RotoWire Partners

According to the Data: How to Stream Defenses

Keith Platte

Keith Platte writes about fantasy sports for RotoWire.


I recently posted an article about defense and kickers and received a comment (thanks Gr8daddyg) that gave me an idea. What if you rotated your defenses every week simply by match-up?

Meaning you seek out the weakest offenses and start whichever defense is playing that offense. If you are thinking about this, I will try to lay out what data to look at and the type of system one could use to assess the match-ups. Keep in mind, this approach might entail dropping and adding a defense on a weekly basis.


For this approach, I suggest looking at three measures of an offense to determine if they're a good match-up for a defense: points per game, sacks allowed and turnovers (TO). Below is a table from the 2013 season with the bottom half of the league in those three measures.

Pts/G Sacks Allowed TO's
1 Jacksonville (15.4) Miami (58) New York Giants (44) 1
2 Houston (17.2) Jacksonville (50) Washington (34) 2
3 Tampa Bay (18) Cleveland (49) Detroit (34) 3
4 New York Jets (18.1) Baltimore (48) Minnesota (32) 4
5 New York Giants (18.4) Buffalo (48) Arizona (31) 5
6 Cleveland (19.2) New York Jets (47) Houston (31) 6
7 Miami (19.8) Tampa Bay (47) Oakland (31) 7
8 Baltimore (20) Philadelphia (46) Cincinnati (30) 8
9 Oakland (20.1) Green Bay (45) Cleveland (29) 9
10 Washington (20.9) Atlanta (44) Baltimore (29) 10
11 Buffalo (21.2) Minnesota (44) New York Jets (29) 11
12 St. Louis (21.8) Oakland (44) Atlanta (28) 12
13 Atlanta (22.1) Seattle (44) Jacksonville (27) 13
14 Tennessee (22.6) Washington (43) Buffalo (27) 14
15 Carolina (22.9) Carolina (43) Denver (26) 15
16 Pittsburgh (23.7) Pittsburgh (43) Miami (26) 16


The next step is to create a rating system based on where each team falls in the table above for the different measures. One way you might do it is by assigning a "plus-2" to a team in the top five, a "plus-1" for a team from 6-10 and a zero for teams from 11-16 for two of the three measures (pts/g and sacks allowed).

For turnovers, the points applied are increased. So, "plus-3" for 1-5, "plus-2" for 6-10, and "plus-1" for 11-16. This increase is because turnovers are typically worth more in fantasy scoring than sacks and Pts/g. In the case where the teams share the same number of the measure but cross the point line (e.g., the turnovers between the 6-10 group and the 11-16 group), the rating is applied to those with the same number. Therefore, Minnesota and Oakland would receive a point even though they are in the 11-16 group. I will use Baltimore as an example for the total rating score: plus-1 for Pts/g (group 6-10), plus-2 for Sacks (group 1-5) and plus-2 for turnovers (group 6-10) totaling a plus-5.

Below is a table with all the team totals.

Team Pts
Baltimore 5
Cleveland 5
Houston 5
Jacksonville 5
New York Giants 5
New York Jets 5
Oakland 5
Miami 4
Minnesota 4
Washington 4
Buffalo 3
Detroit 3
Tampa Bay 3
Atlanta 2
Green Bay 1
Philadelphia 1
Seattle 1
Carolina 0
Pittsburgh 0
Tennessee 0


Based on this rotation approach, any defense that matches up with the teams in yellow or blue would be solid matchups. For week 1 of the NFL season, those teams are below:

Yellow Teams:

Cincinnati @ Baltimore, Pittsburgh vs. Cleveland), Philadelphia vs. Jacksonville, Houston vs. Washington, Detroit vs. New York Giants, Oakland @ New York Jets, New York Jets vs. Oakland

Blue Teams:

Washington vs. Houston, New England @ Miami, St. Louis vs. Minnesota

Before you go running off and employing this approach, there are four points of discussion needed: the point rating system, use of last year's season data, home/road splits and transaction costs.

First consider the point rating system I used. This can be changed, and I would appreciate feedback on what is the most appropriate assignment of points for these three measures. The second point is the use of last year´s data as a starting point. We all know each team had an offseason which entailed some level of change and hopefully upgrade; thus these measures must be re-examined.

I suggest waiting until week three to use the current year's data, then re-examine on a weekly basis for the rest of the season. If there is ample desire (via comments) for having a weekly report on favorable matchups based on this type of analysis, let me know.

Of course, teams typically perform better at home and worse on the road, so tailor your match-up defenses accordingly. That's not to say you should never stream a road defense, but only that it's probably worth something on the order of one point in the rating system.

The last point may not apply to everyone. In many of my leagues, I have transaction fees. While they are not hefty ($2 per transaction), over the season this approach could get pricey. However, what is the true cost for fantasy football championship?