Is the performance of a team defense less predictable than the performance of other fantasy football positions?

October 25, 2008 10:58 AM

For the purposes of fantasy football, is the performance of a defense/special teams harder to predict than the performance of, say, a wide receiver?

The contribution in fantasy points to fantasy team totals in my league is very small relative to how important an entire defense seems in real football. Moreover, our fantasy defense scoring feels uncorrelated to real world defensive team performance.

I think some resistance to modifying the league scoring rules to "correct" this may be founded on the notion that these discrepancies is inherently harder for defenses than for quarterbacks, running backs, etc. . . That is, year to year performance for D's is more variable than that for other positions. My intuition is actually quite the opposite but I have no data one way or the other.

So, please direct me to an analysis of this question. My Access-fu is weak so this option is strongly preferred.

Or, help me do it myself. I can find tables of an individual player's yearly performance. I can find tables of stats for many players for a single year. Is there any place I can get the underlying database used to make those tables? Assuming I can find the data, what analysis should I perform? I guess I would devise some hypothetical defense scoring systems and use it to compare D's to other positions measured by conventional systems. I would compare by measuring the correlation from one year to the next, right? Do I need to adjust anything to account for the differences in the sizes of the data sets? I could also compare multi-year averages to the most recent year, right? What other ways are there to measure this? If my database included average fantasy draft positions I could measure correlations to that, too? Any leads there?

Or, share some anecdotal evidence from your fantasy league. Ultimately, I'm just wondering if there's a reasonable scoring system wherein some rational owners would pick DSTs in the same rounds when other owners are picking starting RBs, QBs and WRs. Does anyone's league operate like that?

posted by stuart_s to Sports, Hobbies, & Recreation (4 answers total)

Because of the way that most standard leagues operate, statistically arbitrary defensive plays tend to be heavily weighted. Fumble recoveries, interceptions and and defensive touchdowns can be gamebreakers, or they can be irrelevant, but they don't usually have nearly as much to do with the defense as they do with the player committing the turnover.
While fantasy baseball considers wins, fantasy football usually doesn't. I've always felt that simple wins should probably be a metric. They should be weighted depending on the points allowed, but they should definitely be a consideration. Sacks aren't super constant, but compared to turnovers, they might as well be written in stone.

The leagues that have the most consistent scoring defenses are the ones in which special teams yardage is figured into the scoring. Not everyone does this, but because punt and kick returners are basically offensive players, their statistics don't change as much from year to year. It should be noted that returners don't generally have long peaks (there are exceptions that prove this rule.)

Add weighted wins to a defensive scoring system and you may get to where you need to go. If the fantasy point differential between a winning defense and a losing defense is large enough, you'll start to find that they'll win leagues.

Most of this information is anecdotal or synthesized from PFP, so for what it's worth, you could look there. I've also not yet been in a league that's done this, so it's theoretical.
posted by Arquimedez Pozo at 11:47 AM on October 25, 2008

Get on over to the Fantasy Football Cafe forums and discuss it there. There is a fellow called "comm1680" on those forums who uses an elaborate mathematical formula to rank defenses every week. He takes a lot of things into account, and his rankings are far superior to chance. He likes to talk about how he does it, too.
posted by ikkyu2 at 6:38 PM on October 25, 2008

I don't think there's any way to answer this question without knowing how your league is structured. Is it a keeper league? If so, how many players do you keep? What's your scoring system, especially for the defenses?

Without knowing this information, there's no way to give an informed answer.

The league I've run for 20 years is a keeper league. There are 16 teams with 15 players each. Teams are allowed to keep 9 players before the draft. Most teams do keep a defense before the draft, but defenses, kickers, and tight ends basically have the same weight. That is, owners will keep QB, RB, or WR before any of the others.

Here's our league scoring system. For defenses, it works like this:

Fumble Recoveries (from Opponent)	2 points each
Interceptions Caught	 2 points each
Sacked a QB	1 point each
Safeties	2 points each	
Total Points Allowed	0-3	6	
Total Points Allowed	4-8	5	
Total Points Allowed	9-10	4	
Total Points Allowed	11-13	3
Total Points Allowed	14-16	2
Total Points Allowed	17-19	1
Number of Defensive & Special Teams TDs	6 points each

We also used to have points based on yardage, but that was overturned because most owners thought it was too arbitrary.

Finally, defensive teams don't add their points to our offensive totals -- they subtract from the opponent's offensive totals. For a long time, we were the only league I knew that did this, but I think it's becoming more common now.

During any given year, there do seem to be a few defenses that consistently score well. There are also defenses that don't. And from year-to-year, there's some (not a lot) of consistency. Chicago has been very good for the past few years, at least in our league. HOWEVER, there's no real long-term consistency, which is why defenses in our league don't get favorable treatment like QB/RB/WR. You know that Tom Brady is going to score you 8-12 points every week for his entire career. (Until he gets knocked out during the first game of this season — argh!). You don't know that about the top defenses.

Long answer, I know, but I used to be really into this stuff, and I've thought a lot about it. I've also run a lot of numbers. (Look around. It used to be easy to find data dumps you could import into Excel to run your own analysis on.) I'm not as into it anymore, but I still have opinions and impressions...
posted by jdroth at 8:36 PM on October 25, 2008

Defenses are harder to predict in my experience simply because, as mentioned, there's a lot of weight on arbitrary plays. You could have a defense with a couple of outstanding CBs that pick off a pass a game, but they may never return an oski for a TD in a season. Meanwhile, your opponent's defense just happens to catch a lucky break and returns a fumble for a TD, so now they've outscored you.

In the NFL, there are maybe five very good to excellent defenses, five poor defenses, and 20 that just kinda depend on the day of the week, weather conditions, whether the defensive coordinator took his smart pills this morning, etc.

WRs seem to be better choices for points... so long as they're #1/#2 targets. Even then, scoring TDs isn't a guarantee, and you're really thinking more about their consistency. In a 1 pt = 10 yrds league, a 1000 yd receiver should average over 6 pts/game before TDs. In reality, of course, they'll have 3-4 point weeks and 10 point (+ 100 yds receiving bonus if your league has that) weeks.
posted by dw at 2:48 PM on October 26, 2008