Hello all, my name is Igor Zolotusky and this is my first blog. One of my favorite hobbies is working with data and doing sports and other types of analysis. I would work on it for a week or two and then forget about it. So a friend of mine suggested I start a blog. So here goes…
A little bit about myself: I’ve been doing research in one form or another for work since 2008 in particular for the media industry. In 2014, I took a break from the real business world and went back to get my MBA in the academic business world. I got my MBA from Carnegie Mellon University’s Tepper School of Business. Tepper allowed me to grow my analytical skill set and get in touch with my inner nerd. In addition to being a numbers geek, I also really like sports, especially hockey and football. So using what I learned at work and at school, I started working with data for various sports. This first blog post is about fantasy football. It was originally a project for my Big Data class at Tepper but I’ve added a few bells and whistles. This first part is simple pivot tables, just ranking NFL quarterbacks by fantasy points in total and in various different conditions.
I start with a general ranking of QBs according to fantasy points average over the last 4 seasons. I felt like 4 seasons is a large enough sample size to be able to analyze data from but small enough to still be relevant. The listed QBs are those that are at or above league average (~18.37). I will filter out the low pass attempt games and below average QBs from here on out, unless stated otherwise. In this list, the highlighted ones are the top 10 (from now on I will highlight the top 10 unless otherwise specified). I also made one other adjustment and excluded any games where a quarterback had 10 or fewer attempts. Those types of games are usually due to injury or rest (at least for the top QBs listed here) and weighing a game where a QB had 10 or fewer attempts with a regular game would artificially pull down the average down. This is especially true of QBs like Dak Prescott who have played few games and had a game where he rested and thus had few fantasy points. One other note is that I included Peyton Manning here for reference. However since he retired prior to the 2016 season, I excluded him from the rest of my analysis since no one is picking him for his/her fantasy team.
The following is the fantasy football points averages for QBs split between Home and Away:
This is a basic split of the QBs at home vs. on the road. The top 10 are highlighted for home but the top 11 are highlighted on the road. The reason is that Matt Moore has only played 2 road games with over 10 passing attempts. However, maybe his coach would like to play him on the road some more with those numbers (I’m half joking). Either way, the average for these above average QBs at home is 21.9 and on the road is 20.2.
QBs that drop from home to away include such household names as Drew Brees, Aaron Rodgers, Ben Roethlisberger, and Tom Brady. QBs that improve from home to away are Russell Wilson, Marcus Mariota, Jameis Winston, and Jay Culter. Particularly surprising is Russell Wilson who has what many consider the best home-field advantage in the sport. It appears the top QBs as a whole drop in play away from their home field.
The following is a clear vs. inclement weather breakdown:
The above lists rank quarterbacks by fantasy points in clear and inclement weather. I classified fog, rain, sleet, and snow as inclement weather. Unlike the previous rankers, the sample size isn’t great for both lists. Meaning, that while at home and away QBs more or less play the same amount of games, there aren’t many games that are played in inclement weather. Therefore, I added games played for reference and will continue to do so in subsequent rankers where this sample size problem applies. Please note: I highlighted the top 9 QBs here because Andrew Luck played only 1 game in inclement weather and therefore wasn’t especially relevant. I kept Tony Romo and Marcus Mariota since it would be difficult to exclude the top 2 QBs on this list.
Some of the data confirms what we expect but some of it is surprising. Tom Brady improving in inclement weather, considering he plays in Foxborough, MA most of his games is not surprising. In fact, not only does he improve in such instances he has a large sample size of games (9) to draw from to confirm that this isn’t a fluke. Ben Roethlisberger, who plays in Pittsburgh has played in a large number of inclement weather games too (9) and also improves from clear to inclement weather. However, Aaron Rodgers the starting QB in Green Bay significantly dropping in inclement weather is surprising. Perhaps, Green Bay is more cold than wet, but that’s analysis for a different time. Of course, overall above average QBs are much better in clear conditions (20.8 vs. 18.7). One final note, for aesthetic purposes I cut off the two bottom QBs in the clear ranker so that the screenshots would be the same size. I will continue doing this unless otherwise stated. The QBs that were cut off were Matthew Stafford at 18.2 and Jay Cutler at 17.7.
Next, we take a look at how wind affects certain QBs:
Low wind is defined as anything 10 MPH and below, and high wind is defined as over 10 MPH. The average of above average fantasy QBs for low wind and high wind is remarkably similar (20.7 for low wind and 20.3 for high wind). Once again I have included the games played for context as the high wind games are much more uncommon than low wind games.
There are some interesting insights to gather from these rankers. For example, Tom Brady who has played in 12 high wind games, the most from eligible QBs in the 4 years of data, is mortal in those types of games. His fantasy points drop from 23 to 20. He doesn’t even make the top 10 of QBs in high wind games (though that’s once again partially due to a small sample size of other QBs). Considering that he plays in New England this is surprising. Many QBs do better in low wind games than Brady does in high wind games, so if you know the forecast is high winds in Foxborough but low winds for your backup QB, it may be wise to go with the latter. Ben Roethlisberger, another QB that plays in tough weather conditions in Pittsburgh, also drops in high wind games. It seems sometimes conditions beat even the greatest QBs. On the other hand, more in line with the expected, Aaron Rodgers’s fantasy points go up in high wind conditions and even though he’s only 3rd, the top 2 QBs played a total of 3 such games, 2 fewer than Rodgers. Finally, another is the curious case of Drew Brees. How poorly does Brees do in low wind games? He does so poorly he got cut off from the above screenshot. His average fantasy points in low wind games are 18.8. However, in high wind games albeit of not the highest sample size (3), he does very well. This improvement from low wind to high wind games is surprising for a dome QB. Analyzing the data further, however, does show that one of those games was against the 2014 Steelers who were 27th in pass D when he gained 40 fantasy points.
Now, we will take a look at QBs playing in different temperatures:
The above are rankers for how the QBs performed in extreme weather (freezing and very hot) and compared them to the standard warm weather that had the QBs played the most games in. I classified as freezing any game where the temperature was 31 degrees and under. I classified very hot as any temperature that was 80 and above (the exact upper bound was 92 degrees as that was the hottest relevant game recorded in the last 4 years). For warm, I somewhat arbitrarily chose 50-69 degrees.
There are some interesting insights that can be gathered from these rankers, however, it’s hard to find players that played a significant amount of games in all three temperature groups. One QB that played a decent sample of games in all three temperature groups is Tom Brady. No surprise that one of the greatest of all time does well in any temperature. Perhaps it’s a bit surprising, however, that the New England QB did so well in very hot temperatures and significantly better than in freezing temperatures. It’s only been 4 games but they all came against teams average or above average in passing D in the year he played them (3x Dolphins and 1x Cowboys). Not surprisingly a QB like Philip Rivers does very well in very hot conditions as he plays in San Diego. He was merely good (and outside of the top 10 in warm conditions). One surprising QB is Ben Roethlisberger and his mediocre stats in freezing temperature, especially compared to his warm temperature stats (15.8 and 11th vs. 23.7 and 4th). For context freezing fantasy points average is 18.3, very hot is 20.4, and warm is 21.2.
The final breakdown of the game conditions is dome stadium vs. open stadium:
As one would imagine playing in a dome correlates with increased fantasy points. The QBs’ fantasy point average in open stadium games is 20.6 and dome games is 22.3. Some of the data confirm what we already suspected. Drew Brees is, in fact, a dome king and a mediocre QB in an open stadium. An interesting case is a player like Ben Roethlisberger who drops significantly in domes to 15 fantasy points from 21.8 to 15.0. On the other hand, it’s only a 3 game sample. Finally, perhaps the most interesting player is Andrew Luck. Luck plays in a dome stadium but has the same fantasy points average in open stadiums. In fact, he’s 2nd in the league in open stadiums, but 4th from all the players that have played significant games in domes. The fact that domes are more quarterback friendly in general makes his stats all the more interesting.
For the final portion of this analysis, I will leave stadium conditions and take a look at the defense a QB faces. This first part is pass D in terms of pass yards allowed per game:
For these rankers, I didn’t include the count of games played since they were consistent and mostly significant. I highlighted the top 11 for two of the rankers (top 10 and 11-20) because while the other QBs played significant games Matt Moore played one or two and ended up in the top 10. As one might expect, as the defense gets worse the average fantasy points get better (top 10: 20.7, 11-20: 20.7, and 21-32: 22.8).
Some quarterbacks you saw the previously mentioned pattern for more so than others. For example, Ben Roethlisberger is great against average to poor defenses, but abysmal against good defenses. He’s 2nd to last in top 10 (the bottom one Marcus Mariota is not seen here as he got cut off, but he’s only 0.1 worse). He improves in both rank and average linearly as he goes to 11-20 and then 21-32. Tom Brady, someone you would think would be immune to these effects also follows a similar pattern. He’s much better than Big Ben against top 10 Ds in both average and rank, but his average improves consistently as defenses get worse (although his rank is lowest for 11-20 defenses). The most interesting case is Drew Brees. He seems to not be affected by the defense he faces. In fact, his rank is highest for top 10 defenses (#1 in the league) and his average actually drops for 11-20 defenses and is virtually identical for 21-32 defenses. Looks like Brees is one of the rare QBs where the matchup doesn’t matter as much, at least from a standpoint of the defenses he faces.
Next, we look at defense from a different angle. How did the QBs do against certain defenses according to passer ratings allowed?
The top 10 are highlighted for the first two groups, but the top 11 is highlighted for 21-32 group since Matt Moore so conspicuously is #1 and only played one game against a 21-32 passer rating defense. The average fantasy points are top 10 (18.7), 11-20 (20.7), and 21-32 (23.2). The difference between standard pass D and passer rating D is that standard pass D takes into account only pass yards allowed per game. Passer rating takes into account completion percentage, yards per attempt, touchdowns per attempt, and interceptions per attempt. Therefore a defense can allow a lot of passing yards (and rank poorly in pass yards allowed per game) but play bend but don’t break defense and allow few TDs and get many INTs and thus be high up in passer rating D.
One interesting case is Ben Roethlisberger. He is almost dead last with 15.3 average fantasy points against top 10 pass yards defense. However, he jumps up to 21.5 and third against teams with top 10 passer rating defense. Perhaps that can be explained by those top passer rating teams being poor in allowing pass yards but getting a lot of interceptions. In that case, if Big Ben is careful with the ball in those games he can take advantage of those teams’ proclivity in allowing a lot of yards. Drew Brees once again is first against top 10 defenses. However, this time he significantly improves against the bottom defenses. For some reason, once again Brees does worst against the middle of the road defenses. Aaron Rodgers is both amazingly consistent and consistently amazing. He is #2 in each group, improving as he gets to each successively worse defensive group.
That’s all for this simple ranker analysis of QBs in different environments and against different defenses in the last 4 seasons. In my next blog entry, I will take a look at a bit more advanced statistical analysis of what factors affect fantasy points.
One final note, I used the following websites as data sources:
Igor Zolotusky's Analysis 