Now, on to today's topic. This is not exactly a topic for the lazy or those in need of instant gratification. You are going to have to read and think about this stuff.
The topic: The Gap Is Widening?
Recent recruiting news has the Skunkbears practically wetting their pants. An over-hyped high school kid has foolishly elected to name that school down the road as his choice. To further exacerbate the matter, another over-hyped kid is considering the same. It is with great glee that the Skunkbears are thumping their collective chests, as if they actually had ANYTHING at all to do with it, and spewing stench over and over again on the many Spartan sports message boards.
The Skunkbears claim that "Hoke is eating Dantonio's lunch" in recruiting.
The Skunkbears claim that "the gap is widening" (and the 'natural order' will be restored).
The facts support just the opposite. The gap, at least as far as recruiting is concerned, has been closing.
How is one to tell how things are going? Most are quick to look at the most recent class or the next incoming class and jump wildly to conclusions. While those class rankings might have some bearing, possibly, one must pause and ponder. Since when did a freshman class make or break a team? Certainly not on a regular basis or as a general rule.
The point is, if one is going to assess the relative strength of program recruiting, based exclusively on recruiting site rankings, one has to include the recruiting classes that make up the team, year by year.
We must digress to make a point.
Statistics is a powerful tool for those who know how to use it and comprehend its inherent limitations. Most football "statistics" are little more than averages of aggregate data, the simplest form of statistics and usually abused.
For example: The running back averaged 5.0 yards per attempt. Take the total net yards and divide by the number of carries. What does that tell you about his performance? Does it tell you the probability that any given play will result in a long rush? No. Done right, statistics can tell you his "mean value" rush, which defines the distance where half of his rushes are shorter and half are longer. It can also tell you the "standard deviation," which is the delta from the mean (+ & -) that contains 68.2% (1 sigma value) of his results. What good is that? It tells you how consistently the running back plays. Two players can have a mean value of 5.0 yards per rush. One of them can have a standard deviation of 1 yard and the other of 2 yards. The first player can be relied on to gain between 4 & 6 yards a play rather reliably. The other is between 3 & 7 yards. Maybe mentioning the "bell curve" will jog a few memories. The point is that statistics can do things more meaningful than just giving out data averages.
What does that have to do with recruiting? Go back to he Skunkbear claim of "closing the gap." That expression literally means they see a trend.
Statistics is very useful in describing trends. Trends are results versus time. This returns us to to the earlier point of needing to assess 5-year multiple class rankings. Doing so creates a rolling (or running) average. Rolling averages are good for assessing trends in situations where the data fluctuations need to be "smoothed out" without degrading the results.
A typical team consists of 85 scholarship players plus another 20-plus walk-on athletes. NCAA, as a rule, allows a school to offer 25 scholarships in a given year. Couple this with the redshirt rule, which allows a coach to sign the kid and delay the use of eligibility for a year, and in today's game, there are recruits from 5 classes on most teams - true freshmen to redshirted seniors.
One could make the point that there are actually only a few freshmen and redshirts that start and so 3 years should be used. The counter point is, when doing rolling averages, a fewer number of time intervals used allows more data fluctuation to shows up in the results. The bottom line is the trend lines become less descriptive and, thus, less useful. In this, a balance needs to be achieved.
Scout, Rivals, ESPN, and 24/7 Sports are the Big Four in ranking recruiting classes. MaxPreps assesses high school players. Each has their own scouts/analysts. Each has their own evaluation criteria. Each has their own ranking algorithm. We will save the discussion of those ranking systems for another time.
A cautionary note is required. There are 11 years of recruiting data available for each team. Anything less that 100 samples is not statistically significant. The sample size is too small to ensure with less than 5% probability that the results did not occur by chance.
The Gap Is Widening?
The data consists of Scout recruiting class rankings for the year given averaged with the 4 preceding classes. 2014 is based on today's rankings and will change as the 2014 recruiting continues. Scout recruiting data goes back to the class of 2002.
Using 5 year rolling averages of the class rankings:
Years UM MSU delta
02-06 8.6 46.2 37.6
03-07 6.8 48.6 41.8
04-08 6.4 40.6 34.2
05-09 8.2 45.4 37.2
06-10 10.2 43.8 33.6
07-11 14.2 40.6 26.4
08-12 13.0 37.8 24.8
09-13 12.2 35.6 23.4
10-14 10.6 31.8 21.1
The most significant single factor that influences class rankings is the number of recruits in the class.
Years UM MSU delta
02-06 102 115 +13
03-07 102 117 +15
04-08 110 121 +11
05-09 110 114 + 4
06-10 114 111 - 3
07-11 113 108 - 5
08-12 118 108 -10
09-13 120 104 -16
10-14 113 93 -20
The delta between the rolling averages is reducing, while the 5 year total number of recruits shows a remarkable shift. MSU is doing a better job of retaining the kids they recruit.
Bottom line: MSU improved their class rankings while reducing the total number of recruits while UM managed to sustain their rankings but with a significant increase in the number of recruits, which has a big impact on class rankings.
It is obvious the Gap Is CLOSING. Moreover, the probability is the gap has closed more than the numbers indicate.
Continued in Part II.
Tweet ------------- Remember to check Spartan Headline links, updated real time, at the bottom of the page. -------------- Please click on COMMENTS below the post to enter and view reader comments. ***If you are receiving this post via automated email, ***you need to go to the site to view headline links and embedded videos in this post: http://spartanresource.blogspot.com/
This is an awesome Statistical Proof that supports what Coach Dantonio has been saying in different ways during the last year or so, while at the same time refuting the specious claims made by skunkbears about their purported trendline during that same time period.ReplyDelete
Most people recognize me as a "stat freak", but some of your work was tickling the top of my cranium. Love it. Go Delta!
Here's a few more simple stats for the more simple-minded of the skunkbear family:
1. The last two games are a 1-1 split.
2. Hoke has a 1-1 record against MSU.
3. Dantonio has a 4-2 record against UM.
4. Dantonio's MSU teams have outscored UM by 45 points.
5. Hoke's UM teams have been outscored by MSU by 12 points.
6. In the last six games of the series, the two closest final scores were the two wins by UM.
In six games against Michigan, Dantonio's teams have won by about a touchdown (7.4 points per game). In two games against MSU, Hoke's teams have lost by about a touchdown (6.0 points per game).
We recognize the abuse of the word "statistics" when referring to sports/athletics. Your use of the word "stats" seems to be just for the benefit of those you are addressing as the items you listed (as you are well aware) are merely data and data averages.Delete
What would be interesting is to do a distribution of the MoVs and find the mean and standard deviation. Yes, keeping in mind that 6 is not a statistically significant number of data.
Good point, as you are actually "The Stats Man", while I am merely a Stat Freak. Did I mention that I love Data, too? (People who know me refer to me as either Spock or Data.)Delete
-Data leads to statistics.
-Statistics lead to information.
-Information leads to knowledge.
-Knowledge leads to wisdom.
The road to wisdom begins with data!