I have that on my main computer instead of Excel. Can that be used or should I try the rich text format?

I think it might be best to produce the top 50, then possibly make a “rest of” list showing in order players that got the most votes but didn’t make the top 50. I think making it an actual top 100 list could look kind of skewed, but maybe I’m wrong.

I suggest 100. People tend to get weird with 50 (a relatively low number of prospects), as they start to put prospects who aren’t quite of appropriate caliber at the end of the list, simply because they really wanted a particular player or two to get some love. Challenging the community to dig deeper than just listing “50 good prospects” is a good way to maximize the usefulness of the results. I imagine if you go to 100, you’ll see some interesting names on there that might surprise some viewers.

Frankly, I’m just not really sure how interesting a top 50 would be.

Email me if you have any trouble manupulating the data in excel and want some help: tje24@drexel.edu

I might need to talk to someone to walk me through all of the 37 steps involving in posting my 50 top prospects, do you have a 24 hour number we can call?

how are you going to handle aggregating the rankings of somebody who made some but not all lists?

Maybe in a separate post, you can post a vote on how many prospects people would want to rank. There does seem to be enough interest.

There’s a logically sound best way, and almost everyone gets it wrong.

Most people give 100 points for a #1, and so on down to 1 point for a #100, and simply add them up. But this asserts that a #1 prospect is 100 times as valuable as a #100 prospect, and that only 100 prospects have value. (Or that Joe Mauer or Dustin Pedroia was 10 times as valuable as whoever was 10 on your MVP ballot, or No Country for Old Men was 10 times as good as Away From Her).

So what needs to be done is to start at 110 or 120 or 150 points and run down to 11 or 21 or 51. Which, respectively, assert that a #1 prospect is 10 times, or 6 times, or 3 times as valuable as a #100 prospect.

Of course, which scale to use is an open question. Fortunately, you don’t have to decide this in advance. Simply count the points using a 100 to 1 scale (Pts), and separately count the number of voters (Votes) that named each prospect. The final ranking is then K * Votes + Pts, where K is the size of the skew you are adding to the 1 to 100 scale to make the results fair.

You are going to want to pick a value for K yourself and report the results that way, but you will also be able to provide a spreadsheet with the separate Votes and Pts figures, and anyone with rudimentary Excel skills will be able to disagree with your choice of K and re-rank the results.

In the meantime, I think I’ll quickly try to determine K objectively. I’m going to look at the career Win Share values of prospects 1-5, 8-12 (because there may be some non-linearity right at the top) and 96-100 for all the BA top 100 lists that are old enough. That should actually answer the question of how many times more valuable a #1 or #10 prospect is than a #100. I’ll report back in this thread.

K = 20. If you give 120 points for a #1 ranking and 21 points for a #100, you will get an almost perfect valuation for prospects 10 through 70 or more. You will undervalue the top 5 prospects and the bottom 20 or 25, but I think you have to live with that.

I looked at the top 5 and bottom 5 and #10, #40, and #70 (i.e., 8-12, 38-42, and 68-72) prospects from the oldest 10 BA Top 100 lists (1990-1999), and calculated average career WS for each prospect. There is, as you would expect, a sigmoid curve. There is an almost perfect linear relationship for the value of #10, #40, and #70 prospects (r^2 = .992), but top 5 and bottom prospects are above that trendline. And quite remarkably, one spot in the rankings is worth almost exactly 1 career WS point. Here are the average career WS points for the 5 group of prospects (using the center ranking of each group):

#3: 148
#10: 115
#40: 79
#70: 52
#97: 43

If you use K = 20, you get the following (with the desired results, smoothing the above data for the three central points)

#3: 118 (should be 148)
#10: 111 (sb 113)
#40: 81 (sb 82)
#70: 51 (sb 51)
#97: 23 (sb 43)

The matches for 10, 40, and 70 are terrific. The top handful of prospects will be underrated, but the problem with “fixing” that is that’s it’s unclear where you start giving the boost. Ditto for the tail — everyone knows that the difference in value becomes a lot less at the bottom of the list, but it’s unclear where that effect kicks in. True sigmoid curves change their slopes gradually, and instituting that correctly would take more math than it’s worth.

So I think using a 120 to 21 point scale will certainly be accurate enough for our purposes here.

We have a few entries already. Keep ’em coming, guys (and gals!).

To check the eligibility of some guys who are very likely to be on everyone’s list if they are eligible:

Junichi Tazawa (Japanese pitcher likely signing with American team)
Dayan Viciendo (Latin American player, signed by White Sox)
Aaron Crow (drafted, did not sign, signed by an independent league team)
Matt Harvey (drafted, did not sign, likely high first round pick in two years)