Again, I am not contending such a player does not exist.  There very well may be players that for one reason or another take a few months to get it going or peter out at the end.  All I am saying is it is not sage to look at what Gordon Beckham or Stephen Drew did the second half of last season and target them now, just as it would be a bad idea to get rid of Dan Haren because he always collapses in the second half.

A year or so ago, I made this point in the forums at our friends at Baseball HQ and a couple of guys much smarter than me in this area had their interest piqued and did some of their own analysis.  The forum is private so I cannot share every intimate detail, but using Bayesian probability analysis (Google it, I had to),  they concluded that for at least an individual season, some first and second half splits feel outside of the expected range of performance level.  The player studied was Adam LaRoche, long considered a second half monster.  The study did not perfectly address my specific hypothesis, but it was enlightening in that for at least one season, it could be strongly argued that LaRoche was statistically better in the second half and it was not just a case of skills not translating into results, which is so often the case in these instances.

I will conclude this discussion by doing something I often frown upon and deem hack analysis, and that is providing selected anecdotal examples to illustrate my point.  The difference is others cite a couple of examples as proof of their argument.  I only want to highlight some rather well known players as a means to help convince those not believing the argument that I at least could be correct, as well as point out how some of this due to perception and not reality in terms of skills not wavering, but rather surface stats not being reflective of said skills.

The first player that piqued my personal interest in this realm was Ichiro Suzuki.  Those of you that have played this game for several seasons might recall that when he first came over to the States, Ichiro enjoyed better first halves, thus the smarts always advised getting rid of him at the All Star break or thereabouts.  Here are his first half and second half numbers, using March to June and July to Oct as the cutoffs.

So in his rookie campaign, Ichiro did not display any difference.  But in his second and third seasons, he indeed had markedly better first halves.  I distinctly recall the pundits suggesting he be dealt before the crash and burn in 2004.  Why you might ask?  Simple – I looked at his first half and determined he was snake-bit and was more than willing to take advantage of others hastiness and acquired him everywhere I could, and 2004 was a very good year for me.  Keep in mind, BABIP had yet to become a household acronym.  My point is, I was more focused on what was happening in 2004, and not at all what transpired the latter part of the two previous seasons.  I felt a low BABIP with a still-stellar contact rate was a more reliable indicator of future performance, in this of the improved variety than the fact Ichiro struggled the second half of ’02 and ’03.

The next guy that got my dander up when he was constantly called a second half pitcher was Johan Santana.  What bothered me was his skills remained consistent one half to the next, but his surface stats, most notably ERA happened to be better for a few second halves.  My argument at the time was it is not a sure thing that if you acquire Santana at the break, your ERA and WHIP would benefit.  Let’s take a look at some numbers:

In 2004, Santana’s first year as a full-time starter, he had a markedly batter second half.  His skills were improved across the board after June.  However, there was also some good fortune involved as his BABIP was quite lucky, leading to more opportunities to pitch from the more comfortable wind-up which may have assisted in the better peripherals.

In 2005, the southpaw’s ERA was better in the second half, but his skills were actually better in the first half, at least in terms of strikeouts and walks.  What improved was he kept the ball in the yard post break.  While this could be part skill, it is also part good fortune.  But, since this was now two straight season of a better ERA in the second half, Santana was labeled as someone to go get for the second half.

But alas, look what happened in 2006.  He was the basically the same pitcher both halves, though walking a couple more hitters leading to a slightly elevated ERA and WHIP.  The key is he did not have a BETTER second half as many that season anticipated.   Knowing many would want to potentially overpay for Santana after the break, I drafted him in a couple of trading leagues and indeed got a king’s ransom for him later.  I also had a pretty good 2006.

Real quickly, in 2007, Santana had a much better ERA in the first half, but the difference in skills was not sufficient to account for the disparity, he was a bit unlucky the second half.  Then is 2008, his second half looked better but was actually just a tad luckier.

The bottom line is these were two famous examples of the herd feeling a player’s second half fate could be anticipated based a two year trend.  But that is all it was, a trend, not a pattern.  If someone in your league is willing to overpay for your Mark Teixeira because he is a second half stud, oblige them.  If Dan Haren’s owner is looking to rid their staff of the impending struggles, help ease their mind and open your arms to the Angel’s ace.  You will not regret it.  Well, at least you shouldn’t anyway.

Today we are going to take a look at the All-Value team based on first half performance.  What I did was take the National Fantasy Baseball Championship ADP and assign dollar values per draft spot based on history.  I opted to use the NFBC ADP as opposed to our pre-season projections as a “wisdom of the crowd” exercise.  You will not find a sharper group of drafters anywhere and the sample is sufficient to weed out those using odd strategies that could skew an ADP.  I kept the analysis simple, just calculating the amount the players earned over what was expected based on their draft spot.  All reserves were assigned an expected value of $0.

While it is interesting, not to mention fun to draw conclusions from the results to aid in future drafts, it must be noted that we are only halfway through the season, the trends may change, not to mention there is no guarantee next year’s player pool acts in the same manner as 2011.  That said, there are some trends I am going to keep my eye on with the hope it gives me an edge come next spring.

Here is your FIRST TEAM ALL-PROFIT FIRST HALF SQUAD:

And your SECOND TEAM ALL-PROFIT FIRST HALF SQUAD:

With the caveat that the following are presently observations and do not yet have any tangible application to game theory, here are some quick thoughts:

1.  I am pleased by the fact that there are not as many pitchers at the upper end of each profit list.  One of the dictums most preach is value pitching always emerges.  At least through the first half of the season, my preseason suggestion that as a populace, we are getting better at valuing pitching may have legs.  Though, we do need to keep in mind the specific populace used here is the NFBC drafter, but, so far, so good.  It will be interesting to determine at season’s end if this pattern still holds.

2.  It is curious that there are no catchers that are significantly outperforming their expectations.  I’m not sure yet how to apply this, but it is worth noting.

3.  Something not shown form the above data is the vast majority of the next group of high profit players are outfielders.  This lends credence to notion of leaving a couple of outfield spots available to be considered fungible, in search of one or two of these value players.

We will broach this again at season’s end.  Good luck to everyone, here is hoping you have a bunch of top second half profit earners on your squad(s)!  Later this week, we will take a look at those on the opposite end of the spectrum, those earning the most negative value the first three months.