Week 9 - Retrospective

Here is the fourth iteration of the retrospective. Some noteworthy points:

- RB and WR remain the most predictable (tiers are most accurate)
- QB also, but there are more high outlier performances than RB/WR
- TE was near perfect last week, for the first time. Past weeks have been noisy.
- K remains noisy, four weeks running

If anyone has spotted anything noteworthy this week or in prior weeks, I'd love to hear it (via twitter or comment below)

Tier 1 median: 25.7
Tier 2 median: 24.7
Tier 3 median: 16.3
Tier 4 median: 25.4
Tier 5 median: 15.6

Rodgers exited with an injury; but injuries can't be predicted. Tiers 1-3 turned out nicely, but there were some real all star numbers in tiers 4 and 5 this week: Foles, Big Ben, Keenum, etc.

Tier 1 median: 17.6
Tier 2 median: 5 (12 w/out Foster)
Tier 3 median: 9.9
Tier 5 median: 7.4
Tier 6 median: 4.2
Tier 7 median: 4.3

Aside from Foster, RB tiers also looked pretty accurate this week.  His injury reporting was rather curious this week.  RB remains one of the more predictable positions.

Tier 1 median: 13.1
Tier 2 median: 8.2
Tier 3 median: 3.4
Tier 4 median: 5.5
Tier 5 median: 2.2
Tier 6 median: 3.6

Usually the top 3 tiers are the elite WR tier, but last week with Denver plus five other teams out, we only had two elite tiers.  Tier 1 and 2 clearly beat out the others by a fair margin again.

Among the lower tiers.  We consistently get a few high outlier performances, but the median of these lower tiers remains low, which is expected.

Both of these patterns have been consistent with the WR retrospective each week.

Tier 1 median: 14.1
Tier 3 median: 5.3
Tier 4 median: 2.6
Tier 5 median: 3.3
Tier 6 median: 1.3

For the first time, TE tiers were near monotonic (almost perfect tier predictions).  In the past few weeks, TE have been noisy, which I had attributed to the fact that a large percentage of TE points are generate by touchdowns (which are inherently more noisy than point by yardage).

While TE performances were more in line with expectations this week, more players in tier 1 this week also helped.

Tier 1 median: 8
Tier 2 median: 6
Tier 3 median: 7
Tier 4 median: 6

Kickers remain noisy.  The amount of predictive power in predicting kickers using this model remains questionable.

Tier 1 median: 6
Tier 2 median: 4
Tier 3 median: 5
Tier 4 median: 4
Tier 5 median: 3.5

DST tiers were somewhat behaved.  Tiers 2, 3 and 4 were a bit of a mix, but tiers 1 and 5 appear to be on target.