On Target 2005: How Far Should They Fall?
by Josh Parks
published March 23, 2005

Josh Parks is a Seattle freelance writer and baseball enthusiast.  This is his third year with Baseball Notebook.

Maybe it's just me, but when I look at my ranking sheets lately, it seems like there is something of a bumper crop out there of players with early-round value who are virtually certain to be available in later rounds. In one league I'm in, with smaller-than-average rosters, I think I could just about put together a whole team of such names.

If you read my "On Target" essay last year at this time you know that in preparing for drafts I like to identify these players ahead of time and specifically target when I might try to select them. I consider this the rough equivalent, in draft league terms, of trying to put together a $333 roster with a $260 budget, since you can still in principle under or overpay for your players, depending on when you draft them.

To illustrate, if the average projected first round value (the first 12 players in a 12-team league) on your ranking sheets is $30, and you can draft one of those players in the fifth round, where the average value is $17, the way I see it, you've successfully "underpaid" for that player by about $13, or 44%, as well as denied that player's services to any other team.

As has already been stated frequently in David Luciani's columns - and this applies whether you're a Baseball Notebook member or not - you do not want to draft players in descending order of forecasted value. While you're likely to wind up with at least a few undervalued players that way, you will not maximize the value on your team. The rule of thumb is, when you think the highest valued player available will be around later, skip over that name.

Of course, the question then becomes, once you've skipped over somebody, how far do you let him fall?

After what I'd call decent results identifying target players last year, but so-so results landing them (can you spell Santana?), I wanted to refine that part of the process. Obviously, I can't speak for every player in every situation, but I am going to share the approach I'm using this year for a number of specific players.

Do Try This At Home

The first thing I did, just like last year, was compare my ranking sheet with some mass media rankings - any well known fantasy magazine or Web site will do. I treat those rankings as the results of a draft and I go through my ranking sheets and keep track of how many times certain players are "skipped over" in each "round" (see my essay on this topic from 2004 for details). I like to do it this way because it gives me a clear visual picture of who is being undervalued.

You might wonder whether reviewing one set of rankings is enough to get a bead on whom the undervalued players are. I believe it is, certainly for a first pass. You may occasionally find a player undervalued by a mass media outlet who is more highly valued by the drafting population, but I expect that would be rare. They call it "mass media" for a reason.

Once I identified the players with the biggest gaps between projected value and where they were taken in my "draft," I looked at the publicly available results for 40 different mixed league drafts and kept track in a spreadsheet of where those players were taken (yes, this was a lot of work). Now 40 is not what I'd call a large sample, but this is by nature an imprecise exercise, since you can only get as close to the target picks I suggest as your particular draft order allows. Plus, if I'd gotten the sample size up to 100, that would have typically only shaved two or three points off the margin of error. While more results are always preferable, given that your ability to make use of the increased precision would be limited by your draft order anyway, the effort-to-benefit didn't seem favorable much past 40.

Using my layman's statistics skills, I then calculated the average draft position and standard deviation for the picks. Finally, I calculated which overall draft pick I would expect to be ahead of about 90% of everybody selecting that player. There's nothing magical about 90%. It's just the level I felt most comfortably balanced the risks of drafting earlier than I needed to and missing out on a player. I'll tell you how to use both a more and less conservative approach later on.

So for the players I've identified, I can now comfortably skip over them while knowing how far I can let them fall and still have what I consider to be about a 90% chance of landing them.

Now for the fun part - the players themselves. For this essay I identified five batters and five pitchers who I think are being drafted significantly later than their forecasted value would indicate, and where I would plan to select them. The ranking sheets I used were for a mythical 12-team mixed 5x5 league with a roster of eight infielders, five outfielders, one utility bat, and nine pitchers. The ranking sheet order I refer to in this article is based on the March 20 projection set.

Bat Men

First off, on the batting side, we've got Andruw Jones. According to my sheet, he has the 25th most valuable forecast, although out of the 40 drafts I looked at, the earliest he was taken was 58th (the fifth round of a 12-team league).

So you know you can skip over him early, but how far do you let him fall? According to my numbers, his average draft position is 99.8, which would fall in the ninth round of a 12-team league. Also, the median for him is 97, pretty close to his average, and also in the ninth round. But if I used either of those numbers as a target I would expect Jones to be available only about half the time.

Once again I can use limited stats skills to leap into action. Armed with Jones' standard deviation of 18.2, I can calculate the 90th percentile of his draft picks. In other words I can figure out when I need to pick him and expect to be ahead of about 90% of everybody else. (You don't really want the math, do you? Well, OK, take the standard deviation and multiply it by 1.28, then subtract the result from the average.  Note that this is not the same as calculating the percentage within a normal curve but rather the percentage of people you will beat to the punch in picking the player in question) For Jones the answer is 76.4, which means I would plan on taking him with whichever of my draft picks was closest to the 76th pick overall. (For the sake of presentation, I'm going to round the rest of the numbers to integers, except for the data at the end.)

Once I've got a target draft pick for the player, I look at what overall pick I have for each round and see which comes closest. If I picked sixth in a 12-team league with a snake draft (where the order reverses each round), my closest overall pick to Jones' target would be 78th, in the seventh round. It's particularly important to know your overall draft pick numbers if you pick early or late in a snake draft format, since you are constantly alternating larger and smaller gaps between picks each round, or if your league has more than 12 teams, since all of the gaps will be larger.

As you'll see below, sometimes players end up with effectively the same target number. In those cases you have to decide for yourself who has priority. Obviously, whoever has the higher forecasted value should get due consideration. Also, if a player is coming on strong recently in spring training (and I note that as of March 21, Jones was leading the NL in spring homers and RBIs, and had boosted his average above .300), you might want to factor that in. Or if one fills a more urgent positional need, you could bump him ahead. You'll also want to consider what you know about your own specific league and how it tends to value players.

If you decide you want to land both players who have similar target numbers, realize that will probably mean either taking one earlier than the target number indicates, or later and increasing the risk of losing him. People may scratch their heads, but the point is to land players with early-round value in later rounds. Let them scratch their heads.

Also keep in mind that you probably won't be able to land all of the target players I mention, so figure out who has priority based on forecasted value and your positional needs. Add up different combinations of player values and go with the one that has the biggest total value and whose target picks fit best into your team's actual draft order.

Now back to the bats. Another player who seems to be undervalued this year after a sub-par 2004 is Shawn Green, who qualifies at first base as well as in the outfield in most leagues. He currently has the 27th most valuable forecast in my mythical league, but his average pick is actually about 109th overall. To land him I'd shoot for something close to the 86th overall pick.

Also in the outfield we've got Pat Burrell and Richard Hidalgo. I list these two together since their numbers are practically identical. They're both forecasted as sixth-rounders in my 12-team league, yet are taken on average with the 175th pick. Since the target picks for these two are so close together (132nd for Hidalgo and 136th for Burrell), you'll have to decide for yourself if you want to land them both and who gets priority.

Turning to the infield, nobody's getting less respect than Kaz Matsui. On average he's taken with about the 153rd pick, while my ranking sheet has him as the 44th most valuable forecast. My target for him is 119, making him somebody I'd watch for late in the 10th round or early in the 11th (again, assuming your league has 12 teams).

A Call to Arms

On the pitching side, probably the biggest gap between forecasted value and perceived value is Greg Maddux, largely because David Luciani is forecasting Maddux to have his best year since 2002. My ranking sheets have him with early second-round value, yet, even without knowing his average draft pick is a whopping 145, I hope you know virtually nobody will draft him that high. My target pick for him is 119 (the same as Matsui), although Maddux does appear to be trending a bit earlier lately. Using the draft order I mentioned above, the closest pick I would have would be 115th overall. If I wanted both Maddux and Matsui, this is an example where I'd use the early pick on Maddux, and let Matsui fall to the next round (126th overall).

Just ahead of Maddux, with the 14th most valuable forecast, is Mike Mussina. After a disappointing season last year, Mussina's perceived value has faltered, with an average pick in the middle of the eighth round. The good news is that Mussina has one of the smallest standard deviations of the players I considered, so my target for him would be 73rd overall.

Next up, with the 39th overall forecast, we have John Smoltz. Whether it's the switch back to being a starter, his age, or both, Smoltz is the 93rd player taken on average. Definitely skip him in the first few rounds, but if he's somebody you want, I wouldn't let him fall much past the 67th pick overall.

Journeyman pitcher Odalis Perez is currently forecasted with the 61st most valuable forecast, but with an average pick of 152. As for his target pick, he's yet another option if you're trying to figure out something for right around 119th overall.

How Lowe Can He Go?

Lastly on the pitching side we've got our old friend Derek Lowe, recently named the Dodgers' opening day starter, as well as the most recent Baseball Notebook Newsletter undervalued pick of the week, and carrying my league's 55th best forecast. With a 2004 ERA that more closely resembles the tab for your morning latte and muffin, it should come as no surprise that Lowe's average pick is 211, which you might recognize as being three points higher than Jason Giambi's 2004 batting average. Just imagine where Lowe would go if he hadn't performed well in the playoffs.

That said, my numbers indicate a pretty wide range of opinion on Lowe. I found one draft where he was taken as early as 131st, and another where he lasted until the 290th pick. In addition, as of March 22 Lowe had a 1.59 spring training ERA, so his perceived value may be on the rise, relatively speaking. At any rate, if the possibilities intrigue you, and I think they should, go after Lowe with something close to the 161st pick. If he performs as projected you're going to look pretty smart, but you can still claim you simply took a flyer on him. And if he doesn't finish the season as your league's 55th most valuable player, at least you didn't try to draft him 55th.

If your league has more or fewer than 12 teams, you'd be right to wonder whether any of this can still help you. My hunch is that it can, as long as you go by the overall picks I'm proposing and take the time to figure out in which round those fall in your particular draft. My other hunch is that in smaller leagues I would expect more surprises. For some reason, the draft picking seems a bit more erratic with smaller leagues. For leagues larger than 12, keep in mind that as the gaps between picks get longer, the consequences of skipping a player can increase, so you'll have a harder time landing players whose target picks are close together. Be more selective in whom you target, and perhaps more aggressive as well.

In addition, your league's scoring format should definitely have an impact. Unfortunately, how to address that circumstance is beyond the scope of this article, so if you're in something besides a 5x5 league, take my recommendations with an even bigger grain of salt. If you're not in a mixed league, you'll have to adjust accordingly for the fact that players will obviously go a lot earlier if you're dealing with just one league.

In any event, no matter which scoring format your league uses, or how many teams are in it, you must have a plan and you must make sure you don't take a player too early.  At the same time, don't let him slide to another team just because you don't know consider when someone else might take him.  More important than anything is recognizing that it's okay to let the occasional player slip through the cracks, even if you're projecting a better season for him in 2005 than everyone else.  It's better to miss out on a player who's projected to be valuable than to waste an early pick on him that could have been used for someone with higher perceived value.  Thinking this way will often net you both the superstar and the undervalued player, which will maximize your roster value.

For The Do-It-Yourself Player

If you'd like to calculate your target picks for the players mentioned at more or less than the 90th percentile, I'm providing the necessary information below.

If you'd like to be less conservative (i.e., you're willing to increase the risk that you'll miss out on a player), you can simply subtract one standard deviation from the average pick. Using the result as your target pick, this should put you ahead of a high percentage of the players out there.

If you want to be more conservative and reduce the risk of missing a player, multiply the standard deviation by two and subtract the result from the average. This should put you at about the 98th percentile.