Monday, February 16, 2009

Implementing a Category Optimization Plan

roll2 suggests that you can't dump steals without making sure that you buy at least one top middle infielder who doesn't run.

If you are going to dump steals, you better nominate and go get the valuable MI who don't steal much, because there are only a few of them, and once your league mates figure out what you are doing, they likely will make them pay a premium for them.
To test roll2's theory, I went back and examined the American League last year, since I have average salaries to look at, and could see whether or not I would have been able to buy middle infielders using Alex Patton's 2008 bid limits. A serendipitous result of this exercise is that I wound up creating a list of bid prices for 2008 as if I was dumping steals. While I recognize that this won't help owners establish prices for this year, it will provide insight into how you might go about tossing a category overboard in your own league.

1) I dialed the SB values down to zero. If you have the Patton software, you can do this rather easily. In the Patton formula for 4x4, steals are worth 22.2% of the pie on offense, so you're technically going to be left with about $1700 across a 12-team league to budget for offense. This comes out to about $142 per team, and isn't enough. Since you're dumping steals, you're going to want to spend a little less for your offense, but not too much less or you won't wind up with enough in the three categories you are competing in. You should budget about $158 per team - or $1896 across a 12-team league - on offense. There's a technical explanation as to why this is so, but explaining it is beyond the scope of this article. (If you want to be even more technical, you should not only dial the SB values to zero, you should skew HR/RBI up so that you actually come out to $158 per team but, again, these calculations are beyond the scope of this article). Put simply, you do want to spend more on pitching than the standard $85 if you're committed to sweeping the category, but not too much - you still need to win three offensive categories to make this work.

2) I calculated how Alex's bids were prorated versus his $ values. If you don't do this, you're making a big mistake, since bids tends to take risk into account and not pay full ticket price for someone like Elvis Andrus or Matt Wieters (assuming those two even make the bigs). I then applied the same prorated formula to the new dollar values from step #1 (the ones that didn't include steals) to create new rankings.

3) I plugged Alex's initial bids into the new rankings, and
4) Multiplied each bid by a factor of .9029 ($1896/$2100).
Both of these steps sound more complicated than they actually are. After step #2, players had a new $ value that didn't include steals and a new bid value that was directly proportional to their old $ value/bid value. By bumping the bids up back up to $2100, I was able to take a peek at how these players ranked under the "old" bid limits. The new bid limits were calculated by multiplying the "ranked" bid limits by .9029 - or the new league hitting budget divided by the standard 67% hitting budget of $175 per team.

Why add the extra steps (the math majors out there are probably asking)? Can't you just take the prorated bids after step #2 and make sure that you adjust these bids so that they equal $1896?

You can do this. However, I like adding steps #3 and #4 because I'm not competing against myself to buy stats; I'm competing against 11 other owners who probably aren't dumping steals and are budgeting across a standard $175/$85 split. I want my prices to be radical; however, I also don't want them to be too radical.

Getting back to Roll2's theory, he's correct. Bumping up against the league average salary of the three expert leagues I use to calculate average salaries (Sportsline, LABR, and Tout Wars), 59 hitters fell in at $1 or more than those average salaries. Only two of these hitters - Marco Scutaro and Juan Uribe - were middle infielders. That's not enough. I don't need to get three top middle infielders, but I probably want one.

You need to do what you always have to do with your bids in competitive leagues: throw out the formulas and manually tweak them. Robinson Cano and Michael Young's adjusted bid prices using my formula above tied the average market price. Knowing this going into a 2008 auction, I might have bumped up Cano/Young $1-2 in order to buy them - or someone like them.

Keep in mind, though, that you don't have to spend big bucks on a middle infielder just because there aren't that many power hitting middle infielders out there. Goign with a Scutaro/Uribe/Jamey Carroll middle infield last year might have long as you made sure to get the power bats you needed elsewhere on your roster. It might have also freed up money for you to chase Miguel Cabrera, which you probably wouldn't have done otherwise.

There is no one right way to optimize a category. I agree with Roll2. I'd feel uncomfortable going in with three middle infield scrubs in a category dump, and would probably want my prices to skew so that I could at least bid aggressively on the likes of Cano and Young. However, if my competitors pushed them to ridiculous heights, I'd probably zig while they zagged and spend more money on my corners and outfielders...where there are non-SB players aplenty.

The nice thing about dumping steals is that you're not stuck chasing a finite amount of players, like you are when implementing a more radical strategy like a no-power plan. Category dumping is no different than playing it straight. It's nice to have a system, and even nicer to have good prices on your sheet. But an important component to any successful auction is adjusting on the fly when your opponents think their catching on to what you're doing.

Don't let them.

No comments: