Now you are choosing a subset of the data to create a convenient argument which defends Kill.
My point encompassed ALL games that Fleck and Kill had coached to try and draw the best comparison.
Most responders who refute my data have been trying to argue on subsets of the data which no doubt favor Kill over Fleck, while at the same time ignore that data that has Fleck over Kill.
If you want to argue Kill or Fleck, there is countless subsets of data you can choose from to make that point, but it's less comprehensive of what I posted.
Someone who wants to defend Fleck over Kill could also choose smaller sets of data to make the argument look better for Fleck than his advantage he currently shows.
(Adjusted for the higher scoring football across the NCAA today would be the first option which would look much worse for Kill).
My data does not say this is a good defense, but the data does show it's not worse than Kill as a whole which is opposite of what we've read this week.
Your argument is that for some of the worst games it is worse which for some of the games it is worse. But in it's entirety they are very similar with a current edge to Fleck.
Wow. Just wow. You really like to skew you some data, don't you. You're arguments to look at the data in total are strong. You're arguments not to consider that the data is made up of sub-sets is just plain wrong. This, my friends is where fake news comes from. What's he old saying? Figures Lie and Liars figure. And of course, one of my all time favorite - "There are three kinds of lies - Lies, Damn Lies and Statistics" - Mark Twain.
Why would it be important to consider Non-conf. games as a subset vs. big ten games?
- Because the quality or lake thereof in the non-conf. schedules could skew the data. Fleck's teams have not played D-1AA teams, Kill's did.
- At the time, the Gophers played 4 Non-Conf. games under kill, and Fleck's squad's have played only three each season
- Since Fleck plays one more Big Ten game than Kill, understanding the differences would be important
- In the big ten season, 6 of the opponents are common which provides a more stable base of data for comparison. While the team strengths/weaknesses fluctuate from season to season, the general status of the pecking order of teams has changed very little, aside from Nebraska.
To lump it all into one bucket and call it good is misleading at best, deceitful at worst. I'm going to give you the benefit of the doubt on this one in that you just don't know better, so this should help you. One other point, you did not specify, but my guess, based on your simple use of aggregate stats to prove your point, you used Kill's data in total for two years. In this look I'm comparing the first 20 games between both coaches - "apples to apples" or the best approximation of that goal possible, showing their progress through 20 games into their status as Head Coach. Note, the data will be shown Kill first, as his coaching tenure was first.
Let's look at the data, shall we?
Through 20 games,:
Average Points given up in aggregate:
Kill: 28.3
Fleck: 25.3
Fleck is giving up a full 3 points per game less than Kill through 20 games. Worse than your skewed numbers show.
Median Points given up in aggregate:
Kill: 28
Fleck: 30
Interesting. Perhaps there is more to these numbers?
The max given up in points per game during this time:
Kill: 58
Fleck: 53
The min given up in points per game during this time:
Kill: 7
Fleck: 3
So in total, the numbers show that Fleck's defenses, overall, look positive when compared to Kill's
When comparing the totals for Non-Conf. vs. Big Ten, some similarities also pop out between the two (mind you not discounting the numbers or removing them, just looking at them in like context)
Average points given up:
Kill, 21.8 Non-conf, 32.6 Big Ten
Fleck, 8.5 Non-conf, 32.4 Big Ten
Breaking it out thusly shows a clear advantage for Fleck in the Non-con. schedule and roughly equal in the Big Ten. Useful to your argument. Until, that is you dig a little deeper.
So, let's compare year 1 vs. year 2 - for kicks and giggles, because in their lies the trends.
Average Points per game:
Kill: 2011, 31.7; 2012, 23.1
Fleck: 2017, 22.8; 2018 28.9
Kill, 11 Non-Conf, 26.8; 12 non-conf, 16.8
Fleck, 17 non-conf, 8.0; 18 non-conf 9.0
Kill 11 Big Ten, 34.1; 12 Big Ten 29.5
Fleck 17 Big Ten, 27.4; 18 Big Ten 40.5
The data shows that year one under Kill, our defense was much worse than Fleck's year one. However, Kill's defense improved significantly in year 2 as measured by average points allowed by 26.97% while Fleck's teams performance was worse by 26.46% in year two. This helps to explain the relatively flat comparison when taken in aggregate and why further diving in to numbers is always a good idea.
When looking at Non-conf numbers, Kill's teams improved significantly in year 2, improving by 37.38%. Fleck's teams were worse by 13%. It's important to note that while a significant percentage decrease in performance, Fleck's delta was only 1.0 points per game. When starting off as low as they did, the percentage can be misleading.
It's in the Big Ten where the data really shows a difference in trend. Kill's teams improved by 13.55%. Fleck's teams have regressed by 46.88% giving up 13 more points per game in year 2.
One more look here on points given up per game, is to break the data into quartiles for evaluation. This look could be done in different ways, but I chose quartiles vs. terciles or quintiles primarily given that this breaks up the data, at the end of the year into first and second halves of the season, which could prove to be interesting in understanding the data, the impact on the length of the season. Particularly for the narrative that Fleck's teams are so young, we might see a decline as the youth wears down against older, and as the argument goes, more physically mature players on the older teams we play. At this point, it breaks things out into four groups of five games, with a split in Q3 between the end of the First season and the beginning of the second. This does allow for some adjustment between the end of Season one and beginning of season two for Fleck given the last two games of the 17 season.
So, here's the data for Kill and Fleck. There are four data point per quartile and are showed in this order, avg points allowed, median points allowed, Max, Min
Kill:
Q1 33.0, 28, 58, 23
Q2 36.0, 41, 45, 21
Q3 18.4, 23, 28, 7
Q4 25.6, 28, 38, 10
Fleck
Q1 17.2, 14, 31, 3
Q2 23.6, 21, 33, 17
Q3 19.4, 14, 39, 3
Q4 40.8, 42, 53, 30
This is an interesting look, but given the way the data is split right now, this isn't as clear as it will be at the seasons end. There are trends in the data that are evident, however, that should become more interesting again, once the season has played out. Comparing Q1 and Q3 vs. each other and Q2 and Q4 vs. each other is perhaps the best way to look at this data and Q1 and Q3 include all of the Non-conf games with 1 Big Ten game for Kill and 2 for Fleck each, and Q2 and Q4 being only conf. games. There are some obvious trends visible in this data, although again, this becomes more interesting after the end of this season.
I have hypothesis about why we see what we see here, but I'll keep those to myself and let you draw your own conclusions on 37Scores claims that Flecks defenses are equal to Kills in their first two years. I will say the data does not support that assertion looking at it in several ways other than simply aggregate average points scored.
One other aspect I looked at is the YDs/Game given up. When complete, I added in the Offensive yards delivered. The was to consider the impact of an effective or ineffective offense on defensive statistics. Once completed I added the disparity between yards given up on defense vs. the yards gained on offense as a measure of effectiveness in the game plan.
Average Yards Allowed
Aggregate of first 20 games
Kill 364.4
Fleck 366.5
Median Yards allowed
Kill 373
Fleck 373
interesting!
Most yards allowed
Kill: 580
Fleck: 659
Least yards allowed
Kill: 160
Fleck: 199
Avg Year 1/Year 2
Kill 392.2/322.6
Fleck 346.7/396.1
Non-Conf/Big Ten
Kill 313.8/398.1
Fleck 247.7/417.4
Non Con Year 1/Year 2
Kill 350.5/277.0
Fleck 239/256.3
Big Ten Year 1/Year 2
Kill 413/368.3
Fleck 382.6/480
Again, the data is clear here that the stats for Kill's defenses improve across the board. Fleck's get worse, across the board. To suggest that the defenses under Kill and Fleck are similar through 20 games is not supported by the data.
For fun, I also did a quick dive through the offense:
Average aggregate Points scored and yards gained per game
Kill 20.4/319.8
Fleck 22.6/338
Pretty similar numbers in aggregate. You'll notice I'm not claiming the offenses were the same.....
Digging deeper
Aggregate Median
Kill 19
Fleck 23
Aggregate Highest score
Kill 44
Fleck 48
Aggregate Lowest Score
Kill 0
Fleck 0
Kill once, Fleck twice
Avg points Year 1/Year 2
Kill 18.4/23.3
Fleck 19.3/27.4
avg points Non Conf/Big Ten
Kill 26.3/16.4
Fleck 32.3/18.4
Non-conf points Year 1/Year 2
Kill 22.8/29.8
Fleck 33.0/31.7
Qualifier - it is arguable that Fleck's Non-Conf year 2 was much more difficult than Non-Conf year 1 given injuries to MTS and the melt down going on under Gary Anderson in OSU. While Kill did open against #25 USC in year 1, that loss was close and low scoring for both teams and did not skew the numbers offensively or defensively.
Big Ten Year 1/Year 2
Kill 16.3/16.8
Fleck 14.8/24.8
Here is where we see real separation in offensive performance to date with suggestions in the upcoming foe's for Fleck this year give the opportunity to at least maintain or improve on the numbers to date.
Quartile Look on offense, again, Avg Points, Median Points, highest game total, lowest game total
Kill
Q1 18.2, 21, 29, 0
Q2 18.0, 17, 24, 13
Q3 28.4, 28, 44, 13
Q4 16.8, 13, 28, 13
Fleck
Q1 28.0, 24, 48, 17
Q2 18.4, 21, 27, 10
Q3 19.0, 21, 48, 0
Q4 24.8, 28, 38, 13
While Fleck's offense has room to grow, it is significantly better than Kills was at this point which may suggest that the defense doesn't need to be equal to Kill's in order to have more success on the field. How much of a drop-off can occur is not yet known.
In the end, we hired Fleck to do better than Jerry Kill. This data suggests on one hand, he's doing much better offensively than Kill did, so, to date, that's working as planned. On defense, however, the data shows a team that is getting significantly worse through 20 games, not better. And the decrease in performance in terms of yards and points is, in fact, huge when looking at the trends and the most recent results.
Main point, 37Score, your assertion that the defenses are similar is just plain uninformed and ignoring the available data to compare the results. Because you refused to look at anything other than the aggregate, you either knew the data showed you were flat wrong, but choose to try and hide behind shoddy statistical analysis OR, most likely, you just stopped when you found data that supported your argument, like most of the people in this country.
