June 9, 2005
Doctoring The Numbers
The Draft, Part Five
I had hoped to complete my series on the draft before, you know...the draft. That didn't happen, but that doesn't mean we can't soldier on. Enough with the whole high school/college thing. Let's just jump into the pitcher vs. hitter angle, shall we?
Here are the 15-year WARP lines for all pitchers and all hitters, from 1984 through 1999:
Player Y0 Y1 Y2 Y3 Y4 Y5 Y6 Y7 Y8 Y9 Y10 Y11 Y12 Y13 Y14 Y15 TOTAL H 0.00 0.04 0.16 0.33 0.51 0.68 0.84 0.89 0.99 0.92 0.83 0.81 0.63 0.65 0.48 0.38 9.13 P 0.01 0.08 0.24 0.48 0.65 0.71 0.76 0.74 0.68 0.62 0.64 0.55 0.49 0.43 0.40 0.34 7.83As you can see, for all years and all players, hitters do enjoy a modest advantage of about 17% over pitchers drafted in the same slot. (Although, if you look at the data through Y10 only, the advantage is just 10%.) This is not surprising; TNSTAAPP didn't come out of nowhere, after all. If anything, the advantage was a little smaller than I would have suspected.
Let's break the data down into early (1984-91) and late (1992-99) groups:
Player Y0 Y1 Y2 Y3 Y4 Y5 Y6 Y7 Y8 Y9 Y10 Y11 Y12 Y13 Y14 Y15 TOTAL H, -91 0.00 0.07 0.23 0.37 0.51 0.67 0.82 0.91 0.98 0.86 0.83 0.80 0.61 0.65 0.48 0.38 9.15 P, -91 0.01 0.11 0.32 0.56 0.73 0.78 0.86 0.75 0.71 0.68 0.64 0.59 0.53 0.43 0.40 0.34 8.45 Player Y0 Y1 Y2 Y3 Y4 Y5 Y6 Y7 Y8 Y9 Y10 TOTAL H, 92+ 0.00 0.02 0.08 0.27 0.51 0.69 0.88 0.88 1.00 1.05 0.83 6.21 P, 92+ 0.00 0.05 0.16 0.41 0.57 0.65 0.63 0.74 0.62 0.49 0.63 4.95The advantage enjoyed by hitters of all stripes, which was a mere 8% in the early half of our study, increased to 25% in the second half of the study, even though the later subset of player is missing years 11 through 15, which significantly favored the hitters in the early group.
This certainly would lend some credence to TNSTAAPP theory; teams may have gotten better at drafting high school talent in the 1990s, but they showed no signs at cracking the code for how to identify which pitchers will go on to stardom and which ones will break down.
Let's break the data down into its four component groups, separating players into high-school pitchers, high-school hitters, college pitchers and college hitters. First, the composite data:
Player Y0 Y1 Y2 Y3 Y4 Y5 Y6 Y7 Y8 Y9 Y10 Y11 Y12 Y13 Y14 Y15 TOTAL HS H 0.00 0.00 0.02 0.12 0.24 0.42 0.59 0.69 0.75 0.76 0.77 0.70 0.52 0.53 0.26 0.17 6.54 HS P 0.00 0.00 0.01 0.20 0.40 0.51 0.58 0.64 0.67 0.55 0.57 0.60 0.48 0.49 0.43 0.34 6.47 COL H 0.00 0.10 0.33 0.60 0.90 1.04 1.20 1.17 1.31 1.11 0.90 0.98 0.79 0.83 0.79 0.68 12.75 COL P 0.01 0.12 0.37 0.63 0.79 0.81 0.85 0.84 0.69 0.62 0.67 0.52 0.49 0.36 0.33 0.27 8.37The first thing that jumps out is the huge difference in value between college and high-school picks; remember, this data looks at all players drafted from 1984 to 1999. But while hitters are, on the whole, significantly more valuable than pitchers when looking at the collegiate pool, there's essentially no difference whatsoever in the value of high-school hitters vs. high-school pitchers.
Before we jump to any conclusions, though, I'm going to break the data down into rounds: Round 1 refers to the first 30 picks in the draft, Round 2 refers to picks 31-70, and Round 3 refers to picks 71-100. Furthermore, I'm going to break the data up to look at 1984-91 separately from 1992-99. Here's how the numbers shook out between 1984 and 1991:
Player 1st Rd 2nd Rd 3rd Rd Average HS H, 84-91 16.13 2.11 1.03 5.99 HS P, 84-91 7.74 6.98 2.51 5.87 COL H, 84-91 26.24 5.91 9.94 13.22 COL P, 84-91 18.48 6.79 4.14 9.50College hitters taken in the first three rounds are significantly better values than college pitchers, who are significantly better value than either species of high-school talent.
Look at the data round by round, though, and a much more interesting pattern emerges.
First off, it's pretty easy to see why the mantra of Thou Shalt Not Draft A High School Pitcher in the First Round came to pass The 15-year WARP value of high-school pitchers taken in the first round is less than half that of any other subset of draft picks. That is an enormously valuable piece of information to know. The spectacular failure of high-school pitchers almost entirely explains the disparity between college and high-school players in the first round; as you can see, high-school hitters were almost exactly as valuable as college pitchers.
It's in the second round where the rules are turned upside down. High-school pitchers have the most value of any group of players drafted in the second round. That's shocking. For every Kurt Miller or Dan Opperman who goes bust as an early first-round pick, there's a second-round gem like Greg Maddux, Tom Glavine, Arthur Rhodes or Al Leiter.
Overall, high-school pitchers drafted in the seconnd round are almost as valuable as their counterparts in the first round. Compare this to college players, where first-round picks were worth between three and five times as much as second-round picks. Or better still, compare that to high-school hitters:
Player 1st Rd 2nd Rd Ratio HS Hitters 16.13 2.11 7.64 HS Pitchers 7.74 6.98 1.11Slight difference there.
Overall in the second round, high-school pitchers, college pitchers, and college hitters all yielded roughly equal amounts of value, while high-school hitters averaged just a third as much value as the other three groups. It's not an exaggeration to say that between 1984 and 1991, the only pick worse than taking a high-school pitcher in the first round was taking a high-school hitter in the second.
Theoretically, if you took high-school hitters in the first round and high-school pitchers in the second round, you could beat the overwhelming odds in favor of collegiate hitters of that period. But your luck ran out in the third round; both cohorts of high-school players got smoked by both college groups. In fact, while it's almost certainly a reflection of small sample size, college hitters were better values in the third round than they were in the second. Guys like John Olerud, Ken Caminiti, Dave Justice and Luis Gonzalez all used the third round as a springboard to stardom (although Olerud fell to the third round in part because he wasn't considered signable.)
To put it in stark terms: on average, the typical college hitter selected in the third round was more valuable than the typical high-school pitcher taken in the first round.
Of course, that's all ancient history. Here's the data from 1992 to 1999. Keep in mind we're looking only at the performance for the first 10 years after each player was drafted, compared to the first 15 years above, so the overall value of these picks is lower.
Player 1st Rd 2nd Rd 3rd Rd Average HS H, 92-99 10.72 3.14 1.76 5.00 HS P, 92-99 9.40 3.20 1.61 4.52 COL H, 92-99 17.28 6.07 3.07 8.53 COL P, 92-99 10.95 4.11 1.11 5.26Once again, college hitters rule the roost. Between 1984 and 1991, the average college hitter had 39% more value than the average college pitcher, which was the next most valuable commodity. Between 1992 and 1999, that edge actually rose slightly to 62%.
Which leads to Draft Rule #8:
Draft Rule #8: Even though the traditional collegiate edge nearly evaporated in the 1990s, collegiate hitters remain far and away the most valuable draft pick on average, enjoying a substantial edge on every other class of draft pick for the entire duration of our study.
You will note, though, that the reason college hitters increased their edge on college pitchers was not because college hitters became better draft picks, but because college pitchers fell back to the pack. Between 1984 and 1991, collegiate pitchers were between 59 and 62% more valuable than high-school hitters and pitchers, respectively. Between 1992 and 1999, that edge dropped all the way to 5% and 16%.
We can isolate the reasons for why the gap closed by directly comparing the 1984-91 data to the 1992-99 data. For this comparison, we need to look at the Y10 data for the early group as well as the late group, so that we're comparing apples to apples.
Player 1984-91 1992-99 % Change HS Hitters 4.31 5.00 +16% HS Pitchers 3.43 4.52 +32% COL Hitters 9.53 8.53 -10% COL Pitchers 7.63 5.26 -31%Both sets of college players lost value from the early years to the later years, and both sets of high-school players gained value. However in both cases, the talent drain from college to high school affected hitters less than it did pitchers. As a result, high-school pitchers no longer present the unacceptable risk they posed in the 1980s, while college pitchers are not significantly more valuable than high-school players of any stripe.
If we look solely at first-round data, the change in draft values becomes even more apparent:
Player 1984-91 1992-99 % Change HS Hitters 10.82 10.72 - 1% HS Pitchers 4.62 9.40 +103% COL Hitters 17.82 17.28 - 3% COL Pitchers 14.39 10.95 - 24%Both college and high-school hitters are almost exactly as valuable in the 1990s as they were in the 1980s. College pitchers, who in the 1980s enjoyed a comfortable advantage over high-school hitters, lost a quarter of their value and are now essentially tied with high-school hitters when it comes to first-round value.
The most striking change has been with high-school pitchers. Once upon a time, these were disastrous picks, earning less than half the value of any other subset of first-round pick. High-school pitchers more than doubled their value in the later set of drafts. From 1992 to 1999, while they were still the riskiest picks overall, high-school pitchers were just 12-14% less valuable than high-school hitters or college pitchers.
The old mantra is dead. Thou Probably Should Not Draft a High School Pitcher in the First Round, but Thou Can.
Here's the second-round data, again comparing Y10 data:
Player 1984-91 1992-99 % Change HS Hitters 1.42 3.14 +121% HS Pitchers 4.17 3.20 - 23% COL Hitters 4.01 6.07 + 51% COL Pitchers 5.29 4.11 - 22%In the second round, college hitters continue to dominate, amassing 48% more WARP than college pitchers and nearly double the value of high-school players. Notice how the values of high-school hitters and pitchers converge. Whereas in the early group, high-school pitchers were nearly three times as valuable as-high school hitters, now the two groups are essentially equal. The most likely explanation I have for this is simply the vagaries of small sample size; we've broken down the data into so many parts that a few outliers in a single group can skew the data one way or the other.
Certainly, the newer data set is smoother and seems more intuitively correct. Yet once again, it appears that the time-honored criticism of high-school pitchers is no longer relevant. High-school players do appear to be riskier than college players in the second round, but the blame for that is shared equally among hitters and pitchers.
Finally, the third-round data:
Player 1984-91 1992-99 % Change HS Hitters 0.69 1.76 +155% HS Pitchers 1.50 1.61 + 7% COL Hitters 6.75 3.07 - 55% COL Pitchers 3.22 1.11 - 66%The third round is where the gains made by high-school players are most obvious. High-school hitters, who were almost worthless in the early group, more than doubled in value, while both collegiate cohorts dropped in value by over half. While college hitters are still the most valuable subset overall, college pitchers were actually the least valuable of the four groups. This is the only instance in the entire study in which one of the college groups was the least valuable of the four.
That's a lot of numbers to throw at you. Let's sum up the important findings here:
If I wanted to sum all five articles into 40 words or less, I could do it like this:
You're going to get about 50% more value from a college hitter than from any other draft pick. High-school pitchers are somewhat riskier than other picks in the first round.That's all you need to know.
So after poring over the musty archives of draft history in the search of some piercing wisdom, what we have found is…there isn't that much wisdom to be had. Teams have done a pretty good job of ferreting out the inefficiencies in the market on their own.
Then again, we're not done searching. Next time, we'll go position by position and see whether there are any inefficiencies to be exploited on a smaller scale. Are high-school catchers really as bad a selection as their reputation suggests? Do college left-handers pan out more than their less genetically deviant brethren? Stay tuned.