P
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
P1
P2
P3
P4

The Weekly Gut Check - Crank Scores Projections

  Posted 7/16 by Matt Waldman, Exclusive for Footballguys.com

The Weekly Gut Check examines the players, strategies and guidelines fantasy football owners use to make personnel decisions.


Part IV of a series devoted to incorporating risk into your draft.

WARNING: If you are looking for the safest way to win a league, this month-long series is not the answer. However, the ideas, trends, and players mentioned here should help you with your individual approach in leagues where your peers also have a strong knowledge base. The key is how much you want to "break the rules." If "high-risk" in your daily life means taking a different way to work, and the word "reach" makes you cringe, then keep your GPS switched on, check out David Dodds' annual piece "The Perfect Draft." His well-conceived strategy should give you a good shot at a contender through by-the-book picks and depth. Consider these articles as a creative place where you see how "pushing the envelope" could work out for you and incorporate what you like/dislike accordingly.

What to Expect

This week, I'll provide a quick review of how Crank Scores work and lay a foundation for projecting consistency. This will involve more data than player discussion. Next week, I'll focus on what type of players achieve a high Crank Score and begin projecting them for the 2009 season with this method.

A Quick Review From Last Week

Crank Scores define consistency in a more advantageous way for fantasy owners. It measures how often a player meets or exceeds a desire baseline of performance. I set those baselines according to league type and they are separated into tiers:

  • Elite: The average fantasy point per game baseline of the very best at each position over a given range of time. For most leagues there are no more than three players at a position that qualify as Elite.
  • No. 1: The baseline value for fantasy No. 1 players in a given week.
  • No. 2: The baseline value considered fantasy No. 2 players in a given week.
  • No. 3: The baseline value for fantasy No. 3 players in a given week.
  • Subpar: The baseline value where players are not starter material if they fall below it.

Each baseline is determined by the player's position, the league size, and the starting lineup. For example, here is a table listing the lowest ranked players qualifying for each tier based on common league sizes and a starting line up of 1 QB, 2 RBs, 3 WRs, and 1 TE.

Size
Elite
No. 1
No. 2
No. 3
Subpar
QB
RB
WR
TE
QB
RB
WR
TE
QB
RB
WR
TE
QB
RB
WR
TE
QB
RB
WR
TE
8
1
2
3
1
8
8
8
8
*
16
16
*
*
*
24
*
9
17
25
9
10
1
2
3
1
10
10
10
10
*
20
20
*
*
*
30
*
11
21
31
11
12
1
2
3
1
12
12
12
12
*
24
24
*
*
*
36
*
13
25
37
13
14
1
2
3
1
14
14
14
14
*
28
28
*
*
*
42
*
15
29
43
15
16
1
2
3
1
16
16
16
16
*
32
32
*
*
*
48
*
17
33
49
17

Notice the asterisks for the QB and TE for the No. 2 and No. 3 tiers? That is because in a league with the starting lineup I mentioned above, there are no, No. 2 or No. 3 quality starters for QB or TE when each team only can field one starter at those positions. If your league allows 2 QBs or 2 TEs, then the QB and TE numbers would appear similar to the RB columns in this example.

I then measure how frequently a player reaches the average fantasy point per game baseline for that tier. If I'm interested how a player did for the 2008 season, then the baselines are determined by the fpts/g of the lowest ranked player according to the chart above. Let's use RBs in 2008 as an example. Here are the top 12 RBs for Crank Score from that season in a 12-team league with the 1 QB, 2 RB, 3 WR, and 1 TE lineup format:

Player
G
FPG
Subpar
Elite
RB #1
RB #2
Crank
DeAngelo Williams
16
17.73
31.25%
56.25%
62.50%
68.75%
25
Michael Turner
16
17.25
37.50%
43.75%
50.00%
62.50%
19
Adrian L. Peterson
16
15.53
37.50%
37.50%
56.25%
62.50%
19
Matt Forte
16
15.22
37.50%
37.50%
50.00%
62.50%
18
Thomas Jones
16
15.12
37.50%
25.00%
56.25%
62.50%
17
Clinton Portis
16
14.03
37.50%
25.00%
37.50%
62.50%
14
Brandon Jacobs
13
15.58
38.46%
38.46%
46.15%
61.54%
14
Chris Johnson
15
13.92
46.67%
40.00%
46.67%
53.33%
14
Steve Slaton
16
14.12
50.00%
31.25%
50.00%
50.00%
13
Marshawn Lynch
15
12.51
40.00%
13.33%
40.00%
60.00%
11
Maurice Jones-Drew
16
13.93
56.25%
31.25%
43.75%
43.75%
10
Brian Westbrook
14
15.56
57.14%
35.71%
42.86%
42.86%
9

Because this is based on a season, the tiers are based on the average fantasy points per game of the lowest scoring position for 16 weeks. Therefore, the baselines look like this:

Elite
FPG
No. 1
FPG
No. 2
FPG
Subpar
FPG
RB2
17.75
RB12
14.12
RB24
12.68
RB25
12.67

These baselines change according to the lineup type of your league, league size, and whether you're measuring Crank for a year, two years or two weeks because Crank uses the average value of that positional ranking for whatever time period you wish to measure. In this case, we see that the lowest ranked RB in the "Elite Tier" is the RB2. The players who scored enough fantasy points to be the second-most productive back for each week in the season had a combined average of 17.75 fantasy points per game.

Here's a simple mock visual of how this baseline works if you're not familiar with Average Value Theory (AVT):

Week
RB2
Points
1
Michael Turner
22.0
2
Adrian Peterson
29.0
3
DeAngelo Williams
15.7
4
LaDainian Tomlinson
14.8
5
Michael Turner
20.0
6
A. Peterson
19.0
7
DeAngelo Williams
18.0
8
DeAngelo Williams
15.0
9
DeAngelo Williams
17.7
10
Matt Forte
18.2
11
Clinton Portis
14.0
12
Marion Barber
16.9
13
Adrian Peterson
17.5
14
Michael Turner
17.0
15
Steve Slaton
15.0
16
Chris Johnson
16.0
17
DeAngelo Williams
16.0
Average Elite Baseline
17.75

The table shows how a different back is the RB2 depending on what he scores for the week and the one back with the second-highest average points per for the season. This 17.75 average baseline is the true number for 2008, but the players and fantasy points are just a simulation of how it's average. The average baseline is used to learn how often a back meets or surpasses this Elite Tier during the year.

When you look at DeAngelo Williams' Crank information, you see he met or surpassed that 17.75 FPG baseline for the Elite Tier a league-high 56.25 percent of the time in 2008. In addition, when you look at the No. 2 column, you'll see Williams was at least performing like one of the top 24 RBs nearly 70 percent of the season.

In contrast, a back like Steve Slaton was an Elite performer less than a third of the games in the season and a No. 2 RB for half the games. Last year, those results were still good enough to be the ninth-most consistent RB in 2008. However, the overall RB consistency was significantly lower than it was in recent seasons. I'll explain this in a bit more detail below, but this is why using two or three seasons of data to create an average value can be useful. I used three seasons, because 2006 and 2007 were far more similar to each other than 2008.

All these Elite, No. 1, No. 2, and Subpar tiers are nice to see when breaking down the players, but a little too cumbersome a quick reference. So I create a single stat that combines this information with a basic formula:

Consistency Ranking (Crank) = (Sum of Elite Gs + Sum of No. 1 Gs + Sum of No. 2 Gs) - Sum of Subpar Gs.

Here's how it works in action using DeAngelo William's 2008 season as an example. Williams played 16 games. Nine of those games were Elite (56.25%); 10 were No. 1 (60.25%); and 11 were No. 2 (68.75%). This means the other five games during Williams' season were not starter worth (Subpar). Therefore, Williams' Crank Score is (9+10+11) - 5 = 25

Crank Average Values: Creating X-Values For Projecting Performance

Most of you reading this probably have familiarity with Value Based Drafting and Average Value Theory. The quick and dirty about these concepts is that you can determine the projected value of a one player versus another across positions with what's called an X-Value. This X-Value is a positive or negative number calculated against a baseline player at each position. These baselines are determined in almost the same way I described in the first table of this article. For the sake of space, if you aren't familiar with VBD or AVT, then I suggest you learn more about them here and here before getting knee deep into my explanations of my process for projecting with Crank.

For projections, I opted to take the average Crank Score over the three-year period of 2006-2008 for each position. Then I created two other figures:

  • Percentage of Perfection (%): This is how close a particular Crank Score is to the highest possible Crank Score at the position for a league. In the 12-team league example with the lineup I've repeatedly used in my examples, the top possible Crank Scores are QB (32); RB (48); WR (64); and TE (32). For instance the average Crank Score from 2006-2008 for the top QB is 22 points. Therefore, 22/32 = 67.7%. In other words, QB1 from 2006-2008 was operating at 67.7% of the top possible score for his position.
  • X-Value (XV): I calculate X-Value just like you do for VBD or AVT. Continuing with the QB1 example in the previous bullet point, I determined the baseline QB for his league set up was the last ranked starting quality player. In a 12-team league starting one QB it is the 12th-ranked QB (QB12). Therefore I took the Crank Score of the QB12 (3) and subtracted that number from the Crank Score of the rest of the QBs to derive this X-Value.

The Percentage of Perfection is just a stat that I think is another interesting way view how much closer one player is to another to maximizing his Crank Score. The X-Value provides the real gauge of a player's value across all positions on a draft board.

When working with Average Value Theory, you create the X-Value based on an average over a determined number of seasons (I used three seasons) and then you plug in the players you believe will match the value. I like using AVT because performance trends haven't changed dramatically for the NFL. There will be years where there are significant outliers in performance, but there hasn't been a huge shift overall. This means if I use average values, I have a lower risk of distorting my entire draft board by over projecting stats. It's possible that four QBs could throw for at least 40 touchdowns in 2009, however if I go that far against the historical grain, I throw off potential accuracy of the rest of my draft board due to my unusual expectations for these four QBs and if I'm wrong I likely miss out on key values in the early rounds of my draft.

When you look at my Average Value Crank Score Cheatsheet listed below, you'll see that the players with the highest X-values are receivers and backs. In fact, five of the top six players from the standpoint of consistency are receivers. I also mentioned in a previous column that the wide receiver position has historically less turnover from year to year and individual receivers in the top tiers hold their value longer than top players at other positions. In many respects, it's a strong argument in favor of my high-risk discussion of investing early and heavy in wide receivers before targeting backs heavily in the mid-to-late rounds.

Generic 2009 AVT-Style, Crank-Based Cheatsheet

Pos
C
%
XV
Pos
C
%
XV
Pos
C
%
XV
QB1
22
67.70%
18
RB1
32
66.70%
27
WR1
44
69.30%
35
QB2
16
49.00%
12
RB2
29
61.10%
24
WR2
39
60.40%
30
QB3
13
40.60%
10
RB3
27
55.60%
22
WR3
37
57.80%
28
QB4
13
40.60%
10
RB4
23
48.60%
18
WR4
35
54.70%
26
QB5
11
33.30%
7
RB5
21
43.10%
16
WR5
33
52.10%
24
QB6
10
31.30%
7
RB6
19
39.60%
14
WR6
32
50.50%
23
QB7
8
26.00%
5
RB7
18
37.50%
13
WR7
30
46.40%
21
QB8
8
25.00%
5
RB8
18
36.80%
13
WR8
28
43.80%
19
QB9
7
22.90%
4
RB9
17
36.10%
12
WR9
26
40.60%
17
QB10
6
18.80%
3
RB10
15
31.30%
10
WR10
25
39.60%
16
QB11
4
11.50%
0
RB11
14
29.20%
9
WR11
25
38.50%
16
QB12
3
10.40%
0
RB12
13
27.80%
8
WR12
24
38.00%
15
QB13
2
7.30%
-1
RB13
13
27.10%
8
WR13
24
37.50%
15
QB14
2
5.20%
-2
RB14
13
26.40%
8
WR14
23
36.50%
14
QB15
2
5.20%
-2
RB15
12
25.00%
7
WR15
23
35.40%
14
QB16
1
3.10%
-2
RB16
11
23.60%
6
WR16
22
33.90%
13
QB17
1
3.10%
-2
RB17
11
22.20%
6
WR17
21
32.80%
12
QB18
1
2.10%
-3
RB18
9
19.40%
4
WR18
20
31.80%
11
QB19
1
2.10%
-3
RB19
9
18.70%
4
WR19
19
29.70%
10
QB20
0
1.00%
-3
RB20
8
16.00%
3
WR20
18
28.60%
9
QB21
0
1.00%
-3
RB21
7
13.90%
2
WR21
18
28.60%
9
QB22
0
1.00%
-3
RB22
7
13.90%
2
WR22
17
27.10%
8
QB23
0
-1.00%
-4
RB23
6
12.50%
1
WR23
17
26.00%
8
QB24
0
-1.00%
-4
RB24
5
10.40%
0
WR24
16
25.50%
7
Pos
C
%
XV
RB25
5
9.70%
0
WR25
16
25.00%
7
TE1
16
50.00%
16
RB26
5
9.70%
0
WR26
14
22.40%
5
TE2
10
31.20%
10
RB27
4
8.30%
-1
WR27
14
21.90%
5
TE3
9
29.20%
9
RB28
4
7.60%
-1
WR28
13
20.80%
4
TE4
8
25.00%
8
RB29
3
6.90%
-2
WR29
13
19.80%
4
TE5
8
24.00%
8
RB30
3
6.90%
-2
WR30
11
17.20%
2
TE6
6
19.80%
6
RB31
3
5.60%
-2
WR31
11
17.20%
2
TE7
4
12.50%
4
RB32
2
4.90%
-3
WR32
11
17.20%
2
TE8
3
10.40%
3
RB33
2
3.50%
-3
WR33
10
16.10%
1
TE9
2
6.30%
2
RB34
1
1.40%
-4
WR34
10
15.10%
1
TE10
1
3.10%
1
RB35
1
1.40%
-4
WR35
9
14.10%
0
TE11
0
1.00%
0
RB36
1
1.40%
-4
WR36
9
14.10%
0
TE12
0
0.00%
0
RB37
0
0.00%
-5
WR37
8
13.00%
-1
TE13
0
-1.00%
0
RB38
0
0.00%
-5
WR38
8
12.00%
-1
TE14
-1
-3.10%
-1
RB39
0
-0.70%
-5
WR39
8
12.00%
-1
TE15
-1
-3.10%
-1
RB40
0
-0.70%
-5
WR40
7
10.90%
-2
TE16
-1
-3.10%
-1
RB41
0
-0.70%
-5
WR41
7
10.40%
-2
TE17
-1
-4.20%
-1
RB42
-1
-1.40%
-6
WR42
7
10.40%
-2
TE18
-3
-8.30%
-3
RB43
-1
-1.40%
-6
WR43
6
9.90%
-3
TE19
-2
-7.30%
-2
RB44
-2
-3.50%
-7
WR44
6
8.90%
-3
TE20
-3
-9.40%
-3
RB45
-2
-3.50%
-7
WR45
6
8.90%
-3
TE21
-3
-9.40%
-3
RB46
-2
-3.50%
-7
WR46
5
8.30%
-4
TE22
-3
-10.40%
-3
RB47
-2
-3.50%
-7
WR47
5
7.30%
-4
TE23
-3
-8.30%
-3
RB48
-2
-4.20%
-7
WR48
4
6.30%
-5
TE24
-4
-12.50%
-4
-
-
-
-
-
-
-
-

If you were to formulate a draft order by round and pick based on this information here is how it would look. I used the color-coding to show which positions had the same X-Value:

Pick Order Based on Crank X-Value

Rnd
Pick
Rnd
Pick
Rnd
Pick
Rnd
Pick
Rnd
Pick
Rnd
Pick
1.01
WR1
2.01
QB1
3.01
RB8
4.01
WR20
5.01
WR25
6.01
WR28
1.02
WR2
2.02
WR9
3.02
WR16
4.02
WR21
5.02
QB5
6.02
WR29
1.03
WR3
2.03
RB5
3.03
RB9
4.03
TE3
5.03
QB6
6.03
QB9
1.04
RB1
2.04
WR10
3.04
WR17
4.04
RB12
5.04
RB16
6.04
TE7
1.05
WR4
2.05
WR11
3.05
QB2
4.05
RB13
5.05
RB17
6.05
RB20
1.06
RB2
2.06
TE1
3.06
WR18
4.06
RB14
5.06
TE6
6.06
QB10
1.07
WR5
2.07
WR12
3.07
RB10
4.07
WR22
5.07
WR26
6.07
TE8
1.08
WR6
2.08
WR13
3.08
WR19
4.08
WR23
5.08
WR27
6.08
RB21
1.09
RB3
2.09
RB6
3.09
QB3
4.09
TE4
5.09
QB7
6.09
RB22
1.10
WR7
2.10
WR14
3.10
QB4
4.10
TE5
5.10
QB8
6.10
WR30
1.11
WR8
2.11
WR15
3.11
TE2
4.11
RB15
5.11
RB18
6.11
WR31
1.12
RB4
2.12
RB7
3.12
RB11
4.12
WR24
5.12
RB19
6.12
WR32
Rnd
Pick
Rnd
Pick
Rnd
Pick
Rnd
Pick
Rnd
Pick
Rnd
Pick
7.01
TE9
8.01
TE11
9.01
TE17
10.01
TE19
11.01
TE21
12.01
RB38
7.02
RB23
8.02
TE12
9.02
RB29
10.02
RB32
11.02
TE22
12.02
RB39
7.03
WR33
8.03
TE13
9.03
RB30
10.03
RB33
11.03
TE23
12.03
RB40
7.04
WR34
8.04
RB27
9.04
RB31
10.04
WR43
11.04
RB34
12.04
RB41
7.05
TE10
8.05
RB28
9.05
WR40
10.05
WR44
11.05
RB35
12.05
WR48
7.06
RB24
8.06
WR37
9.06
WR41
10.06
WR45
11.06
RB36
12.06
RB42
7.07
RB25
8.07
WR38
9.07
WR42
10.07
QB18
11.07
WR46
12.07
RB43
7.08
RB26
8.08
WR39
9.08
QB14
10.08
QB19
11.08
WR47
12.08
RB44
7.09
WR35
8.09
QB13
9.09
QB15
10.09
QB20
11.09
QB23
12.09
RB45
7.10
WR36
8.10
TE14
9.10
QB16
10.10
QB21
11.10
QB24
12.10
RB46
7.11
QB11
8.11
TE15
9.11
QB17
10.11
QB22
11.11
TE24
12.11
RB47
7.12
QB12
8.12
TE16
9.12
TE18
10.12
TE20
11.12
RB37
12.12
RB48

Since most of your competition will be taking a far more traditional approach to valuing players, it's unlikely you will need to adhere to this order if you're going to try Crank Scores as your primary draft strategy. I have never seen anyone draft receivers this early in most traditional formats, however I think it could be an interesting experiment if you're feeling like blowing the minds of your competition without worrying about the outcome - for instance in a mock draft with an early pick. I'd consider executing this strategy in a mock of one league and then really use it in a different league with a similar set up if I liked the results.

Next week, I'll address one of the most important questions pertaining to Crank Scores: What makes a player consistent? I'll look at each offensive position and provide factors to consider when projecting consistency for the upcoming season. Next week's column will provide more examples with players and not strictly stats.