Example of the Drain that Power Only Players Have on a Team
A while ago, I ranked 2B according to their possible 2012 return using the Bill James Projections. There was some discussion in the comments on how detrimental it is to have a low AVG and SB number with a power hitter. I have decided to rank some 2B ‘roto style’ to help show how players that only have value in RBIs, Runs and HRs can be a drag on a team.
In the discussion, it was stated that HRs from 2B are nice and players like Dan Uggla are extremely valuable. Other people pointed out that a horrible AVG and effectively 0 SB make him a liability. While most teams can be balanced through multiple players, having one player not pulling his weight in a couple of categories can be a huge drag.
I picked 20 top ranked 2B going into the 2012 season to examine. I took each player’s stats rated for 600 PAs. I am not a huge fan of the Bill James projections, but they will work just fine for the example being shown. I ranked each player from 1 to 20 in each of the 5 roto categories with the 1st place person getting 20 points and so on down with the 20th rated player getting 1 point. The totals were added up and here are the results using just HRs, RBIs, and Runs.
| Rank | Name | Runs | RBI | HR | Runs | RBI | HR | Total |
| 1 | Dan Uggla | 85 | 82 | 29 | 14 | 17 | 20 | 51 |
| 2.5 | Chase Utley | 89 | 83 | 22 | 16.5 | 18.5 | 15 | 50 |
| 2.5 | Robinson Cano | 86 | 90 | 22 | 15 | 20 | 15 | 50 |
| 4 | Ian Kinsler | 100 | 72 | 23 | 20 | 12 | 17.5 | 49.5 |
| 5 | Ryan Raburn | 81 | 83 | 23 | 11 | 18.5 | 17.5 | 47 |
| 6 | Jason Kipnis | 95 | 74 | 19 | 19 | 14.5 | 12.5 | 46 |
| 7 | Danny Espinosa | 80 | 71 | 24 | 9 | 10 | 19 | 38 |
| 8 | Rickie Weeks | 94 | 61 | 22 | 18 | 4.5 | 15 | 37.5 |
| 9 | Ben Zobrist | 82 | 74 | 17 | 12.5 | 14.5 | 9.5 | 36.5 |
| 10.5 | Dustin Pedroia | 89 | 68 | 16 | 16.5 | 8 | 7.5 | 32 |
| 10.5 | Brandon Phillips | 82 | 71 | 17 | 12.5 | 10 | 9.5 | 32 |
| 12 | Howie Kendrick | 80 | 74 | 14 | 9 | 13 | 5.5 | 27.5 |
| 13 | Kelly Johnson | 79 | 63 | 19 | 7 | 6 | 12.5 | 25.5 |
| 14 | Neil Walker | 71 | 81 | 14 | 2.5 | 16 | 5.5 | 24 |
| 15 | Ryan Roberts | 78 | 65 | 18 | 5.5 | 7 | 11 | 23.5 |
| 16 | Aaron Hill | 72 | 71 | 16 | 4 | 10 | 7.5 | 21.5 |
| 17 | Brian Roberts | 80 | 53 | 9 | 9 | 3 | 2.5 | 14.5 |
| 18 | Martin Prado | 78 | 61 | 12 | 5.5 | 4.5 | 4 | 14 |
| 19 | Dustin Ackley | 71 | 52 | 9 | 2.5 | 1.5 | 2.5 | 6.5 |
| 20 | Jemile Weeks | 70 | 52 | 4 | 1 | 1.5 | 1 | 3.5 |
The two players that really stick out place are Uggla at #1 and Pedroia at #10. They both seem like they have been swapped in the rankings. Uggla’s power number push him to the top of the rankings by being nearly 1st in all 3 categories.
Now here is the rankings with AVG added in:
| Rank | Name | Runs | RBI | HR | AVG | Runs | RBI | HR | AVG | Total | Rank Change |
| 1 | Robinson Cano | 86 | 90 | 22 | 0.302 | 15 | 20 | 15 | 20 | 70 | 2 |
| 2 | Chase Utley | 89 | 83 | 22 | 0.280 | 16.5 | 18.5 | 15 | 16 | 66 | 0 |
| 3 | Ian Kinsler | 100 | 72 | 23 | 0.271 | 20 | 12 | 17.5 | 10 | 59.5 | 1 |
| 4 | Jason Kipnis | 95 | 74 | 19 | 0.272 | 19 | 14.5 | 12.5 | 11 | 57 | 2 |
| 5 | Ryan Raburn | 81 | 83 | 23 | 0.264 | 11 | 18.5 | 17.5 | 9 | 56 | 0 |
| 6 | Dan Uggla | 85 | 82 | 29 | 0.251 | 14 | 17 | 20 | 2.5 | 53.5 | -5 |
| 7 | Dustin Pedroia | 89 | 68 | 16 | 0.299 | 16.5 | 8 | 7.5 | 19 | 51 | 3 |
| 8 | Brandon Phillips | 82 | 71 | 17 | 0.279 | 12.5 | 10 | 9.5 | 14.5 | 46.5 | 3 |
| 9 | Rickie Weeks | 94 | 61 | 22 | 0.262 | 18 | 4.5 | 15 | 7.5 | 45 | -1 |
| 10 | Howie Kendrick | 80 | 74 | 14 | 0.287 | 9 | 13 | 5.5 | 17 | 44.5 | 2 |
| 11 | Ben Zobrist | 82 | 74 | 17 | 0.262 | 12.5 | 14.5 | 9.5 | 7.5 | 44 | -2 |
| 12 | Danny Espinosa | 80 | 71 | 24 | 0.248 | 9 | 10 | 19 | 1 | 39 | -5 |
| 13 | Neil Walker | 71 | 81 | 14 | 0.273 | 2.5 | 16 | 5.5 | 12 | 36 | 1 |
| 14 | Martin Prado | 78 | 61 | 12 | 0.289 | 5.5 | 4.5 | 4 | 18 | 32 | 4 |
| 15.5 | Ryan Roberts | 78 | 65 | 18 | 0.255 | 5.5 | 7 | 11 | 4.5 | 28 | 0 |
| 15.5 | Kelly Johnson | 79 | 63 | 19 | 0.251 | 7 | 6 | 12.5 | 2.5 | 28 | -3 |
| 17.5 | Brian Roberts | 80 | 53 | 9 | 0.274 | 9 | 3 | 2.5 | 13 | 27.5 | 0 |
| 17.5 | Aaron Hill | 72 | 71 | 16 | 0.256 | 4 | 10 | 7.5 | 6 | 27.5 | -2 |
| 19 | Jemile Weeks | 70 | 52 | 4 | 0.279 | 1 | 1.5 | 1 | 14.5 | 18 | 1 |
| 20 | Dustin Ackley | 71 | 52 | 9 | 0.255 | 2.5 | 1.5 | 2.5 | 4.5 | 11 | -1 |
The two players that dropped the most was Danny Espinosa and Dan Uggla by 5 spots. Dustin Pedroia and Brandon Phillips are now just behind Uggla in the rankings.
Finally, here are the rankings with SB added:
| Rank | Name | Runs | RBI | HR | AVG | SB | Runs | RBI | HR | AVG | SB | Total | Rank Change |
| 1 | Chase Utley | 89 | 83 | 22 | 0.280 | 15 | 16.5 | 18.5 | 15 | 16 | 12 | 78 | 1 |
| 2 | Ian Kinsler | 100 | 72 | 23 | 0.271 | 23 | 20 | 12 | 17.5 | 10 | 18 | 77.5 | 2 |
| 3 | Jason Kipnis | 95 | 74 | 19 | 0.272 | 19 | 19 | 14.5 | 12.5 | 11 | 17 | 74 | 3 |
| 4 | Robinson Cano | 86 | 90 | 22 | 0.302 | 4 | 15 | 20 | 15 | 20 | 2.5 | 72.5 | -1 |
| 5 | Dustin Pedroia | 89 | 68 | 16 | 0.299 | 16 | 16.5 | 8 | 7.5 | 19 | 13 | 64 | 5 |
| 6 | Ryan Raburn | 81 | 83 | 23 | 0.264 | 5 | 11 | 18.5 | 17.5 | 9 | 4 | 60 | -1 |
| 7 | Ben Zobrist | 82 | 74 | 17 | 0.262 | 17 | 12.5 | 14.5 | 9.5 | 7.5 | 14 | 58 | 2 |
| 8 | Brandon Phillips | 82 | 71 | 17 | 0.279 | 14 | 12.5 | 10 | 9.5 | 14.5 | 11 | 57.5 | 3 |
| 9.5 | Danny Espinosa | 80 | 71 | 24 | 0.248 | 18 | 9 | 10 | 19 | 1 | 15.5 | 54.5 | -2 |
| 9.5 | Dan Uggla | 85 | 82 | 29 | 0.251 | 2 | 14 | 17 | 20 | 2.5 | 1 | 54.5 | -9 |
| 11 | Howie Kendrick | 80 | 74 | 14 | 0.287 | 13 | 9 | 13 | 5.5 | 17 | 9 | 53.5 | 1 |
| 12 | Rickie Weeks | 94 | 61 | 22 | 0.262 | 12 | 18 | 4.5 | 15 | 7.5 | 6.5 | 51.5 | -4 |
| 13 | Brian Roberts | 80 | 53 | 9 | 0.274 | 24 | 9 | 3 | 2.5 | 13 | 19 | 46.5 | 4 |
| 14 | Ryan Roberts | 78 | 65 | 18 | 0.255 | 18 | 5.5 | 7 | 11 | 4.5 | 15.5 | 43.5 | 1 |
| 15 | Neil Walker | 71 | 81 | 14 | 0.273 | 8 | 2.5 | 16 | 5.5 | 12 | 5 | 41 | -1 |
| 16 | Jemile Weeks | 70 | 52 | 4 | 0.279 | 29 | 1 | 1.5 | 1 | 14.5 | 20 | 38 | 4 |
| 17 | Kelly Johnson | 79 | 63 | 19 | 0.251 | 13 | 7 | 6 | 12.5 | 2.5 | 9 | 37 | -4 |
| 18 | Aaron Hill | 72 | 71 | 16 | 0.256 | 13 | 4 | 10 | 7.5 | 6 | 9 | 36.5 | -2 |
| 19 | Martin Prado | 78 | 61 | 12 | 0.289 | 4 | 5.5 | 4.5 | 4 | 18 | 2.5 | 34.5 | -1 |
| 20 | Dustin Ackley | 71 | 52 | 9 | 0.255 | 12 | 2.5 | 1.5 | 2.5 | 4.5 | 6.5 | 17.5 | -1 |
Uggla moves down another 4 spots, while Espinosa halts his downward trend. Uggla goes from the top 2B to middle of the road once AVG and SB are added.
Power hitters like Uggla seem good with all the counting stats, but they can be a huge drag if they have extremely below average numbers in AVG and SBs. A respectable value in just one of the categories, like SBs with Espinosa, helps boost how much the player is worth. An owner should always be aware of how much of a drain a ‘power only’ hitter is on a team.
Last word in third paragraph should be RUNS, not AVG…….however, a really nice teaching lesson-type article!
Thanks. Fixed.
“I ranked each player from 1 to 20 in each of the 5 roto categories with the 1st place person getting 20 points and so on down with the 20th rated player getting 1 point.”
I think this is a less useful method than comparing the output in a category to the median output produced by a starter in your league (thus varying by league configuration). I hate Uggla personally and have never owned him in a league, but in this case he is not just ahead of the 2nd place guy, he is projected 5 HR ahead of the 2nd place guy and about 10 HR ahead of the pack in general. All you can really conclude about Uggla is that if you draft him, you probably have more need to balance him with a guy who can give you plus speed/average for their position.
One way to think about this is to convert AVG to a counting stat–hits.
Dan Uggla hit .233 last year. The median 2B with 500 ABs AVG was .269 (Weeks/Zobrist). To make up for Uggla’s BA drain you’ve got to find 22 hits OVER a .269 average to get back to the median. To put that in perspective, its the difference in hits between Zobrist and Brandon Phillips.
This is a pretty bad way to estimate value, actually, unless your league somehow scores you according to each of your players’ rank ordering of various stats for players in their position.
I am exceptionally confused why you didn’t just find a linear weights scheme for their stats? Some stats are inherently more valuable than others. Considering you’ll usually end up with like 800-1000 RBI and about 1/4 that number HR, having an advantage of 5 HR is huge compared to 5 RBI.
By this scheme, a guy at 2B could be projected for 100 HR and 1000 RBI, but still end up around Uggla’s position if he had a worse average and 0 SB. Doesn’t that seem inherently like a broken way of looking at players?
To check on this, I just applied a different methodology to the 2011 data. I would have applied it to the James projections but I didn’t know where to export them from and didn’t feel like hunting for them.
I did the following for 2B with > 450 AB:
1. Take every counting category value x and adjusted it by doing:
x* = x-median(all x for 2B)/median(all x for 2B)
So then each player’s values are the fraction that they are above or below their peers.
2. For average, I calculated the typical average as:
avgAVG = sum(hits for 2B)/sum(AB for 2B)
and the typical AB as:
avgAB = median(AB for 2B)
Using this, I calculated weighted fraction of how far above or below their peers a player landed in avg by:
wAvg = ((avg-avgAVG)/AVG)*(AB/avgAB)
That way you’re looking not only at their average, but it’s weighted by their plate appearances a bit. I normally do something a bit different where I figure out how much it drags my total team average up or down, but you need to estimate your rest-of-team AB and AVG to do that.
3. So now you have everyone’s normalized DISTANCE from the median, rather than just some ranking and you can meaningfully sum the values to determine how good a player is. Summing ranks is entirely meaningless, as per measurement theory. You need at least an interval scale.
The results:
Kinsler 2.797643524
Pedroia 2.085093555
Cano 1.578411404
Zobrist 1.308306573
Phillips 0.742887616
Uggla 0.428107726
Espinosa 0.327558444
Kendrick 0.259815757
Johnson 0.118619769
Hill -0.278994068
Walker -0.294383141
Weeks -0.403721051
Ellis -1.24322222
Andino -1.354214497
Barney -1.619917947
Beckham -1.691286885
Infante -1.758162929
Carroll -2.224260036
So if you were drafting 2B players last year, this would give a reasonable understanding of their total value.
With that said, even this is very flawed because the normalization is wrong. Normalizing against position is stupid, unless you’re in an “all 2B league.” You want to normalize your categories against the average player in all positions.
In which case, my statement of:
———————-
1. Take every counting category value x and adjusted it by doing:
x* = x-median(all x for 2B)/median(all x for 2B)
———————-
Turns into:
———————-
x* = x-median(all x for fantasy-relevant players)/median(all x for fantasy-relevant players)
———————-
That way you’re looking at how much he’s contributing to the category in total. You can also use means, which may actually be more useful than medians. It doesn’t impact the ranking much, and medians lose information in this context. But people like medians (shrug). Generally not a big difference.
Finally, one needs to calculate a replacement-level player for guys who are projected to miss games to know a guy’s true value. A guy likely to play 135 games at an elite level is almost certainly worth more than one who does the same production in 163. This gives guys like Utley a bit more value than otherwise expected.
Why (X-Med)/Med instead of doing Z-Scores [(X-Avg)/StdDev]?
Could do that too, actually. I considered it. I felt that this would be a bit more understandable. For valuation purposes, it would amount to about the same results.
With that said, I’m now leaning highly toward mean actually. If the mean was 16 HR and you had two guys with 20 (2*1.25) HR, that would be as good as one guy with 24 HR (1.5). You can know where a guy will be via the mean and his absolute deviation from it. If you use z-scores, you can’t say that.
While the z-score tells you more about the actual distribution, I’m not sure why we care in this case. So why not use an intuitive normalization factor rather than a non-intuitive one?
I’m no statistician, but I think we care because the value of a 24-HR guy over a 23-HR guy in a set with an average of 20 HRs is extremely dependent on the standard deviation.
Kipnis ahead of Cano just seems so wrong, no matter what method you took to get there….
The part of the method you can blame for that silly result is using Bill James’s projections.
Lack of steals and a low batting average matter in rotisserie — that should be obvious to anyone. What’s important is how to quantify their relative importance, and on that count, simply ranking second basemen by category is terribly inadequate.
B N has the right idea, though that method can be better fine-tuned since we all play different wrinkles of roto ball and median production in one league is not the same as in another. So I use actual yearly stats in my league as a starting point.
Howie Kendrick is projected for only 5.5 HR’s? Really? He has had 10 or more the last 3 seasons?
Ettin, you’re looking at the wrong column.
Nice use of tables! The content was useless.
Was anyone really surprised that Uggla ranked highest in the one table that didn’t include his two weaknesses in a roto-league? I was shocked to see he falls down the rankings when you add in his liabilities?
Wow, ground breaking stuff.
The only way to accurately show the “drain” a player like Uggla would have on a team would be to take a full league standings, and just move the players around, while keeping every other aspect constant, and see how they affect the standings.
Even then, that is subject to the league’s make-up and the competency of the GMs involved. I’m sure there are teams that did well with Uggla and poorly with Cano or Pedroia.