Statistics can come packaged several different ways. Some should come packaged with a warning label. It is easy to formulate a statistical statement to help bolster a weak argument or hide behind a shaming fact. Product success rates are a simple and good example of this.
“[Insert workout program name here] is one of the most successful workout programs on the market. 90% of our users have dropped several pounds in the first few weeks while only working out minutes each day!”
Groundbreaking!
If what was implied here were true, I would buy it. “New Year – New Me” is off to a very slow start. Unfortunately for my physical appearance, I should not take that statement at its face value. I should first dig into it to see what it really says. How do they define “users”? Are “users” all of the people who bought the program? What if “users” were all of the people who followed the program exactly and the program consisted of a 90-minute workout and carb-free diet plan. In that case, the “statistical statement” doesn`t endorse the workout program as much as it endorses the idea that working out for an hour and half every day and dieting will increase my odds of losing weight. Not so groundbreaking.
That example is a marketing ploy that is done intentionally and created to mislead. Sports statistics, although not done intentionally, can sometimes also be misleading. I think most novice players will allow a stat to sway their opinion. I am by no means discounting the usefulness of data. My approach is very data driven. I use data even when trying to develop or test new game theories; but learning how to approach certain pieces of data is a useful skill to develop. (How many times in one article do you think I can use the word data? I would put the over/under at 15. New writer trick – use the same word over and over again.)
Fantasy Labs gives us all the tools needed to take a real look at data. I highly recommend any newer users of Fantasy Labs take the time to watch some of the video tutorials on how to utilize each tool. I saw an interesting tweet from Mike Clay after hearing it mentioned on Adam Levitan’s podcast. The tweet mentioned that over the past six weeks the top four defenses in the NFL in relation to fantasy points allowed to QBs are the Houston Texans, Seattle Seahawks, Green Bay Packers, and Cincinnati Bengals.
All four of those teams are playing this weekend, none against one another. I think any statistic is relevant in a short slate, where I am looking to find any edge possible. I decided I would take a little deeper look into the past six weeks for these defenses. I was already considering Ben Roethlisberger and Kirk Cousins going into the week, and they are both playing one of these four teams. I would not let a single statistic seen on twitter scare me off of a play, but I do think it is worth taking a deeper look.
This is the performances of quarterbacks when playing the Packers in the past six weeks.
That is not the strongest group of quarterbacks. In relation to each player’s Plus/Minus on the season, only three of these players are comparable to Kirk Cousins, who draws the matchup against the Packers this weekend.
Those players are Carson Palmer (5.43), Derek Carr (5.4), and Matthew Stafford (5.26). This season, Kirk Cousins has a Plus/Minus of +7.74. Don’t believe me? Take a look.
Matthew Stafford had a horrible showing against the Packers. Looking into this, I see that a high variance QB had a poor game in the second meeting between divisional rivals. I didn’t play Stafford that week and can’t remember how I thought he would do, but I hope I was not very surprised by this performance. Derek Carr had a good game while playing from behind for almost the entirety of the game, which is a great situation for your quarterback to be in.
Carson Palmer barely met his salary-based expectation, something he has only failed to reach in two games this year. In this game, the Cardinals were up 17 at half and up 31 half way through the third quarter. The defense scored two late touchdowns, allowing the passing game to really take their foot off the gas. I think his performance can be pointed more towards game flow instead of the Packers defense.
I wouldn’t have let the Packers recent performance sway my opinions of Kirk Cousins without at least taking a deeper look. After doing so, I do not see any reason to sway me off of him in a game with a one-point spread that is tied for the highest total on the slate. I did not find anything too surprising when peeling apart Green Bay’s most recent performances, but it is something I do often with varying results. Results aside, I think it is a way of viewing data that should be explored by more novice players who take data at face value. (Data count is now at 11.)
I am going to get into another example of this. Before that, I will put the same data below for the other three defenses that were previously mentioned.
Another example where I think taking the time to look further into some data is when using the Trends tool. I mentioned in a previous article about how I used this tool to take a look at how QBs on the west coast perform when playing at home against an Eastern Time zone team. The Trends tool allows us to quickly take a look at this. I had this trend already made, but I remade it for the sake of this article and it took me less than a minute.
I think this is most common way someone would go about creating a trend for this situation. In fact, it is the exact way I first went about creating it. However, there is a problem with this. Fantasy Labs allows you to take a closer look at the individual past results. And when looking into this specific trends’ past results I notice three things right away.
I underlined the three players who are creating noise in the “Past Results” tab. I need to eliminate these results. I should be clear that I am not eliminating them because they scored poorly and are driving down the success of this trend. I am eliminating them because they are not part of the player pool I am interested in. My interest lies in how the starting quarterback is going to perform.
All three of these quarterbacks show up in this trend because they did play during these games. Kellen Clemens and Tarvaris Jackson both entered at the end of the game — they both took a knee to finish with a box score of 1 rush for -1 yards. Colin Kaepernick came into the game after a hit on Blaine Gabbert that was flagged for roughing the passer. Kaepernick went on to hand the ball off once and throw an incomplete pass before Gabbert re-entered the game.
This could very easily go overlooked. Now that it has been noticed, I have a few options to clean this trend up. The method I chose was to enter all the true data into a spread sheet and get the information myself. Another method would be to be a bit more thoughtful when decided what to factor into building the trend.
The only difference here is that I added that the player’s floor is between 1 and 20. I know that aside from a very unusual circumstance Fantasy Labs is not going to project a quarterback’s floor above 0 if they are not going to be starting. A quick look through the past results yields no obvious noise or outliers. A quick simple click of the mouse eliminated six past results and gave me a much better look at how this trend has recently performed.
My “process” is a fluid one. I change where a lot of my time will be spent based on a few different factors. For example, early this season it became evident that pricing on DraftKings was soft at the tight end and quarterback positions. I spent a lot of time trying to best identify the values at those positions because I thought that accurately finding those values would offer me a nice edge on the field. I had good success rostering two tight ends for the first half of the season. I used the Trends tool a ton in both of those instances. I could go on forever about how powerful a tool this is and I think trying to squeeze every ounce of usefulness out of it requires getting to know it a little bit. Again, I strongly recommend checking out some of the tutorial videos or simply asking for help, if needed. Data is our friend and if you are on this site, you have access to tons of it – we just need to find ways to make it as helpful as possible.
Good news if you took the under — including this sentence the final “data” count comes in at 14.
