Position scarcity

[Jackie Treehorn]

OK here's my question. How do you take projections for a point-based league and normalize the projected points for pitchers and hitters to get an effective overall ranking put together? Right now I have an overall pitcher ranking and an overall hitter ranking. Albert Pujols is the top hitter with 649 points, whereas Randy Johnson is the top pitcher with 1029. That's a pretty big difference in points. I obviously can't just combine the rankings based on points, otherwise I'd have 35 pitchers ranked ahead of Albert Pujols. So what would be a good way to normalize these point values in order to compare hitters and pitchers reasonably?

I think you're looking for standard deviation. In other words, to find out in that type of scoring system which players are most valuable, you'd have to compare them to the average. So, I'd imagine you could take the number of teams in your league and multiply that by the number of hitters they would have on their team. (For example, 12 teams X 12 hitters each=144 hitters. Then take the top 144 hitting point getters and find the average of ALL of their points combined. Then compare each hitter to the average to find how much their score deviates from the average. (For example, the average of the top 144 hitters was 400 points and Pujols scored 649). His score deviates 249 from the average. You could then do the same for the pitchers and compare their scores to the average for all pitchers. (Example, the top 144 pitchers average 600 points and Unit scores 1029). His score deviates 429 from the average. Then you would rank all players together (hitters and pitchers), but do so in order of highest deviation from the average. So..Unit would be ranked ahead of Pujols with a score of 429 to Pujols' 249. This would allow you to get a combined pitchers/hitters rankings list. You could take it a step further, I'd imagine, and take each position (1B, 2B, etc.) and find an average for it and compare each player at that position to that average to get a deviation.

[aperri]

Hi Guys,

A couple of Cafe Members who are also our subscribers invited me to comment on this topic. As many of you already know, I support this site with ad revenue and enjoy to cruise by occasionally, as many of you drive the intensity that makes this sport so fun.

We've been using the Position Scarcity model since 1999, and I'd like to begin by saying that almost every rankings list you see today use some form of position scarcity to come up with it's overall ranking. They might not quantify it, but I rarely see overall lists that are based on straight valuation.

I don't know if the folks who have said they have done back testing on the topic really used the right parameters, because we've done over 100,000 simulations and the results are not even close when comparing the results to straight drafting using $ values (to the tune of 25% more effective within the first 80-100 picks…after that the value of position scarcity reduces greatly and become par with straight valuation) If the ranking or dollar valuation service is not computing $ values based off of the projected statistics this is another matter entirely.

I can offer you results of these tests in a statistical format, but that really will not mean much to the non mathematician. So I will present it in a non scientific format, and leave you to judge if it passes the test of logic:

As an example of our drafting strategy, last year we recommended 3B Scott Rolen as an overall top 10 selection based on VAM. If however, we based our rankings strictly on his 5x5 Roto valuation he would have fell outside of the overall top 20.

Do you believe that Rolen’s draft rank should be elevated based on the mediocrity at his position…when you consider the Hillenbrands and Ensbergs that muddle up the position…does it make sense that Rolen's value is more than his stats contribute (intrinsic value). Let's assume that you draft in a serpentine type of league: Based on last year's stats, Todd Helton finished with a similar fantasy worth...but who would you rather have? If you selected Helton to fill in you 1st base slot, more than likely you would have to fill in your 3rd base slot with a mediocre player by the time your turn came up again. Whereas if you selected Rolen, you'd more than likely have a solid player to choose from at 1st base since there is a much greater depth at that position.

If you draft in auction league, then you might ask "what is the difference if Rolen plays 1st or 3rd, as long as he's on your team?" Since most fantasy leagues require that a roster include a standard number of players at a pre determined number of positions, then there is a replacement cost in filling a starting roster position with a middle of the road 3B like Morgan Ensberg over the likes of a middle of the road 1B like Frank Thomas. As an example, if you let 3B Rolen slip by last year and instead invested an equal number of draft dollars on someone with similar stats (let's say 1B Todd Helton again). This has two immediate and obvious effects: 1. It significantly reduces your spending capital and 2. It puts you in a position were you'll have to fill in some other positions with average players. If the other position that you fill in with an average player is a position that is not deep in quality, then it's going to hurt your team's accumulated production. A counter argument might be "if you did choose 3B Rolen over 1B Helton, then to fill in your 1st base slot isn't 1B Frank Thomas going to cost you more than 3B Morgan Ensberg....thus making the apparent gain a mute point". Not necessarily, since there is a greater degree of draft deflation when it comes to the mid tier player, thus the difference in draft cost/position would not equate with the gain in productive stats. Auction Drafts typically follow a barbell structure (most of the money is spent upfront on star players) which creates a deflation of prices for the mid tier player. Thus you’ll be able to pick up a middle of the road player 1st baseman like Frank Thomas at relative discount based on the strength of the other players at his qualified position.

Additionally you must apply a predictability of each position into the scarcity model. As almost everyone knows, pitchers and catchers are very unstable positions to forecast, mainly due to the excessive injury rates among the groups. In fact each position should have a predictability quotient that is attached to the position scarcity model (or you would be drafting a stud catcher in the #1 slot). And this is exactly what we do and needs to be done to maximize the value of the position scarcity model.

For more on our concept of Position Scarcity, visit: http://www.insiderbaseball.com/VAM.htm

Best of luck this fantasy season!

Anthony A. Perri
Publisher and Statistician
Fantistics Fantasy Sports

[baseballgeek7]

Anthony, thanks for the putting it in that light. The position scarcity theory makes considerable sense to me.