is the world's most
complete repository of interesting college football statistics and
statistical analysis. When used properly, it will make readers smarter,
more knowleadgeable, more insightful, and, in most cases, better
looking. This brief guide explains the basic features and provides
more detailed explanations. Links
followed by (g) point to a glossary entry.
In the top bar you will
find links to key
summary statistics, sorted
by individual, team, and conference.
, and Receiving
listed in sortable columns. In each case,
there are two options for setting the number of attempts/receptions
necessary for a player to qualify.
c) Completion Percentage
completions per pass attempt
: completion% adjusted for the opponent's
: completion% adjusted for yards per completion
: completion% adjusted
for yards/comp. and opp. pass D rating
e) Opp. Comp% Allowed(g)
Average completion% allowed by opponents, weighted by the number of
passes thrown against that opponent.
f) Average Opp. Rating(g)
individual quarterback's strength of schedule
b) Rec Yards
e) % of Team's Receptions
f) % of Team's Receiving Yards
B) Heisman (g)
- The Stat-Wise
Rankings rates every offensive
awarded for offensive
production and contribution to a team's success.
Fantasy projections and rankings for the upcoming week.
ranked. All rating and ranking systems except the cRPI are based on
iterative fit models. Click "(g)" for more information.
i) BPR (g) -
- More team rankings.
ii) Power (g) -
iii) Hybrid (g)
iv) SOS-BE (g) -
v) SOS-Und (g)
- An Adjusted rating of the scoring
offense and defense, adjusted for game tempo.
vii) Run/Pass/Run D/Pass D Eff (g)
and pass efficiencies based on
yards per play.
viii) cRPI (g) -
B) Summary -
i) Season, Season+ - Projected record for regular
season games and regular season plus any
conference championship game.
ii) Win Prob. Div, Conf -
represent the division or conference.
iii) Bowl Eligible - Probability
be bowl eligible
iv) Undefeated Reg/Post - Probability
be undefeated at the end of the regular
after any potential conference championship game.
v) Win Probability
Probability that a team will finish the regular season with that
number of wins
- Score predictions for any
possible pairing of FBS teams at a neutral site.
probability of winning a national championship. Play - a
team's odds of playing in the national championship game. Champs - the
odds the team wins the championship game. Opponent - the team's most
- see Team:Rank
. SOS is the strength of the
i) P-Top, -Mid, -Bot - the
strength of the top, middle and bottom of the conference, respectively.
You can find all past blog entries
in the middle of the right column in the dropdown menu labeled "More
1) Team Pages - Links to team pages.
2) Past Rankings -
all teams (with sufficient sample size) since 1900. Teams are ranked by
hybrid(g), cRPI(g), performance(g), elo(g), offense and
3) Maps: Off, Def (g)
- Maps locate teams base on their
style of play on offense and defense. Teams with similar styles are
close to one another.
4) In-Game Odds (g) - Allows the
user to track the the probability that a team will win and cover, and
the expected margin of victory (updated line) throughout the course of
5) Links - A list of important external
Adj. Yards/Carry - yards per carry
is adjusted for the strength of the opponent's run D.
Adjusted+ - A measure of quarterback
accuracy. The adjusted+ completion percentage is a quarterback's raw
percentage adjusted for both the yards per completion - longer passes,
ceterus paribus, are more difficult to complete - and the strength of
the pass defense the quarterback faced.
- A BCS style ranking system with an
twist. This is my preferred system for ranking teams, but it is not
intended to be predictive.
Champs - Essentially, this is the
probability that a team will play in/win the national championship.
These probabilities are calculated in three steps. First, the current
season, up to the bowl games, is simulated thousands of times. Second,
team's are ranked by their performance in the season, and the top two
are selected to play for the national championship. Finally, the
national championship game is simulated. By far the most
computationally intensive and error-prone step is in ranking the
team's, but champ odds uses a slightly modified variation of the hybrid
ranking that is very successful in identifying the top two teams using
[*] - The cRPI is a technical variation on the traditional RPI. Like
the RPI, the cRPI is a formulaic rating system that considers only wins
and losses, not the margin of victory. The cRPI, though, corrects one
critical flaw of the RPI. A team's RPI can be hurt after a win, or
improved after a loss if the opponent's win percentage is sufficiently
low or high, respectively. Instead, the cRPI awards or subtracts points
based on the quality of the opponent, so a win against a soft opponent
earns a team few points while a win against a tough opponent is worth
more points. "cRPI*" = cRPI*100.
Explosive - A measure of a backs
explosiveness. A higher value means that a running back is more likely
a long run. It does not mean that a ball carrier will have a high
average yards per carry.
Hybrid - The
hybrid is the rating system that
best reflects national polls. The Hybrid rating is a compilation of the
elo and power
ratings, win/loss record, and strength of
schedule. I post updated ratings for every FBS team this season and
since 1900, and also compute ratings for all NCAA teams in the current
In-Game Odds - During a game, the odds that
a team will win or lose can rise and fall. With the in-game calculator,
you can calculate at any time the odds that a team will win and the
expected score differential at the end of the game. Given the pre-game
line, you can also calculate the odds that a team will win before the
based on tempo,
run to pass ratio, run and pass effectiveness, etc., the maps locate
teams so that similar teams are close to one another and dissimilar
teams are far apart. The maps are used in team pages to visualize
potential matchup issues.
Ratings - these ratings are calculated using an iterative fit
model that minimizes the error between predicted points scored based on
these ratings and actual points scored. These are then adjusted for the
number of possessions in a game.
Offense/Defense ratings are a key component of the power
ratings, hybrid rankings (as a componenet of
the power ratings), and predicted game results.
Opp Adj. - A measure of a quarterback's
accuracy. The Opp. Adj. comp% begins with the raw comp% and adjusts it
for the quality of the opponent's pass defense. Opponents' pass
defenses are rated using an iterative fit model to minimize the error
between the expected completion percentage and the real completion
Opp Run D Rating - The opponent run D rating
is the average rating of the defenses a runner faces. This rating is
calculating by minimizing the error between the estimated yards per
carry (based on the run D rating and the adj
Opp Comp% Allowed - Opponent pass D ratings
are measured in two ways. The more simple is the opp. comp% allowed.
This is the average completion percentage allowed by a quarterback's
opponents, weighted by the number of passes thrown against that
opponent. The second method, the average opp pass rating, adjusts the
opp. comp% allowed based on the accuracy (Yards Adj.)
Eff - The efficiency ratings measure a team's
yards-per-play proficiency. An iterative fit model is used to minimize
error between the expected yards per pass/run (calculated by subracting
the defense's efficiency from the offense's efficiency) and the actual
yards per pass/run. These values are also used to adjust for game
specific-matchup issues - a team that depends heavily on the run will
perform relatively poorly if they are playing a team that is relatively
proficient at stopping the run.
Picks - Every week, CFBTN posts
predictions for every NCAA football game. For FBS games, these picks
are based on historical patterns, power ratings, performance trends, and matchups based on run and
pass efficiencies and ratios. For FCS, division II and division III
games are based on performance ratings.
- The power ratings ignore wins and losses and focusing on measuring
how a team will perform in upcoming games.
Pick All - The same methodology
that is used to make picks for real games is
applied to any possible FBS matchup.
Probability of Gaining . . . - based
on the distribution of yards per run for a back, we can calculate the
probability that a back will gain a certain number of yards on a given
play. We can also calculate the probability that a back will gain 10
yards on three plays.
- A team's strength of schedule in terms of becoming bowl eligible.
- A team's strength of schedule in terms of going undefeated.
Standings - In addition to posting
the standings, CFBTN also posts the probability for each team that it
will win outright or share a conference or division title. These
probabilities are calculating by simulating every possible outcome and
estimating the probability of that outcome.
Stat-Wise Heisman Rankings -
Individuals are rated based on the their cumulative statistics
(rushing, passing and receiving yards and touchdowns), their
performance in big games and losses, and the quality of their team and
the opposition faced.
- The Trend-O-Meter tracks a team's performance over the course of a
season using non-linear regression techniques.
- A measure of a quarterback's accuracy as a passer. Yards adj. begins
with the raw comp% and adjusts that percentage based on the yards per
completion. For example, a quarterback that completes a high percent of
short passes will see his completion percentage adjusted downward.