How to find draft value at Defense in 2017

Defenses are hard to predict, but we put a lot of work into doing just that. According to both MathBox™ and the Fantasy Outliers team, these Defenses could return higher value than ADP and ECR this year.

Chris Seal
Aug 25, 2017 · 7 min read

Listen to our draft prep monster, all-you-need-to-know podcast at Fantasy Outliers Podcast Ep. 5 — The Big 2017 Predictions Episode!

It refers to our draft kit, which is available free here. (desktop friendly)

Defenses are hard to predict

Figure 1. Average Weekly Value versus Draft Order by Position historical returns in 10-Team Standard leagues since 2011. The Average Weekly Value, Percent Within Position, Scatterplot in the link provided.

This holds true for our models, too. When we conducted our expected error analysis of each of our models, we found that we were wrong by about +/- 1.3 points per week on average when predicting DST performance. Yikes. That’s not good, but hey, neither is the status quo (and the ‘status quo’ probably doesn’t even know how wrong it is). A lot went into these models, and I think they provide a unique insight into this difficult-to-predict fantasy position.

So, what are some good Defenses to target in 2017?

Figure 2. MathBox™ Defense/Special Teams rankings vs. Expert Consensus Rankings

The above chart compares MathBox™’s projections with Expert Consensus Rankings (ECR). MathBox™ predicts Points Per Game, then calculates an estimated Value vs. Tier 1 & 2 / Top 20 Defenses. For example, Carolina’s +1.29 means that we think they will beat the typical Top 20 Defense by 1.29 points per week on average. After a quick look at our draft kit, they’re projected to beat Tier 1 / Top 10 Defenses by +0.95 points/week.

Key takeaways for finding value at Defense/Special Teams

Dave Seal: Everyone says that defenses are hard to predict. It’s not that they’re not valuable, if you find a good one — just that the risk is so high that conventional wisdom says not to reach for one. Instead, ‘they’ say wait until the end of the draft right before you choose a kicker. By and large, our predictions are actually validating this strategy. MathBox™ is telling us that it’s less sure about these than the projections for some of the other position groups, in part, because the expected error rates are higher. Also, there is only about a 2-point range from the highest projection to the 20th-rated defense. So it’s not really taking as big of a stand as it is on some of the other positions.

That said, the biggest stand it’s making is the Carolina Panthers where MathBox™ has at #1 and ECR is #11. That’s a pretty big difference. The depth behind that rank is meaningful, too— a +1.3 points/week value versus +0.9 for the next team down, the Denver Broncos.

After Carolina, Denver, Kansas City, and Houston all fall in line with ECR (which is kind of crazy if you think about it).

Some slight bargains this year could be the Seattle Seahawks, as MathBox™ has them at #5 and their ECR is #9. Another one is the Cincinnati Bengals come in right after them at MathBox™ #6 and ECR #12.

Personally, I’m a big fan of streaming defenses. In our research, we found that there is almost a 1-point difference between home and away points scored for Defenses — the most of any position. There are usually good defenses available on the waiver wire (the Minnesota Vikings last year, come to mind).

Figure 3. Home/Away average weekly points differences for each position. It’s the Home/Away Difference graph in the link provided.

David Harbarger: So far, we’ve talked about teams that are all projected to be starters. We might have the order different, but the masses will be drafting all of these teams. Here are a few potential sleeper teams:

The Jacksonville Jaguars come in at #7 for MathBox™ and #18 ECR. My intuition would be closer to what ECR is saying, so I’m not sure what’s going on here. Another one is the Buffalo Bills, who MathBox™ has at #10 vs. an ECR of #22.

Teams MathBox™ isn’t a fan of this year are the Arizona Cardinals (MathBox™ #11 vs ECR #7) and the New England Patriots (MathBox™ #17 vs. ECR #4). Is this is an error in the prediction or if there are some things that MathBox™ is seeing that the masses aren’t? It does give me pause. But to me, while I’m not sure the Patriots are as bad as #17, I’m still going to be avoiding them at #4.

In our analysis of historical leagues, there’s always a defense that gives back +4–5 points/week. In 2015, the Denver defense was the 6th most valuable ‘player’ in any position in competitive leagues. So points are out there to be had, if you get the right one(s).

That said, if you assume 12 defenses will be drafted, that leaves 20 defenses on the waiver wire. It seems to me that one of those is more likely to pop than one of the ones that was drafted. So I’m going to draft one defense and be looking on the waiver wire in the first few weeks in an attempt to try to be the first to snag a promising D/ST.

Clay J. Seal: I can’t draft defenses/special teams. I’m terrible at it to the point that MathBox™ told me to draft New Orleans first and I would do it. I think the issue with D/STs is that a good defense in the real game doesn’t seem to correlate as much to fantasy and it’s very difficult to analyze when the eye test doesn’t gleam any information. However, even with a little bit higher of an error rate for this position, our predictions seem to like the usual suspects a lot (Denver, Kansas City, Houston, Seattle) as well as Carolina, so trying to snag one of those is probably your best bet.

Chris Seal: As I described in the section above, a lot of inputs and a lot of years went into these models, so if any of the projections give you pause, it’s not like it’s pulling random numbers out of thin air. They are based in something meaningful. Now, some of the projections will be wrong, but they’re rooted in some pretty robust data and methodologies. I like to think that MathBox™ is more likely to be correct than what else is out there.

As you might’ve seen in our Wide Receiver’s Draft Value analysis (or any of the position breakdowns on our Medium homepage), I like messy charts. Basically, it’s comparing Expert Consensus Rankings with MathBox™’s rankings in order to find players who can give you a good draft value. Anybody at or above the dotted gray diagonal line is projected to be a good draft value. Bigger green circles are projected to have positive Weekly Value vs. Tiers 1 & 2 and smaller red circles are projected to have the negative Weekly Value.

Figure 4. Which Defenses to target in what part of the draft.

While I don’t think you can go wrong with Denver, KC, or Houston at the top of ECR, I also don’t see how you can miss Carolina at ECR’s 11th-rated Defense. If we’re looking at 10-team leagues, they should be available at pick 140. If you don’t like them for whatever reason, I’d go with Den, KC, Hou, Sea, or Cin. Maybe I will actually draft two defenses this year. It depends how my drafts go, but it’s a valuable position, if you get it right. I do like some of the waiver wire ideas the guys described above.


To get a unique draft kit that will give you an advantage in your league this year, go to our 2017 fantasy projections here. (desktop friendly)

If not, you can also connect with us by: Listening to our podcast, reading our articles, following us on Facebook, Twitter, Instagram, or emailing us at: podcast@fantasyoutliers.com.

Fantasy Outliers

Where machines and humans team up to win at fantasy sports

)

Chris Seal

Written by

Co-Founder, Lead Data Scientist at Fantasy Outliers

Fantasy Outliers

Where machines and humans team up to win at fantasy sports

Welcome to a place where words matter. On Medium, smart voices and original ideas take center stage - with no ads in sight. Watch
Follow all the topics you care about, and we’ll deliver the best stories for you to your homepage and inbox. Explore
Get unlimited access to the best stories on Medium — and support writers while you’re at it. Just $5/month. Upgrade