## Quantifying performance

Moderator: Baseball Moderators

### Quantifying performance

In preparation for this year, I'm thinking a lot about position scarcity and how a certain level of performance at one position is not always equal to that same level of performance at a different position. Ergo, a SS is more valuable than a 1B if their stats are equal. My idea here though is to quantify this concept. Forgive me if this is a bit dorky:

The idea is to put the stats of all qualified players at each position in a distribution (normal?) and then give each player a score based on how many standard deviations they are from the mean. Then add each score from all the stats that you look at (traditional 5x5 for example) to give each player a total score. What I'd imagine you'd see is that players like Chase Utley would have the highest scores because of position scarcity. And I'd now have an easier way to rank players for draft day.

Any analytical minds, tell me what you think; if you would make any improvements to this method.
LloydChristmas
College Coach

Posts: 157
Joined: 6 Jan 2009
Home Cafe: Baseball

### Re: Quantifying performance

Just one more note:

I don't know too much about it, but I have a striking suspicion that WAR and similar stats already do this type of quantification.

What I like about my method is that it's flexible for whatever stats different leagues use.

If someone's versed on sabermetrics, please chime in.

EDIT: I'm investigating more into ESPN's Player Rater. I've liked using it for a long time so I'm gonna try to learn a little bit more about it.
Last edited by LloydChristmas on Mon Jan 31, 2011 10:19 pm, edited 1 time in total.
LloydChristmas
College Coach

Posts: 157
Joined: 6 Jan 2009
Home Cafe: Baseball

### Re: Quantifying performance

First, you don't have to apologize for being dorky on a fantasy baseball message board. It's expected.

Second, it's standard to establish a replacement level for each position and give players value on their stats that go beyond replacement level. That doesn't mean it's the only way to valuate, and it's essentially the same thing but with a different baseline. The issue is that you'd have players in the player pool with negative value. Lots of them. And it might falsely indicate that a player with a value of 10 is worth twice as much as a player with a value of 5, since you're not working on an absolute scale. Really, the player worth 10 is probably only 25% as valuable as the guy worth 5. This is a very simple hurdle to overcome, but there might be others that aren't as obvious. Doesn't hurt to crunch the numbers though, and see how it looks after being tweaked. It's good to stray from the pack in terms of assigning value, but you don't want to be be taking Chase Utley #1 overall and laughing at the next guy for taking a 1B so high. Basically: make sure it passes the sniff test.
Skin Blues
General Manager

Posts: 3082
(Past Year: 82)
Joined: 11 Apr 2010
Home Cafe: Baseball

### Re: Quantifying performance

It's topics like this that make fantasy baseball INFINITELY better than fantasy football... and why it drives me NUTS when people try to replicate football by making their baseball leagues into points leagues...

Two main things that stick out as "unquantifiable" right away, whether you're playing H2H or Roto...

1.) Every league setup creates different player values
You can't create one set of values for players until you have one specific way of how their scoring counts. The second you change anything in the league setup, whether that's min-starts, max-innings, two catchers, add MI/CI, #4/5OF, one or two UT, how many bench spots, etc, etc, etc... it all changes. Likewise how you select those players can change that as well... auctions and drafts require much different preparation. Of course more experienced players usually prefer the auction because of how much freedom it provides to the owners, but the dollar values you assign to players before the auction are going to fluctuate wildly through the auction.

2.) Positional scarcity's value and a player's "overall" value only matter at certain times of the draft/auction
Anybody who's been through a few years of drafts realizes that at some point during that draft you're going to be switching gears and going from "best available" to "best available X", or "any player that can provide me with X stats". Positional scarcity really only controls player values through the first few rounds. After that the runs on certain positions come into play and owners start chasing certain positions to fill out their roster. If you're playing roto and you've got a good idea about your target numbers in each category you're right away looking at what areas you're lacking in and drafting players that will support those categories.

After only a few rounds of the draft the remaining players' values are dictated by the players who have already been selected. There's really no way to account for this numerically before the draft, it's more of a "feel" thing that you develop through mocks and such. Last year after about a dozen mocks I had a good feeling that if I didn't have 1B/3B/OF/OF/SP filled in the first five rounds, I struggled to complete a squad I felt comfortable with. It didn't matter which specific players I got, I was just never happy with my team if I went outside those five positions in the first five rounds. This year will likely be different, of course.

All that being said.... this game is all about stats. If I were to try this method (which I've attempted in the past), I would start with:
A.) Finding a "replacement level" stat-line for each position.
B.) Then I'd look at my statistical targets in each category,
C.) Subtract out the performance of a "replacement level" team,
D.) Take the remaining required stats (that theoretically separate the first-place and last-place teams) and start assigning percentages to how much each player provides towards those versus the rest of the players available at that position.
E.) Assign a standard-deviation to those players at each position.
F.) Sort the total player-pool by their value vs. their piers.

This way you'd at least limit the player pool you're looking at to about 100 to 150 players, the players that realistically need to be compared against all other players, rather than just against their own position. After you get beyond that it's all going to fall apart anyway.
MTUCache
Major League Manager

Posts: 1263
(Past Year: -21)
Joined: 16 Mar 2007
Home Cafe: Baseball

### Re: Quantifying performance

That's pretty much correct. Here is one of the most underrated threads of all time, IMO.

viewtopic.php?t=92352

The above describes a way to assign a dollar value to each player. Even if you're not in an auction format, it can still be used as a way to rank players. The beauty of all of it is that incorporates position scarcity. That thread will forever be bookmarked!
Element
General Manager

Posts: 4904
(Past Year: 28)
Joined: 25 Feb 2006
Home Cafe: Baseball

### Re: Quantifying performance

When you draft a player higher than his numbers would warrant if there were only one position the stakes are raised. If you draft Utley and he meets your projection you're halfway there. But if Weeks and Kinsler exceed your projections and put up similar value to Utley you lose ('09). If Utley only plays 115 games and the rest of 2b put up numbers like they did last year you lose big time. Same deal if there is a random breakout like Kinsler in '08. You have to go back to '07 to find a year where taking Utley higher due to PS paid off, but the guy who drafted BJ Upton probably got more out of his pick. If you look back at the recent PS history at Catcher it's ever uglier.

In seasons where I've done all the adjustments for PS and drafted accordingly I've been worse off than not. My advice is to use a list of PS values as a secondary tool on draft day because if you aren't dead on with your nearly all of your projections and you draft accordingly you're setting your team up for failure. There is more room to succeed in taking the best available player while keeping track of how fast a given position goes on draft day. Those PS scarcity values change in real time during the draft according to the unpredictable nature of your fellow drafters. But don't take it too far.

One season I went to the length of using std dev. based PS values in real time, deleting a player from my spreadsheet and thus immediately recalculating PS relative values to look at for my next pick...that lead to some weird picks and to me concentrating on a more "traditional" approach to drafting in the future, while keeping in mind the fancier valuation methods.
GoodOl'Days
College Coach

Posts: 130
Joined: 14 Jun 2010

### Re: Quantifying performance

Yeah, in traditional roto, there's really no such thing as position scarcity. Replacement level is pretty much the same across the board. The only exception is if you start two catchers. In shallow leagues position scarcity comes more into play, too.
Skin Blues
General Manager

Posts: 3082
(Past Year: 82)
Joined: 11 Apr 2010
Home Cafe: Baseball