This week, I looked at the idea of streaming players against favorable matchups in an effort to outperform their season-long average. Using real-world data from the most competitive leagues I’ve been a part of, I concluded that the theory fell apart in practice.
From this, one could conclude that streaming quarterbacks and defenses is not a great primary plan. And I think that’s a reasonable conclusion. But this doesn’t necessarily mean that investing a high draft pick into quarterbacks and defenses is a great primary plan, either.
So let’s look at that. Instead of telling you what doesn’t work, let’s examine the four FESL leagues to see if we can figure out what did work at quarterback and defense.
Does Draft Position Predict Positional Scoring?
For starters, I wanted to test the hypothesis that spending an early draft pick at the quarterback or defense position was the way to go for top production.
To quantify this, I looked at where each team drafted its first quarterback. I then assigned a team a score of 150 minus that draft pick. As an example, if a team drafted a quarterback at 23rd overall, they are said to have invested 127 points of value into their top starter.
Did teams that drafted their top starter early outscore teams that drafted their top starter late? Actually, they did not; the correlation between draft capital and positional scoring was -0.079, which doesn’t just indicate a weak relationship, but a weakly negative one; teams that waited longer at quarterback wound up scoring more at quarterback.
To some extent, that’s just a function of the season. It was a weird year at quarterback. The top three scorers at the position in FESL’s scoring system were Tom Brady, Carson Palmer, and Cam Newton; Palmer and Newton were late round picks, while Brady was expected to serve a 4-game suspension at the time of the draft and fell as a result.
At the same time, injury and underperformance struck the consensus top three quarterbacks in Andrew Luck, Aaron Rodgers, and Ben Roethlisberger. This meant that the top of the order was a minefield. But as I said, 2015 was an odd year; historically, quarterback has been one of the most predictable positions.
It will be interesting to see next year whether the lack of predictability at the position was a one-year fluke, or a harbinger of things to come.
At defense, I repeated a similar procedure, but this time I assigned each pick a value of 200 minus the overall draft pick, (to better reflect that defenses were going later). This time, the correlation between when an owner took his first defense and how many points he scored at the position was 0.208— a weak relationship, but at least a positive one.
Unlike at quarterback, the weak relationship between draft position and season-ending production at defense is hardly an aberration. Other than placekicker, no position is less predictable on a season-by-season and week-by-week basis. This is why defenses are typically drafted so late every year; we have little idea of which ones will actually be any good.
Do Number of Starts Predict Positional Scoring?
This, to me, was by far the more interesting of the two questions. I tallied how often each team started its most-often-started quarterback and defense. 4 of the 42 teams, for instance, started the same quarterback 12 times in the first 13 weeks— every week except for the bye, in other words. At the other end of the spectrum, there were six teams that did not start any individual quarterback more than 6 times.
Number of starts by a team’s primary starter is a good proxy for how “stream-heavy” that team was. Teams that gave a lot of starts to their top option were essentially anti-streamers— they identified a top option and rode him all year. Teams that gave fewer starts to their top option, on the other hand, were spending much more time and effort seeking out favorable matchups or juggling multiple players in and out of lineups.
How did the number of starts given to the top option correlate to production at the position? Across all 42 teams, the correlation was 0.361 at quarterback and 0.418 at defense. Starting a single quality performer, it turns out, was a much more productive strategy than juggling starters based on matchup.
How much more productive? At quarterback, each additional start given to the top starter corresponded to about 9.77 extra points over the full season. At defense, ignoring the two teams that started a defense 13 times, (opting to take a zero on the bye rather than dropping a skill player for a second defense), each additional start corresponded to about 8.74 more points over the full season.
On defense, of the 40 teams that did not start a defense on bye, 19 started a single defense 10-12 times. Those 19 teams averaged 175.6 points from the defense position. Nine more teams started a single defense either 8 or 9 times; those nine teams scored, on average, 159.1 points at the position.
That left twelve teams starting their top defense 7 or fewer games on the season, the most matchup-heavy cohort. Those twelve teams averaged just 118.1 points at the position for the year.
By far the most active defense streamer of the 42 teams was Justin Howe, who started 11 different defenses over 13 weeks, with no defense starting more than twice. Using the methodology outlined in my earlier article about streaming, Howe’s matchup-seeking cost him 24.83 fantasy points against expectations over the season. And the expectations were already low to begin with; Howe was cycling through 2nd- and 3rd-tier defenses. In total, his 94 points on defense ranked 40th out of the 42 teams.
Here are charts of average fantasy points scored against number of starts given to the top player at both quarterback and defense. I've omitted the two outlier teams who started a defense thirteen times and the one outlier team with a maximum of two defensive starts.
|Maximum Starts||Total Quarterback Scoring||Total Defense Scoring|
So What Does This Mean
The big takeaway is that streaming should always be a strategy of last resort. For every one fewer start teams in FESL gave their top quarterback or defense, they scored about 9-10 fewer points on the season.
The difference between teams that started their top defense 10+ times and the teams that started their top defense seven or fewer times was a hair over 4.4 points per game. The difference between teams that started their top quarterback 10+ times and teams that started their top quarterback seven or fewer times was a hair under 3 points per game.
This isn’t to suggest that the key is to just keep starting the same quarterback, ignoring everything else. The causal arrow runs the other way- the teams with a lot of starts accumulate them because they have strong quarterback production; they don’t have strong quarterback production because they allocate a lot of starts.
If you have a bad quarterback, continuing to roll him out will not magically make him good. In that case, it makes sense to start throwing darts at the board and hoping one will strike the bullseye. Playing matchups and churning through waivers are desperation moves, but that doesn’t mean they’re the wrong move when you’re truly desperate.
To avoid getting stuck with a bad quarterback in the first place, the solution isn’t necessarily sinking a lot of draft capital into the position, either. The Carson Palmer and Cam Newton owners might have started the season out intending to stream, but they quickly shifted gears and became a one-quarterback franchise as soon as they realized what they had.
At the end of the day, the goal should be to acquire a top performer at quarterback or defense and ride him or it. Streaming is not an equally good second option, it is a last resort should you fail at plan A.
There are a lot of ways to get a star. You can draft one early. You can draft one late. You can acquire one off of waivers. You can trade for one. However you do it, though, fantasy football is still all about the difference-makers. Unfortunately, there are no easy fixes to make up for their lack.
More articles from Adam HarstadSee all
More articles on: DynastySee all
Cracking FanDuel: Week 7 - Knotts
The New Reality No.152: Quarterback Touchdown Regression Candidates - Parsons
Dynasty News, Week 7 - Tefertiller
More articles on: StatsSee all
IDP Matchups to Exploit and Avoid: Week 7 - Larkin
For The Win: Week 7 - Larkin
Normalized Strength of Schedule: Week 7 - Lee