1. When did you start writing your system?
2. How does the system assess ratings to teams?
3. How does the system figure out predictions?
4. What is the overall winning percentage for your predictions?
5. Are there any biases towards a specific team or conference?
6. What do the fields mean on the ratings page?
7. What do the fields mean on the predictions page?
8. Who do I contact if I see an error in the ratings or for more information?
9. What language is your system written in?
10. Do you know of a website where I can find historical scores?
11. Do you know of a website where I can find historical Vegas spreads?
12. Do you back up or guarantee any of the information provided in this website?
13. What are your personal opinions on sports, football, and the BCS?

Below are the answers to a set of frequently asked questions (FAQ) that I receive regarding my website. If you have a question about the site, please check here first, and if you don't find the answer to your question, send me an email at This e-mail address is being protected from spam bots, you need JavaScript enabled to view it .

1. When did you start writing your system? I started writing the system during the 1995 college football season. My initial reason for doing so was that I got fed up with reading the weekly polls for college football and seeing teams ranked much higher or lower than I thought they should have been. I had known about the Sagarin and NY Times computer systems since the late 80's and had heard that more people were beginning to try their hand at this artform and decided to give it a go.

1. How does the system assess ratings to teams? First off, I look at each game and how each team fared score-wise. Then I compare the each team's performances in that game to their opponents average against that statistic. IOW, if Team A beats Team B 30-10, and Team B usually gives up 25 pts/game, then Team A earns 5 points for the game for having scored 5 points better than Team B's average pts/allowed statistic. I also assess a win bonus or a loss penalty (no bonus or penalty for ties) to each team, give a modest bonus to away teams (or a lesser bonus for playing on a neutral field), assign a "game difficulty bonus" for playing good teams, then tally up all of these calculations into a game score. I then take a weighted average of the games played with more recent games counting more than less recent games. Strength of schedule is then determined on a similarly weighted average, and then I publish the results on the website.

1. How does the system figure out predictions? This is much harder to describe. Jeff Sagarin's system is purported to be able to get a spread by simply subtracting the lower ranked team from the higher one, but such is not possible, nor even recommended with my system. In Sagarin's system, the higher ranked team is almost always going to be favored by him, however there are cases where a much lower ranked team in my system can be favored, based on the types of teams playing, where they play, and when they play. I find it much harder to win big at the end of the season compared to at the beginning, usually from loss of key personnel due to injury or due to colder weather that occurs in many parts of the country in starting in late October. That being said, I make no direct accomodation for injuries or weather, but use a reduction algorithm to lower the expected spread of every game later in the season. In addition to that, I take into account the ratings of the teams, their average offensive and defensive statistics and the use of the standard deviation of that average, the site of the game, and a few other factors. Overall, the final predictions usually favor the higher ranked team, but not always.
2. What is the overall winning percentage for your system? In college football, my straight up winning percentage averages around 75-80% for the year without too much deviation from that average. Against the spread, I typically run around 53-56%, however there are usually some big deviations from week to week in that category -- it's not uncommon to have a 65% week followed by a 35% week. Overall, the predictions are fairly accurate, IMHO.
3. Are there any biases towards a specific team or conference? There are no inherent biases to any one team. I am a Ohio State fan and alumnus, but in my system, OSU is just another meaningless name for a team that plays college football. They could just as well be called Team #73, or Upper Arlington, or Bob -- the system could care less. However, I do have certain biases within the system that are built into the algorithms in order to determine if a team is playing well or not. In fact, every single computer rating system has this set of biases as well, otherwise it would be pointless to try to determine strength and fitness. For a better description of my biases, please check out my Opinions page.
4. What do the fields mean in the ratings page? Going from right to left, you have Rank, Team, W-L-T, Rating, and Schedule. The first three sets of columns are obvious, and just for consistency, I leave a Ties column in regardless of whether ties are possible, only because it would be too time consuming to build and maintain vastly different programs for each sport. Following that is the Ratings section. I start with the current rating and in parentheses, I give the change in rank and rating from the previous day's or week's ratings. Finally is the Schedule column which is Strength of Schedule. Overall, it's fairly easy to read, in my opinion.
5. What do the fields mean in the prediction page? The prediction page is a bit more confusing. For starters, I always list the Home team first instead of putting the favorite first. I feel this makes the layout more consistent, although some still have problems reading the layout. Just remember, Home First. Inbetween the teams, I have the Vegas line for that matchup. A negative line means that the home team is favored points, and a positive spread means the away team is favored. An alternative way to look at it is that the spread refers to the amount of points that the home team either gives up or receives. Following the away team is the "CR Line" which stands for "CompRank Line" (cheezy, I know...). That is the system's prediction of the game's final spread. Next is the "Conf" column which is a measure of my confidence that the team I've picked with the CR Line is going to beat the Vegas spread. Finally, I compare my CR Line vs. the Vegas line and put the team in CAPS that the system predicts is favored vs. the Vegas line.

   For example, if Team A is at home, and is favored by Vegas by 3 points over Team B, and the system predicts Team A will win by 7, then overall, it picks Team A to cover the 3 point Vegas Line and capitalizes their team name, as such:

Home Line Away CR Line Conf

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

TEAM A -3.0 Team B -7.0 0.612

1. However, if the system figures that Team B will come within the 3 point Vegas line, such as losing by 2 or less or winning straight up, then it favors Team B, and the line looks like this:

   Home                     Line  Away                   CR Line   Conf ==================================================================== Team A                   -3.0  TEAM B                  +2.4    0.732 

   Of course, there are situations where the two teams already have all their letters in ALL CAPS, such as UCLA, BYU, or TCU, and if they play (or even play each other!), then the sure fire way to tell who is favored is to use the formula (Line - CR Line) and if positive, pick the home team, if negative, pick the away team. This always works.

   On the post-game analysis, I also add four more columns to the page, "Score", "Str", "ATS" and "Anl" columns, which stand for Score (duh), Straight Up, Against The Spread, and Analysis. The Score column is self-explanatory. "Str" denotes whether my CR Line pick was correct (W or L), ATS for whether my CR Line pick won, lost or tied against the vegas spread, and "Anl" is my analysis of the CR Line pick. A "U" in this column denotes an upset vs. the vegas spread occurred, a "U+" means the upset resulted in a win ATS for my pick, a "U-" resulted in a loss ATS. An "O" means that one or both teams either overplayed or underplayed compared to how they had been playing so far that season in that game. And a "CR" in that column means that my system picked wrong (Vegas spread was either more accurate, or the system predicted an upset that did not occur). At the bottom of the analysis page is the overall results for that set of games and it is mostly self-explanatory.

2. Who do I contact if I see an error in the ratings or for more information? You can always contact me at This e-mail address is being protected from spam bots, you need JavaScript enabled to view it for any question regarding the Comprank system. I read email almost daily and will respond to every email about the system.

   Telemarketers beware -- I track and reserve the right to prosecute all senders of unsolicited commercial email sent to me. It is a waste of my monthly bandwidth and I have no time to deal with unwanted garbage.

3. What language is your system written in? The current implementation of my program is written in Java. I rewrote the program from its original coding in C at the beginning of the 2001 college football season. The original code began to become subject to the Big Ball of Mud phenomenon that many programs tend to happen to programs over their lifetimes, and I decided that it was time to rewrite. I chose Java because it was clean, well-supported, had possible marketable features for having that knowledge, but most of all, to learn a new language and keep my coding skills current. Nothing like new syntax to help refine your skills. The new code is now highly coherent, loosely coupled, modularized, portable, and doggone it, it's just pretty.
4. Do you know of a website where I can find historical scores? James Howell, custodian of the Howell Power Ratings System has a comprehensive database of nearly every college football game ever played, dating back to 1869. You can find the scores here.
5. Do you know of a website where I can find historical Vegas spreads? This is harder to find. I don't know why, but I have found it incredibly difficult to find historical spreads on the Internet. Just about every other piece of information known to man can be found on the Internet, but spreads remain elusive. My advice is to spend some time at the local library and pull the microfiches from your local newspaper and scan the sports section for the lines. If you do find and compile a nice database of spreads, blog it, publish the information to the rec.sport.football.college newsgroup, or submit the finding to one of the search portals, like Yahoo or Google. This is one piece of information that many are seeking, but not many have.
6. Do you back up or guarantee any of the information provided in this website? There are no guarantees provided or implied in any part of this website. All information contained within is for entertainment purposes only. Any use of this information for purposes other than entertainment are the sole responsibility of the user.