NFL Punters: Is there a difference

Watching the Broncos this season, one of the most oft heard complaints is about Britton Colquitt, the punter.  His average through the first fifteen weeks of the season was at 43.58 yards per punt, and Broncos fans were finding him lacking.  However, it bugged me that almost every qualifying punter in the league was averaging somewhere in the 40 + yards per punt region.  How much of difference can you truly glean from looking at the averages?

Using some data from the NFL Savant website, where they have gathered play by play data for the NFL this season, I pulled out every punt through week 15 and how far it traveled.  From there I computed the descriptives for each punter.  hen, performing an ANOVA test, I was able to conclude that there was in fact a difference amongst punter means.

ANOVA Results
Yardage
	Sum of Squares	df	Mean Square	F	Sig.
Between Groups	6578.996	31	212.226	2.749	.000
Within Groups	164029.928	2125	77.191
Total	170608.924	2156

 I also then tested for a difference in variances, ie whether or not the difference in means could be accounted for by the variance in their punts.  This was not the case though, as the Levene test gave us the inclination that punter to punter variance is the same

Test of Homogeneity of Variances
Yardage
Levene Statistic	df1	df2	Sig.
1.282	31	2125	.138

Followign this gives us that we needed to do some Post Hoc work to determine exactly where each punter falls in the hierarchy of the league.

Tier	Punter	Average
1		48.39
	J.HEKKER	48.39
2		47.93
	M.BOSHER	47.93
3		45.65
	K.HUBER	46
	P.O'DONNELL	43.95
	B.ANGER	45.9
	P.MCAFEE	47.81
	D.COLQUITT	45.71
	B.COLQUITT	43.58
	M.DARR	47.48
	R.QUIGLEY	43.66
	S.KOCH	47.84
	B.PINION	43.96
	M.SCIFRES	44.7
	T.WAY	45.59
	B.KERN	47.17
	C.JONES	45.75
	C.SCHMIDT	45.59
	R.ALLEN	45.65
	S.MARTIN	46.07
	T.MORSTEAD	44.94
	M.KING	44.27
	A.LEE	46.66
	B.NORTMAN	45.83
	D.JONES	47.41
	T.MASTHAY	43.68
	B.WING	44.84
	J.RYAN	45.82
	S.LECHLER	46.93
4		42.27
	D.BUTLER	42.37
	J.BERRY	42.19
	J.SCHUM	42.25
5		42.05
	J.LOCKE	42.05

The above table breaks down the Tukey test results.  There is a decent break down of punters in the league, with there being 5 different groupings.  However, much like place kickers where there are some very good, and some very bad players earning money in the league, a majority of them all fall together, with a bit of luck helping to determine a punt travelling either 45 or 43 yards on a play.

This of course is a fairly rudimentary analysis.  A punt that travels 35 yards might force a fair catch when one that goes 60 outkicks the coverage units and allows a big return.  IT would be interesting to see as well with more detailed data if some punters are sacrificing some distance for more control, like pinning in a returner on the sideline versus kicking it straight down the middle.

College Football Rankngs

After 5 weeks of play, its time for preseason projections to be removed from the program's system.  Team's are who they are now, no longer can fans be buyoed by the small sample size of games and plays. 

Top 10:

1. Alabama
2. USC
3. Oklahoma
4. Michigan
5. Ole Miss
6. Boise State
7. Stanford
8. Georgia
9. Baylor
10. Texas A&M

The only two here in the Top 10 that I really have any issue with are Boise and Georgia.  Boise I think is sneaking in by virtue of blowing out a bad Virginia team, and following it up by blowing out Hawaii.  Georgia is a very dangerous team, but Lambert has been a mixed bag.  When the game has been close, he's been quite good completing 67% of his passes for over 8 yards per attempt.  But he's also only thrown for more than 150 yards once this season, and when his team really needed a great performance against Bama he fell very flat. 

The Bottom 10:

1. North Texas
2. Georgia State
3. UTEP
4. New Mexico State
5. Wyoming
6. Kansas
7. Charolotte
8. Old Dominion
9. Idaho
10. South Alabama

Both of the bottom two schools are currently rated as worse than an average FCS opponent.How far the Mean Green have fallen.  In 2013 this team was able to average more than 30 points a game, and play solid defense.  This year they gave up 62 points to Iowa, and are currently averaging just .216 points per play.  I'm very disappointed in Georgia State.  I really thought this fledgling program could take a bit of a step forward, but a loss to an FCS school after dropping the opener to FBS newcomer Charlotte has doomed this campaign to never reaching the 3 win mark.

College football Over and Under Achievers So far

So, I really should post more.  However, life continues to get in the way, as work schedules and house moving have conspired to keeping my normally free weekends quite packed.  I have however carved out enough time to look through some of the programs ratings, and see how far some teams are from the preseason projected rating, for better or for worse.

At the Top:

1. Air Force
2. UNLV
3. MTSU
4. Boston College
5. Ole Miss'

Most of these teams are playing much better than the program had predicted them to be, which is most of them were supposed to be pretty dreadful instead of slightly above average.The Falcons have certainly looked like a very solid team in the first three weeks, crushing FCS Morgan State, and running away from San Jose State.  However, it was probably the absolute stuffing of Michigan State's run game that has brought the most shock, as Sparty managed just 1.8 yards per carry and suffered 8 tackles for loss to the Falcons surprisingly stingy defense.  UNLV is really a surprise on this list, but its mostly because the program expected them to be god awful, and they've only been miserable.  The small sample size is also probably helping them with that 80-8 win over Idaho State surpassing what they should have done against an FCS school.  Ole Miss is here largely for beating Alabama (predicted National Champion) handily.

Those at the bottom:

1.  Oregon
2. North Texas
3. Washington State
4. New Mexico State
5. Georgia State

 As Bill Connelly mentioned in his article, Oregon's defense has yet to stop anybody.  The offense is averaging an acceptable .443 points per play (only good enough for 45th in the nation), while ranking 108th by allowing .543.  North Texas has the distinction of being one of the five teams rated worse than an FCS school (on a neutral site, an FCS school would be favored), and is the second biggest underachiever by just a small margin.  Wazzu actually lost to an FCS school, at home this year, and Georgia State opened the season by dropping a game to FBS newcomer Charolotte.  That loss is really disappointing, since I had Georgia State going over 2.5 wins this year, and its going to be tough to get it now. 

Best Bowl matchups

The folks over at the SBNation main site released their big college football preview.  One of the most interesting pages that they do annually is their bowl projections.  I went ahead and categorized the matchups they gave by how close the projections put the bowl game being, as well as what the spread should be.

Team 1	Opponent	Team 1Win Chance	Average Spread	Departure from Even
Michigan State	UCLA	47.40%	-1	2.60%
Western Michigan	Appalachian State	52.90%	1	2.90%
Georgia Tech	Texas	53.00%	1	3.00%
Wisconsin	Arizona State	45.40%	-2	4.60%
Michigan	Georgia	44.60%	-2	5.40%
Oklahoma	Southern California	56.00%	2	6.00%
Central Florida	Bowling Green State	56.00%	2	6.00%
Kansas State	California	56.90%	3	6.90%
Minnesota	Arkansas	41.50%	-3	8.50%
Nebraska	Texas A&M	40.80%	-3	9.20%
Utah State	Utah	40.10%	-4	9.90%
West Virginia	Missouri	39.30%	-4	10.70%
Oklahoma State	Louisiana State	38.50%	-4	11.50%
Louisville	Tennessee	38.30%	-4	11.70%
Alabama	Oregon	62.70%	4	12.70%
Texas-El Paso	New Mexico	35.80%	-5	14.20%
Boise State	Notre Dame	33.90%	-6	16.10%
Virginia Tech	Penn State	66.40%	6	16.40%
Toledo	Georgia Southern	67.00%	6	17.00%
Duke	Mississippi State	31.90%	-7	18.10%
Iowa	Arizona State	31.40%	-7	18.60%
East Carolina	Louisiana-Lafayette	69.70%	7	19.70%
Florida State	Auburn	70.40%	7	20.40%
Baylor	Mississippi	72.20%	8	22.20%
Arkansas State	San Jose State	72.90%	8	22.90%
Houston	South Carolina	25.20%	-9	24.80%
Temple	Rice	75.50%	10	25.50%
North Carolina State	Stanford	23.20%	-10	26.80%
Northern Illinois	Colorado State	77.20%	11	27.20%
Cincinnati	Western Kentucky	78.90%	11	28.90%
San Diego State	Texas State	79.40%	12	29.40%
Ohio State	Texas Christian	79.50%	11	29.50%
Pittsburgh	Memphis	81.00%	12	31.00%
Maryland	Air Force	83.60%	14	33.60%
Miami(FL)	Marshall	85.80%	15	35.80%
Middle Tennessee State	Akron	86.10%	15	36.10%
Old Dominion	Brigham Young	10.90%	-17	39.10%
North Carolina	Navy	93.20%	21	43.20%

The best matchup in their projection is a battle of old school powers in Michigan State and UCLA, while the UNC-Navy game would be one to skip.

NFL Preseason to Regular season correlation

So, there have been a few posts over on MileHighReport talking about the team's preseason performance so far.  There's been some speculation that in recent years defensive stats have been a decent indicator of regular season performance.  I wanted to test this a bit, checking to see if there was any real link between the two.

Going back to 2012, I found total yards allowed per game in pre and regular season, as well as yards per play.  The results were not encouraging.  In the following chart, we'll see Yards per game has an R^2 of just .11, an extremely basic correlation.

 

Yards per play is even worse, with an R^2 of just .03.

We shouldn't expect there to be any real predictive power for preseason play.  Schemes offensively and defensively are vanilla through and through, with no game planning to attack weaknesses.  You also get extreme variance with a lot of guys playing extended snaps, who will likely be selling cars in a few weeks.

College football Top 25 Projection

This summer has been busy.  I have managed however to find some time to tweak the formulas I've used to predict team strength and come up with some composite rankings.  The changes centered a lot on removing some emphasis on turnover luck (it really has a very low significance for explaining year to year variability), and redid my returning experience formula.  It should help give a much better view of a team over a season, since the formula had a new reduced standard error equal to about half of what it had been.

In season I'll be making a few tweaks to what happens for a team afflicted with injuries, which of course will as the old military adage goes, ruin this ranking at first contact.

Team	Ranking
Alabama	1
Ohio State	2
Southern California	3
UCLA	4
Oregon	5
Oklahoma	6
Baylor	7
Stanford	8
Florida State	9
Notre Dame	10
Georgia	11
Texas A&M	12
Texas Christian	13
Mississippi	14
Arizona State	15
Clemson	16
Oklahoma State	17
Michigan	18
Texas	19
Missouri	20
Texas Tech	21
Michigan State	22
Florida	23
Arizona	24
Miami(FL)	25

It loves the Pac-12, which I think will continue to shine.  I do doubt though we see USC, UCLA and Oregon in the top 5 or so at the end of the year.  I'm also not nearly as high on Florida, but they could be a darkhorse with that defensive secondary.

The program also generates odds for a team to win a range of games, which I'll sort by below, going from the best chances to win at least 11 games in the regular season.

Team	Ranking
Ohio State	1
Florida State	2
UCLA	3
Alabama	4
Southern California	5
Oklahoma	6
Notre Dame	7
Baylor	8
Marshall	9
Nebraska	10
Boise State	11
Penn State	12
Georgia	13
Cincinnati	14
Stanford	15
Oregon	16
Massachusetts	17
Texas Christian	18
Wisconsin	19
Michigan	20
Northern Illinois	21
Michigan State	22
Clemson	23
Texas A&M	24
Oklahoma State	25

Yetisburg

 Sometimes you see a game at the store and the name just begs you to buy it.  That game a couple weeks ago was Yetisburg, a civil war combat game with a fantasy setting twist.  The Union and the Confederacy square off with the aid of melee combat specialist Yetis and long range artillery Mastodons.  

Each player has access to identical armies, each with infantry and cavalry (regular and elite troopers, the elite being able to withstand more damage before needing to head to the hospital), Generals, Mastodons and Yetis.  Game play is pretty straightforward.  Each player has 5 columns of troops, 3 rows deep.  The first and second rows, the front and reserve lines, are face up and can make attacks or take advantage of abilities.  The third row is the supply line, face down cards that will be used to replenish the first two rows when when those cards are killed in combat.  Player's take turns attacking by playing a card from their hand, and ordering all troops of that card type to attack.  Results of the attack are created randomly, by drawing a meat token and flipping it over to see what direction the shot went, and what the range of the shot is.  Damage is dealt by counting the number of rows in the direction listed, according to the units ability. Regular troops all deal one damage while yetis and mastodons deal damage equal to the range. Killed troops are sent to the discard pile, and lines are replenished.  This goes back and forth until both players have either exhausted all their troops or both players pass their turn to attack. After that, players can use their Generals in the field to refresh troops for the next turn, swap positions, or bring troops back from the discard pile into the fight. The game continues on until someone cannot reinforce their lines from the supply line, or when drawing cards for their hand cant draw anymore.  

 Yeti combat is a little different from regular troops. They move to a row in between the players front lines called the trench.  Yetis crash through their own lines, killing friendly troops to get into the trench, and if their is an opposing Yeti in the trench for their column, they will knock them back to the opponents front lien, killing any troops that are there.  If a Yeti gets hit by friendly fire, they swap sides, going under control of your opponent until they get hit again.  Yetis only deal damage to the units directly next to them.

My game group has played this game now a few times and has come to a pretty much unanimous consensus: The game is a a fun, quickly paced short game, but the randomness doesn't leave you really wanting to play it again and again.  The combat mechanic makes it extremely likely you're going to on more than one occasion shoot your own troops, especially those at the left and right of them.  If there was a way to somehow get a little more strategic control of how units were shooting the game would be a bit better for replay value, especially since the artwork on the cards is so fantastic.