============================================================

2008 COLLEGE FOOTBALL - ALL DIVISION CONSENSUS RANKINGS

2008 Consensus College Ranking

============================================================

HOW IT WORKS (Updated 01/29/06)

HOW IT WORKS

-------------------------------------------------------------------------------

EXPLORING HOME FIELD ADVANTAGE

Sooner or later, most sports handicappers resort to using some type of value which is applied to teams playing at home in order to assist them with game predictions. But ask 10 different ranking experts how many points they award for home field advantage and chances are you’ll get a minimum of five or six different answers. Over the years, I’ve come across systems that allow as many as seven to 10 points and some that allow as few as none. Others employ calculated, unique point assignments for each team. However, the most common response you’re apt to hear is simply, “three.” Why three? Is this an average? Or perhaps this is a magical number a mathematician devised that just seems to work? Chances are, the number “3” was derived from a little bit of each and passed on as a consensus over the years.

COLLEGES

The LazIndex has used a three point standard for college teams playing at home and a one point average for high school home teams for several years. But just how accurate is this method?

A re-examination of the 2004 football season brought to light some very interesting data. By extracting the Power Rating performances of each school at home vs. on the road for all 707 college teams, it was found that 544 (77%) did perform better at home than on the road.

A decent percentage but this still leaves 163 teams that played better as road warriors.

In 2004, the average variance between home and away power ratings for all 707 colleges was found to be +2.19.

A deeper look revealed the most startling information. Of the 707 schools, only 277 (39 %) actually played three or more points better at home than on the road. The problem now is that the LazIndex power rating values assigned to each team are a COMBINATION of home and road performances.

For example, Hawaii had the greatest home/away variance in all of college football.

Hawaii 109.36 Combined 100.65 –Away 113.23 – Home +12.58

However, because the “combined” power rating calculated for Hawaii was 109.36 we cannot just add 12.58 points for their home performance because at no time did they play to a level of 121.94. But should Hawaii’s overall power rating be adjusted to 100.65 and add 12.58 for home field? Sounds possible but let’s continue. If we consider the reverse scenario, Georgia Tech had the worst home/away variance of all Division 1A schools.

Georgia Tech 118.74 Combined 120.58 – Away 116.89 – Home -3.69

Is it conceivable to SUBTRACT 3.69 points from their away power rating of 120.58 or even worse from their combined power rating of 118.74? The latter method would create a level of play lower than they performed all season.

Let’s move on. There is a noticeable drop in margins as we move down through the various classifications of college football. Here are the average margins for each division

D1A 112.64 Combined 110.82 – Away 114.57 – Home +3.75

D1AA 86.19 Combined 85.07 – Away 87.28 – Home +2.21

D2 74.87 Combined 74.03 – Away 75.78 – Home +1.75

D3 52.45 Combined 51.56 – Away 53.34 – Home +1.78

NAIA 55.91 Combined 54.98 – Away 56.87 – Home +1.89

I believe there are two major factors that can be attributed to the decline in home field advantage

throughout the lower divisions of college football:

1) Travel Distance

2) Attendance

Geographically speaking, Division 1A conferences cover vast expanses of the country as compared to the lower divisions in college football which are typically confined to smaller regional areas. For example: The Division 1A conference with the highest home/away variance is the Western Athletic Conference (+ 6.40). This conference consists of 10 schools in seven states covering an area ranging from Hawaii to Idaho to Louisiana. Traveling these distances has to be a tremendous burden on football team’s week in and week out. Compare that with more regionally located D1 conferences, the Southeastern (+1.88) and the Atlantic Coast (+1.89) and you begin to get the picture.

In addition, the enormous home attendance in Division 1A easily dwarfs in comparison the size of the home crowds at the lower level universities. The combination of noise level and intensity (emotion) level both play vital roles in the home performance of D1A teams.

HIGH SCHOOLS

I reviewed the results of 475 Florida school teams on all levels that played games both at home and on the road. Of the 2,095 Regular Season games analyzed, Home teams won 1103 games (52.6%) and Away teams won 992 games (47.4%).

However, of the 475 schools only 210 (44%) actually played one or more points better at home than on the road. In fact, 209 (also 44%) of the 475 schools actually had a LOWER home field power rating than they did when playing away from home. Here is the Florida high school summary from 2004:

43.34 Combined Power Rating 43.08 – Away 43.65 – Home +.57

ANALYSIS SUMMARY

The data extracted from 2004 clearly indicates the home field factors I have been using are overstated. As mentioned above, The Lazindex deployed a three point home field advantage for college and allocated one point for high school home teams. This analysis proved that of the 1182 football teams I ranked last season, 695 (59%) did not achieve their assigned level of home field advantage power rating points. It gets worse.

Assuming each team plays half of its games on the road and the other half at home, and because the Lazindex power rating is a combination of both those results, the actual variance should have shown much higher results. For example, if college Team A has a power rating of 90 and is allotted 3 points for home field, the formula is “assuming” that Team A plays at a level of 87 on the road and at a 93 level at home. This means there should be a six point margin in the performance levels for college teams depending on where the game is played. In actuality, only 80 (11%) of the 707 colleges were able to achieve a six point or greater margin. In high school, where there was a 1 point home field allotment, teams should have experienced a two point rating variance between home and away. The study showed only 160 (34%) schools actually achieved that margin.

THE IMPACT

It’s important to understand that the sole purpose for developing home field advantage values is to assist in predicting outcomes and point spreads. These values do not affect the Lazindex ranking process or the overall power rating formula. However, these rating variances are probably significant enough to alter at least a handful of contest predictions each week. This can add up over time and could have substantial impacts on year long predicting accuracy.

THE SOLUTION

I have examined the data and there is definitive evidence that nearly 70% of all the teams ranked by the Lazindex experienced some degree of home field advantage during 2004. And although there is no sure fire solution to this problem, rest assured improvements will be made next season. It is my belief that each team should be assigned some additional value for playing at home and therefore I will continue to do so. However, these values will be monitored as the season progresses and the assignments will change during the course of the year, perhaps as often as each week. Initial values will possibly be assigned in the following manner:

D1A colleges +2.0

Other colleges +1.0

High Schools +0.3

Thanks for checking in- Laz January 2005.

--------------------------------------------------------------------------

HIGH SCHOOLS - PRIVATE vs. PUBLIC 2004

Review The Numbers

----------------------------------------------------------------------------

TESTING THE NEW FORMAT

In attempt to prove the new format will result in improved ranking accuracy, I took 11 of the 12 ranking systems that rank ALL levels of college football and ran consensus numbers. The link for the 12th system (Greg Holland) was not available so I could not include it in the consensus test.

The remaining 11 systems (Armstrong, Born, Dolphin, LazIndex, Massey, Montgomery, Rothman, Saucedo, Sorensen, Wilson and Wolfe) were used for the test. Each system ranked at least 688 teams from all levels of college football in 2003.

Taking the data from each of the systems, I compiled a consensus of the 688 teams using the “old” LazIndex format (Power Rankings Only). Once calculated, I found that the Saucedo system was closest to the consensus with Dolphin coming in a close second. The LazIndex came in 8th.

The season was then recalculated utilizing the “NEW” LazIndex format (With Winning Edge Factor) and a new consensus was run among the 11 systems. The results exceeded expectations. The “new” LazIndex was closest to the consensus with Dolphin maintaining second.

System Fact. Deviation W/Old LAZ Format

Saucedo 0.411

Dolphin 0.430

Sorensen 0.447

Rothman 0.466

Massey 0.531

Wilson 0.548

Montgomery 0.565

Laz - Old 0.668

Born 0.918

Armstrong 1.124

Wolfe 1.156

=======================================================================

System Fact. Deviation W/New Laz Format

Laz - New 0.351

Dolphin 0.421

Saucedo 0.445

Sorensen 0.456

Rothman 0.487

Massey 0.524

Wilson 0.539

Montgomery 0.592

Born 0.938

Armstrong 1.094

Wolfe 1.117

HOW CONCLUSIVE WAS THE TEST ? AND WILL LAZ BE THE BEST ?

Although these results were definitely encouraging, only time will tell how the new format will work. There is much more to the ranking process than these subjective 11 systems. This test was conducted among systems that rank ALL the schools. There are VISIBLE FLAWS in several of these models, which are somewhat disturbing. Not necessarily with how each Division is ranked internally, but how each team fits into the overall ranking. In other words, where does each school rank among the 688, regardless of division ?

For example, St. John’s (MN) was the consensus # 1 in Division III. Yet, among these 11 systems, the Johnnies’ overall ranking ranged from # 103 to # 212. That’s quite a range of variance. Even worse, take the case of Siena the consensus sorriest team in 1AA. Their overall ranking varied almost 200 positions (# 467 to # 656) among these 11 systems. Where should they really be plotted in the overall scope of 688 schools?

Further evidence of these flaws can be easily pointed out simply by looking at Kenneth Massey’s performance against the consensus of these 11 systems. It is accurately documented that Massey’s system ranked #1 overall against the consensus of 97 systems which ranked Div 1A in 2003. Yet, among these 11, he finished in the middle of the pack. There’s no way he can be that far off.

While I am very anxious and optimistic about the new format, we’ll have to wait to see what lays ahead, The LazIndex remains committed to achieving continued ranking accuracy improvements and will strive to be the most relied on system which ranks ALL levels of college and high school football.

March 28, 2004

From Jan. 31, 2004 -

WHY THE CHANGE?

This past season, the LAZ INDEX consistently performed in the 75% to 80% predictive range, however there has been a great deal of negative feedback and criticism in the way the system “ranks” the teams. This is because the rankings are calculated utilizing only the power rating.

What is a POWER RATING?

A power rating is the value assigned to each team based upon the scoring margin and schedule strength of the subject team and it’s opponents.

Several times over the past two seasons, I have explained that the power rating each team establishes, is an average representation of performance capabilities and potential. This rating is used to determine the predictive outcome of each contest. Typically, the power rating assigned by the LAZ INDEX can accurately predict the winner in about 80% of regular season games and is slightly less accurate (75% range) in the postseason.

What is the PROBLEM?

The power rating, though consistently accurate picking winners, does not actually incorporate wins and losses into the ranking equation. This leaves certain teams “out in the cold”. In college, let’s take Ohio State for example. This team has caused handicappers a great deal of headaches in the past because OSU plays to the level of their opponents each week. This causes their power rating to be lower than most good teams. They don’t win by much but they WIN.

On the other hand, the power rating rewards teams for keeping games close.

In Florida High School play, let’s look at Bradenton Manatee. They play a tough schedule and perform to the level of their opponent each week but often come up short in the “W” column. Their power rating is higher than most average teams.

The SOLUTION

Next season, The LAZ POWER INDEX will incorporate a “Winning Edge Factor” into its ranking calculation. This factor will utilize the power rating and the actual win-loss performance to create an overall ranking for each team. You will not be able to mathematically predict results using the Ranking number however, the Power Rating will still be provided for the sole purpose of performing that calculation.

WILL IT WORK?

Initial results have been very encouraging. You can review the actual new format I am using for NCAA Division 3 Men’s Basketball on the D3 Basketball page.

Additionally, I “replayed” the 2003 NCAA and Florida High School seasons and the results were definitely significant. I will not publish the entire listing because the season is in the past, but here is a look at how the top 3 football teams in both college and Florida high school would have looked had this system been implemented:

COLLEGE

1* LSU* 1.8405* 13* 1* 138.89

2* Okla* 1.8221* 12* 2* 139.30

3* USC* 1.8085* 12* 1* 138.02

HIGH SCHOOL

1 Mia Carol City* 1.1608* 14* 1* 78.82

2 Armwood* 1.1580* 15* 0* 81.72

3 Chaminade* 1.1323* 13* 1* 81.33

This change is the first major change the LAZINDEX has ever gone through.

It won’t be the last. –

Laz ...

January 31, 2004