Process Analysis: How to Go About Your Fantasy Football Draft Part 2

I’ve been using process analysis to develop a draft strategy for Fantasy Football. I’ve already shown that getting a top running back should be your first priority. But what if the top running backs are gone by the time you get your first pick? When should you switch to other positions? I’ll use Minitab to find out.

Analyzing the Data

I’ll look at standard deviations, which measure the spread in the data. The greater the standard deviation is, the more the data is spread out. The position with the greatest spread is the one you want to draft early on. I collected fantasy points over the last 3 years for the top 40 running backs and receivers, and for the top 20 tight ends and quarterbacks (I selected only 20 because you draft fewer of them). I divided each position up into four groups and calculated the standard deviation for each group:

Standard deviations for each group

Let Minitab help you with your cheat sheet.

Your Top Priority: Top 10 Running Backs

We already established that getting a top running back should be your highest priority. But if the top running backs are gone, we need to look at the next highest group.

The Second Best: Top 5 Tight Ends

Yep, it’s important to grab a top tight because there is a big drop off from the top guy to the 5th guy. Obviously you don’t need to do this in the first few rounds. But if there are any stud tight ends, the numbers say to make sure you get one.

Coming in Third: Top 5 Quarterbacks

This is the position to go for in the first round if you think the top running backs have already been taken. In fact, you should make it a priority to get a quarterback even if your first pick was a running back. You’ll notice the standard deviations in the other positions decrease as we get to the more average players. But this is not true for quarterbacks. Even if you miss out on the top 5 guys, you want to grab the best available because their value keeps dropping at a large rate the entire way through the draft.

The Least Important Position: Wide Receivers

The biggest takeaway from this analysis is that you should ignore receivers early in your draft. Of all the positions, they have the least amount of drop off in fantasy points from top to average players. In fact, the numbers say it’s more important to grab the 11th best running back than it is to grab the top wide receiver! Sure, you’re not going to get Roddy White or Andre Johnson, but you’ll do just fine taking Mike Williams or Jeremy Maclin later on.

Keep Things in Perspective

Now this doesn’t mean you should always blindly choose quarterbacks and running backs over a receiver. Maybe you’re really high on BenJarvus Green-Ellis as your second running back. Because he’ll still be available later in the draft, it would be smarter to take a receiver first and pick up Green-Ellis in a later round.

And remember, last year Arian Foster was on average the 20th running back selected. Two years ago, Chris Johnson was on average the 8th running back selected. And three years ago, DeAngelo Williams was the 31st running back selected! You don’t have to pick a running back early to get a top running back!

So the statistics say to concentrate on elite running backs and quarterbacks early. Even though you will sacrifice getting an elite wide receiver, the amount of points you give up will not be nearly as much as the other way around.


Name: Jerry • Thursday, December 29, 2011

So the fantasy football season is over and I had a good year, but lost the championship. I have read a few of your blogs and am wondering if you have done any work with stats like targets and carries. It seems to me like, you might be able to make some predition based on this information.

Name: Kevin • Tuesday, January 10, 2012


Yes, I have definitely looked into more stats than just average fantasy points per game. I've looked at targets and receptions for receivers. And for running backs I've looked at carries, total touches, red zone touches, and fumbles. One of the problems is that sometimes they just don’t explain much of the variation in a player’s final average.

For example, this year Brandon Lloyd has the 4th most targets in the NFL, but is ranked 32nd in fantasy points per game. Dwayne Bowe is 6th in targets and 24th in fantasy points per game. Pierre Garcon is 8th in targets, and 27th in fantasy points per game. Just because a wide receiver is getting a lot of looks doesn’t mean they are scoring a lot of fantasy points. You can point out the same thing with running backs and carries.

Another problem is that in my regression model the predictor I use (average fantasy points through the first 5 weeks) already accounts for most of the variation that targets and carries can explain. If you’re getting a lot of carries and targets, odds are you’re scoring more fantasy points than those with less (both currently and going forward the rest of the year). But this variation is already accounted for in your fantasy average through the first 5 weeks. Adding targets and carries to the model doesn’t really add anything.

Hope that answers your question and thanks for reading!

Name: mike • Wednesday, May 9, 2012

Here is the problem.... Predictability of stats.
What is the standard deviation from one year to the next in terms of rushing yards. In otherwords, a runningback who gets 1000 yards this year, will he get between 800-1200 yards? 600-1400? The greater the standard deviation the lower probability of your player not even being starting quality. Players at position with high standard deviation can be drafted later, but you should also draft MORE players at that position because you are more likely to be able to get a starter out of at least one of them. Although a player with an expected score of 200 +/-20 points has an equal chance of getting 180 as he does 220 (assuming normal distribution) by grabbing a more predictable player early, you effectively are gaining later round picks when you don't have as much of a need for a backup. Additionaly, due to regression to the mean, many of the top performers won't be at their average, but will have came off a very high year of stats and are more likely to regress to the mean, which may skew the data left. As a result, players at positions with lower year to year standard deviation players should be taken earlier, players at position with higher std dev should be taken later. Additionally, you should get MORE players at a position with a high standard deviation because you will increase your probability of getting at least ONE player at that position that ends up being a star and more likely to get 2 or 3 starters..

blog comments powered by Disqus