The Greatest Top 27 Twins Prospect List Ever...

by B-Ville on Feb 20, 2009 11:05 AM EST Comment 6 comments

… or quite possibly the greatest waste of time ever. With all the top Twins prospect lists out there from various bloggers and websites, I thought it would be interesting to use some of those to compile an average list. I used lists from Aaron Gleeman, Baseball Prospectus, John Sickels, Seth Stohs, Josh Johnson, Baseball America and the community list here at Twinkie Town. The rankings from each of those respective lists are listed in that same order below alongside each player. The average list is somewhat flawed because we don’t know where the prospects outside of the top 10 or 11 would have ranked on the BA and BP lists because they only go to 10 and 11. I gave guys who missed lists an NR for that list and threw that ranking out of their average, which clearly helped a few guys. Swarzak, who ended up at #7 for example, didn’t make the top 11 in BP’s list so I gave him an NR instead of what would have been a 12 or above had the BP list been longer, which would have dropped him down the average list. I stopped at 25 (27 w/ two ties), because the rankings really started to vary from list to list around that area… and because I felt I had wasted enough time at that point. I’d probably swap a few players around here and there, but a pretty solid list overall. Interesting to see the tiers formed by the avg ranking # for each prospect.

1. Hicks - 1, 2, 2, 1, 1, 1, 3 – 1.57

2. Revere - 2, 1, 1, 2, 2, 2, 2 – 1.71

------------------------------------------------------------------

3. Valencia - 6, 3, 5, 4, 3, 5, 1 – 3.85

------------------------------------------------------------------
4. Ramos - 3, 5, 3, 7, 4, 3, 5 – 4.28

5. Morales - 4, 4, 11, 3, 6, 10, 4 – 6.00

------------------------------------------------------------------

6. Hunt - 5, 6, 4, 16, 8, 7, 11 – 8.14

7. Swarzak - 7, NR, 16, 5, 9, 6, 7 – 8.33
8. Mulvey - 12, 7, 7, 8, 7, 8, 12 – 8.71
9. Mijares - 10, NR, 6, 22, 5, 4, 6 – 8.83
10. Robertson - 8, NR, 13, 6, 10, NR, 8 – 9.00
------------------------------------------------------------------

11. Guerra - 15, 10, NR , 12, 12, NR, 9 – 11.60
12. Gutierrez - 13, 9, 8, 10, 9, 18, 16 – 11.85
13. (tie) Parmelee - 11, NR, 10, 13, 13, NR, 15 – 12.40
13. (tie) Hughes - 17, NR, 15, 9, 11, NR, 10 – 12.40
14. Manship - 9, 11, 18, 14, 14, NR, 14 – 13.33
------------------------------------------------------------------
15. Plouffe - 23, NR, NR, 11, 15, NR, 13 – 15.50
------------------------------------------------------------------
16. Bromberg - 14, 8, 12, 25, 26, NR, 24 – 18. 16
17. Delaney - 18, NR, 9, 24, 16, NR, 25 – 18.40
18. McCardell - 16, NR, NR, 18, 19, NR, 22 – 18.75
19. Slama - 19, NR, 14, 27, 17, NR, 17 – 18.8
20. Tolleson - 20, NR, 17, 19, 21, NR, 20 – 19.40
21. Tosoni - 26, NR, NR, 15, 20, NR, 21 – 20.50
------------------------------------------------------------------
22. Duensing - 25, NR, NR, 20, 25, NR, 19 – 22.25
23. Benson - 21, NR, NR, 17, 27, NR, 26 – 22.75
------------------------------------------------------------------
24. Pridie - 29, 19, NR, 31, 22, NR, 27 – 25.60
25. (tie) Burnett - 22, NR, NR, 30, 23, NR, 29 – 26.00
25. (tie) Martin - 27, NR, NR, 21, 28, NR, 28 – 26.00

0 recs | Comment 6 comments |

Story-email Email Printer Print

Comments

Terrific list

I think it’s really helpful to compile stuff like this together. There’s an old story about an experiment that was done in England decades ago where everyone at a county fair could view a cow and write down their estimate of it’s weight. Whoever got closest won a prize. What the researchers discovered was that while very few people actually guessed the correct weight, when taking all the guesses together and averaging it out the collective guess was within a 1/2 pound of the correct weight. My point is that the Twinkietown list and this list are probably pretty accurate, not only in player positioning but also for relative groupings of players.

My only other thought would be to look at the median—might help differentiate some of the ties.

by biggity2bit on Feb 20, 2009 12:44 PM EST reply actions

Ahhhh mean, median, mode...

… forgot about those. (Makes me sad to think about all of the math I’ve forgotten over the years.) Good point on the median. It would help with some of the outliers & ties. As I made it further down the list(and into my work day!) I thought about doing something to adjust for the outliers. Now I suppose I’ll have to spend more time on this.

by B-Ville on Feb 20, 2009 2:01 PM EST up reply actions

More numbers

Alright, so I burned a little time at work and run some numbers on the top 15 (through Manship).

The median was interesting, in that it is the point at which 50% of the data set is above it, and 50% is below. This reduces the impact of statistical outliers, and in this case makes an argument for how players should be grouped. Using the median, the top 15 prospects are:
median value, player name

1. Hicks
2. Revere

4. Valencia
4. Ramos
4. Morales

6. Mijares (previously ranked 9th)

7. Hunt
7. Swarzak

8. Mulvey
8. Robertson

10. Gutierrez
11. Hughes
12. Guerra
13. Parmelee
14. Manship

The other set of numbers I ran was a simple standard deviation for each player’s rankings. What the standard deviation measures is how variable a given data set is—the higher the number, the more it varies. In the case of these player rankings, a player with a low standard deviation AND a low ranking can serve as a benchmark, as it means that there is high agreement that this is where this player belongs (ex. Parmelee). On the flip side, a high standard deviation means that there is probably lots of talent there, but also some pretty serious questions (ex. Mijares). With that said, here are the standard deviations for the original top 15 rankings:

1. Hicks – .787
2. Revere – .488
3. Valencia – 1.676
4. Ramos – 1.496
5. Morales – 3.215
6. Hunt – 4.14
7. Swarzak – 3.983
8. Mulvey – 2.289
9. Mijares – 6.765
10. Robertson – 2.646
11. Guerra – 2.302
12. Gutierrez – 3.891
13. Parmelee – 1.949
13. Hughes – 3.435
15. Manship – 3.077

And finally, if we break the top 15 into three groups of five, we find the standard deviation for each group as a whole:

1-5: 1.06
6-10: 1.76
11-15: .80

Which means to me that there’s a pretty high level of agreement of who belongs in the top 5, 10, etc. etc.

by biggity2bit on Feb 20, 2009 5:08 PM EST up reply actions

Great Stuff

Thanks. I started working on the median, but you’ve done a better job. It’s good to see the standard deviation numbers. It makes a lot of sense seeing guys like Mijares and Hunt having higher numbers there.

by B-Ville on Feb 20, 2009 6:11 PM EST reply actions