A Little Help…

First off, there is probably going to be a Poll Dancing post this week, but quite honestly, I haven’t had the time to get around to it.  But there is something I’d like to throw out for your input.  Today on Yahoo!, Dan Wetzel has a piece about what to do with the college postseason.  As we know, he favors a playoff but he offers something of an olive branch to the BCS stalwarts: an overhaul of the BCS ranking system.  Of the two ideas he suggests, I am most intrigued by the first: an open, objective computer-based system for comparing teams.  I had already started working on something like he proposes to feature here in the coming weeks, so why not take it to the next level?

That’s where you come in.  I’ve already started tweaking some initial ideas, but I’d like to know what you feel are the most important factors in determining a team’s success.  For example, I have already included third-down conversion/opponent third-down conversion, turnover margin, and points for and against.  I decided to filter these stats to include only games against FBS teams with a winning record, so blowouts of weak opponents will be ignored.

What else do you think makes a good team good?  Anything at all, even it isn’t statistical in nature, is welcome.  If there’s a way to quantify something, I’ll do my best to find it.


  1. My initial response to your filtering idea is to wonder about the fairness/consistency. Winning record… at the time the two teams play? At the time the algorithm is generated (dynamically changes from week to week?)

    And what if a team is tough one week because they have awesome personnel, but later get decimated with injuries and their record suffers? The team that lost to them early when they were stacked would be unfairly penalized in comparison. (Example: 2007 Oregon before/after Dennis Dixon got hurt.)

    But back to ideas on quantifying a “good team”:

    1. Some clutch variable would be good. (The adage of a good team “finding a way to win”.)Like 4th quarter comebacks, or games won on last drive, or scores in close time proximity to opponent scores (answering an opponent’s score quickly).

    2. Maybe something about 2nd half scoring? If scoring drops off in 2nd half… sign that coaching is weak. Tricky because dominant teams often play scrubs in 2nd half…

    3. A ratio of offensive-to-defensive/PR/KR scoring? A team with a high amount of non-offensive scores is likely more well-rounded than a team that only gets points via it’s offense.

    4. Some ratio that compares points with length of scoring drives? E.g., an offense that scored 45 points on drives averaging 40 yards may not be better than an offense that scored 35 points on drives averaging 80. The former was likely “helped out” by acts of chance while the latter means the team had long stretches of success.

    5. Acts of chance… Weather: how do you compare one team’s performance in a snowstorm against another team’s in a dome? A “weather impact factor” of some sort?

    6. Acts of chance… Officiating: What about games that are decided by poor officiating (OSU/Illinois 2007)? A good start for that might be to create some ratio based on points gained and/or allowed vs “opponent penalties” statistic. If a team has success and has a bad opponent penalty ranking, it’s an indication that the officials are trying to balance things out for the weaker teams.

    That’s just a start… wonder if anybody else has others…

  2. This would be intended for after the completion of the regular season only, when it actually matters, so all numbers will be final numbers not time-of-game numbers. However, some consideration for critical injuries or roster changes is not out of the question. Ideally (in my opinion) it would be used to fill in at-large slots for a 16-team playoff, but I would like it to produce a reasonable top 2 as well.

    #1 is worth exploring. I also thought of a “record-balancing” concept based on Phil Steele’s close wins/losses research. Basically, a team’s actual record would be compared to theoretical records if all close wins or losses had gone the other way. For example, OSU was 6-6 with 2 close wins & 5 close losses, so we could have been 4-8 or 11-1. Averaging all 3 gives us 7-5, which may be a truer representation of the team’s talent.

    #2 is tough but something with tracking scoring throughout a game would be good.

    #3 I like. Speaks to balance, which is definitely crucial. Maybe also look at passing vs. rushing balance.

    #4 is good too. Sustaining drives is key, but advantageous field position could just be defensive/special teams success.

    #5 if we can quantify weather impact, we deserve an award. At least a Grammy.

    #6 is intriguing. Will definitely mess around with that as it has been an issue with successful teams for years.

    • I think weather could be quantified… take a few statistical measures and create a metric for them. Precip, precip type, wind, temp. The sum or average the metrics. The ranges might be arbitrary but it’s better than nothing.

      The trick would be to baseline, wouldn’t it? So we define the perfect weather for football: dry field, winds of 10mph or less, 65-75 degrees. That impact factor is “1” or something. Anything above or below those baselines alters the factor.

  3. Monkey covered a lot of the thoughts I had but I was also thinking of a way to quantify the momentum of team, like Georgia this year. Early losses count “less” in the current system but should also count in the definition of a “good team”. Some kind of “improvement factor”.

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.

%d bloggers like this: