I apologize, my intent was to post this at the end of last week, but as I was crunching the data, the project took on a mind of its own and I kept thinking of better ways to do it, which led to more questions to answer and before I knew it, the weekend was upon me and I was in the dark for a day, literally, as Irene rendered me powerless all day Sunday. The irony is that even after spending extra time on the project, I am still in the dark. But, I decided to present what I have done and see where it takes us.
One of the more popular notions that has evolved in fantasy football circles the past several seasons is that a consistent team is preferred over one that one week kicks tail and the next scuffles. Some analysts have developed metrics to gauge the consistency. I admit, I should have done some more due diligence before embarking on this study, reading up on exactly what others have done, but alas, I did not. I just read and hear consistency alluded to in a lot of fantasy advice so I thought I would take a look at the principle.
Intuitively, I can make a strong argument that consistency is better. Using the guise that you want to win, the assumption is your team scores better than the league average. I mean, who cares if it better to score the league average number of points by being consistent or having some great weeks mixed in with poor ones. Regardless of the schedule, unless you are extremely lucky, a squad averaging the league number of points is going to be a .500 team, give or take a little. Anyway, as just alluded to, the schedule and luck of the draw plays a major role in fantasy football. Without looking at the data like we will do soon, my gut tells me there is something to the consistency concept. By consistently scoring above the league average, this forces your opponent EVERY WEEK  to score even higher than the league average. Sure, the luck of the draw is still a factor, but your score is fixed and only teams that do really well will defeat you. If your seasonal average is ten percent above the league average but you have some really strong weeks and some really week ones, you are opening yourself up to losing to a team that scores at or below league average that week as well as scoring a ton of wasted points during a matchup you would have one by scoring considerably fewer. Since I am someone that prefers data over my gut, no matter how large it may be, I thought this would make a neat little study.
I chose the weekly score to be investigated to be ten percent above the league average. My data showed that a team averaging ten percent above the league average would finish 95 or 104 in a 14game fantasy season. This seemed quite reasonable. Enter my first conundrum. Should I do the study with the team scoring ten percent each week’s league average, or use a fixed number equal to ten percent more than the average score of all fourteen weeks? I opted to do both, though my lean is to favor the former, assuming ten percent above each individual week’s average.
I used three private leagues as my model. They are all twelve team leagues but use different scoring systems so the league averages are all different. The sample of three leagues is way too small to make any definitive conclusion, but my thinking was the impact of consistency may differ depending on the average points scored in each league, perhaps based on the scoring system. Usually, the lower the points scored, the fewer points are assigned to yardage bonus meaning touchdowns are paramount. The point being, the consistency of players scoring touchdowns and getting yardage points could be different. Are you sort of getting where I was coming from in the introduction? The more I got into it, the more I realized the question was not so black and white; there is some serious gray area.
The study was comparing the appropriate points scored total to what actually occurred, and determining the corresponding winning percentage. You then compare all the different winning percentages to examine the impact of consistency.
What I will do is slowly walk you through one league, presenting the data along the way, then present the data for the other two leagues. I will begin with assuming that the team averages ten percent above the league total each week.
In the first league I studied, the average for all teams was 84.3 points. A team scoring exactly the league average each week had a .504 winning percentage. A team scoring ten percent above each weekly average was .649 (about a 95 record). This is the most consistent team possible and their yearly average was 91.7 points. What I then did was access varying degrees of consistency. I used four percent increments, adding and subtracting that percentage from the team’s “consistent” score, which was itself ten percent above the weekly average. I determined how a team scoring that percent above and below would fare in terms of winning percentage. I looked at +/ 4%, 8%, 12%, 16% and 20%. In each case, the team’s yearly average is the same; the assumption is in half the weeks the team did better, the other half it did worse. Here are the associated winning percentages:
0% 
0.649 
4% 
0.646 
8% 
0.646 
12% 
0.637 
16% 
0.643 
20% 
0.631 
This data indeed suggests there is something to consistency. Remember, in each of those instances, the total number of points the team scored is exactly the same.
These results led me to a question. Would it be better to field a consistent team that averaged fewer points than this team? To look at this, I subtracted 2%, 4%, 6% and 8% from the team’s weekly totals. Keep in mind that s subtracting 10% results in a team scoring the league average every week. Here is a table showing how a team consistently scoring that number of would fare. The average is included to give a perspective to the numbers, remember that this study uses +/ each week’s average. That is, the number provided is the yearly average.
0% 
83.4 
0.506 
2% 
86.7 
0.530 
4% 
85.1 
0.565 
6% 
88.4 
0.607 
8% 
90.1 
0.631 
10% 
91.7 
0.649 
At least for this league, a consistent team scoring fewer points was not better than a more inconsistent one averaging more points. Hang on, you’ll see why this is relevant in a bit, don’t forget, we have two more leagues to look at.
Here is the data for this league assuming that the team scores the same number of points each week, exactly ten percent above the seasonal team average:
0% 
0.667 
4% 
0.670 
8% 
0.646 
12% 
0.634 
16% 
0.649 
20% 
0.631 
Here, the slightly inconsistent team actually fared a smidge better, but obviously not enough to be significant. And now, here is a look at consistent teams scoring a little less than ten percent above the seasonal average each week:
0% 
83.4 
0.524 
2% 
86.7 
0.554 
4% 
85.1 
0.601 
6% 
88.4 
0.619 
8% 
90.1 
0.637 
10% 
91.7 
0.667 
Here, a consistent team scoring a couple fewer points is about the same as a more inconsistent team scoring a few more points.
I will now present the data for the second league. This one averaged 126.9 points per team, so the study was done on a team averaging ten percent more, or 139.6.
0% 
0.708 
4% 
0.682 
8% 
0.667 
12% 
0.664 
16% 
0.649 
20% 
0.619 
Chalk another one up for consistency. Here are the results of the team scoring incrementally fewer points:
0% 
126.9 
0.494 
2% 
132.0 
0.536 
4% 
129.4 
0.565 
6% 
134.5 
0.607 
8% 
137.1 
0.667 
10% 
139.6 
0.708 
A slightly worse (in terms of average points) but more consistent team can hang with a slightly inconsistent team but has a better record than a more inconsistent squad scoring more points.
And here is the data, side by side, for a team averaging the same number of points every week:
0% 
0.667 
0% 
126.9 
0.524 

4% 
0.667 
2% 
132.0 
0.565 

8% 
0.682 
4% 
129.4 
0.607 

12% 
0.673 
6% 
134.5 
0.613 

16% 
0.649 
8% 
137.1 
0.643 

20% 
0.628 
10% 
139.6 
0.667 
Now the third league, with the team scoring ten percent (146.5) above the league average (133.3), side by side:
0% 
0.690 
0% 
133.2 
0.536 

4% 
0.679 
2% 
138.5 
0.565 

8% 
0.682 
4% 
135.8 
0.583 

12% 
0.661 
6% 
141.1 
0.631 

16% 
0.631 
8% 
143.8 
0.679 

20% 
0.604 
10% 
146.5 
0.690 
The trends are similar. The first set of data shows a consistent team scoring the same number of points as an inconsistent team is desired. The chart on the right suggests that a team scoring even fewer points, on a relative basis, than the first two leagues is still better than an inconsistent teal scoring more points. There is actually a trend forming here, but the sample is too small to draw any concrete conclusions, but at least with respect to these three leagues, the more points the league scored, the more you want to be consistent since you can average even fewer points and be better than a consistent team scoring more.
And finally, the data from the third league with the team averaging the same total every week:
0% 
0.708 
0% 
133.2 
0.494 

4% 
0.682 
2% 
138.5 
0.536 

8% 
0.667 
4% 
135.8 
0.565 

12% 
0.664 
6% 
141.1 
0.607 

16% 
0.649 
8% 
143.8 
0.667 

20% 
0.619 
10% 
146.5 
0.708 
Are you still with me? If so, thanks. So we have presented a ton of numbers, probably in a confusing manner only to corroborate what was initially assumed in the first place. But like I said, it is always nice when data supports intuition. That said, I do think the study needs to be expanded. Even in a small sample, I see enough to believe in the consistency factor. My problem is, I am not quite sure how to incorporate this into drafting. Do you forgo upside for a player with higher floor? Do you build a consistent foundation then take your chances? The same thing goes when setting your lineup. Do you opt for the player that is more likely to score a moderate number of points or the one that can go off any week, but can also be invisible? All this bandwidth and we end up with more questions than answers.
Thus ends my foray into the realm of fantasy football. Thanks for indulging me and good luck with your drafts!!