This site is supported by the advertisements on it, please disable your AdBlocker so we can continue to provide you with the quality content you expect.
  1. Follow us on Twitter @buckeyeplanet and @bp_recruiting, like us on Facebook! Enjoy a post or article, recommend it to others! BP is only as strong as its community, and we only promote by word of mouth, so share away!
    Dismiss Notice
  2. Consider registering! Fewer and higher quality ads, no emails you don't want, access to all the forums, download game torrents, private messages, polls, Sportsbook, etc. Even if you just want to lurk, there are a lot of good reasons to register!
    Dismiss Notice

Who Have They Played?! (Comparing teams through statistical analysis)

Discussion in 'Buckeye Football' started by DaddyBigBucks, Oct 27, 2022.

By DaddyBigBucks on Oct 27, 2022 at 11:29 PM
  1. DaddyBigBucks

    DaddyBigBucks Administrator Staff Member Bookie

    Who Have They Played?!

    The content here at Buckeye Planet is created (with apologies to Mr. Lincoln) by fans and for fans. The media will throw context-free numbers at you and call it analysis. Their marketers tell them to assume that their viewers have a 7th to 8th grade level of education. Seriously. Here at BP, we don't make any assumptions about who you are, except perhaps that you're like us. We like our guests with thick skin and sarcastic wit and our numbers with enough context to make them relevant.

    The talking heads of college football spew all sorts of stats all the time. When there are a few seconds of dead air to fill, some coiffed and tanned former jock will casually mention a statistic or factoid that's supposed to make himself sound intelligent and informed, and you to feel out of touch with reality, something like: "James Madison has the best rushing defense in the country." You hear it, and your preconditioned response is to throw something at your television and yell (scream it with me, everyone): "Who Have They Played?!"

    Who Have They Played?! That's always a fair question. Yes, we think that's even a fair question when it's asked of our beloved Buckeyes. It also happens to be an especially fair (and hilarious) question when the person doing the asking is Desmond Howard (whose Wolverines currently have the 78th ranked schedule according to Sagarin).

    One of the ways to dig into the Who Have They Played question is to compare a certain team to everyone else their opponents have played. While each team has played only six or seven games so far, that team's opponents have played about three dozen combined opponents at this point of the season. This comparison - opponents of opponents - injects some context into the statistics while also increasing the number of data points.

    Some notes on methodology before we get started:
    • Games against FCS opponents are factored out. For example, Toledo's shut-out of the Long Island Sharks is not considered. So the 27.25 points allowed per game that the NCAA credits to them becomes 31.14 points allowed per game.
    • Head to Head is factored out. So the Buckeyes' 77-burger is also removed.
    • The practical upshot is that, for this analysis, Toledo gives up 23.5 points per game to all FBS opponents not named Ohio State.
    • Ohio State's 77-burger thus amounts to a juicy 3.277 times as many points as Toledo's other opponents average against the Rockets. We call this the Buckeyes' Differential Scoring Offense (DSO) for that game.
    Just as one can calculate Differential Scoring Offense (DSO) for a single game, you can do the same for the season as a whole. The same can be done for Differential Scoring Defense (DSD); the difference being that higher numbers are better for DSO and lower numbers are better for DSD.

    Using these ratios to compare the Buckeyes and Nittany Lions, we get the following:

    Statistical Analysis MetricOhio StatePenn State
    Points Per Game - Offense49.57133.429
    Differential Scoring - Offense2.2351.344
    Points Per Game - Defense14.85718.857
    Differential Scoring - Defense0.5380.661
    We could stop here and feel secure that the above analysis shows Ohio State's superiority. Or we could go a step further and combine the above numbers to give us a crude prediction for Saturday's game.

    Because Ohio State scores about 2.2 times as many points as teams usually give up (their DSO), and as Penn State typically gives up about 18.8 points per game, this gives us a prediction of about 42 points for Ohio State's output. Doing similar calculations with all of the other numbers in the above table renders the following prediction:

    Ohio State: 33-42 points
    Penn State: 18-20 points​

    If we stop here, we observe that these numbers comport fairly well with the Vegas predictions (using spread and over/under) of Ohio State -15.5, and total points of 60.5 (a score at the top ends of the predicted ranges - say Ohio State 41, Penn State 20 - would give us both the cover and the over). But what if we could dig deeper into the numbers for that next level of context?

    The obvious next question is, of course: Who Have They Played?! Did you build up your stats against weaklings, or are your best performances against better competition? This is difficult with traditional stats because everyone has better numbers against the Little Sisters of the Poor (I'm looking at you Boise State). Comparing differential numbers achieved versus varying levels of competition gives us a uniquely suitable way of determining whether a team is a bully or a hero, elite or not elite. (James Franklin himself may have provided a spoiler as to where Penn State falls on the elite vs not elite scale).

    Without going into tedious detail, there is a way of determining how well a team holds up to better competition, and of reducing that analysis down to a single number. The following table gives this number, called “rigidity” for the purpose of this discussion, along with their differential numbers from the previous table.

    Team/Unit (Offense/Defense)DSO/DSDRigidity
    Ohio State - Offense2.23574.5
    Penn State - Defense0.661-53.5
    Penn State - Offense1.34469.9
    Ohio State - Defense0.53817.7
    It is always interesting when the results are unexpected. In this case, it may come as a surprise that the Penn State offense is nearly as rigid as the Ohio State offense (69.9 vs 74.5). But being able to maintain your performance against better competition is more impressive when you (Ohio State) are scoring more than 2.23 times as many points as your competition usually gives up, while your opponent (Penn State) has a factor of only 1.34 times (this is another way of expressing DSO).

    Leaving out a bit more of that pesky tedious detail, we now have a means of modifying our prediction. Suffice to say it is similar to a best-fit analysis, but modified so as to work with data that is non-linear and multi-modal. To put it in practical terms, Penn State's offense is more rigid than the Ohio State defense, so the prediction for their point total is modified upward. Ohio State's offense is vastly more rigid than Penn State's abysmally flacid defense, so Ohio State's point total is adjusted upward by quite a bit more. This gives us our penultimate prediction:

    Ohio State 51
    Penn State 21
    The above was called our penultimate prediction, partly because I really like that word and partly because there is one final bit of context to add, one that affects only Ohio State's point total. While by far the most rigid unit on the field is the Buckeye offense, it turns out they are much more rigid in terms of the passing game than in the running game. The same is true of the Penn State defense. As anyone who watched them against Michigan or freshman-quarterbacked Minnesota could tell you, the Nittany Lions are hysterically flacid against the run. The simple truth though is that they are quite rigid against the pass. By being most rigid in the area where Ohio State is most rigid, this makes Penn State a slightly better match up against Ohio State than they might be otherwise, so Ohio State's point total has to be adjusted downward, giving us our final prediction (for entertainment purposes only).

    Ohio State 45
    Penn State 21​
    Last edited by a moderator: Oct 28, 2022


Discussion in 'Buckeye Football' started by DaddyBigBucks, Oct 27, 2022.

    1. muffler dragon
    2. brodybuck21
      this is the stuff iim talking about. give me more
    3. DaddyBigBucks
      A lot of that belongs to LJB. He should be cited as co-author at least.
    4. charlohiottean
      Hear me fellow denizens of our beloved BP! Do not be seduced by the bright shiny words this statistical charlatan slathers across his spreadsheetery like “factored”, “ methodology” or “sharks”! We must remain diligent in our refusal to allow the outlandish claims made by these algorithmic wizards to turn us from the venerable predictolators we’ve lovingly used and relied upon through the ages.






      Are these tried and true systems to be shamefully abandoned in our mad pursuit of the ever elusive fantasyland called statistical surety? May it never be!

      Hold fast, I say, to the old ways. Do not forsake the solid foundations of unchallenged groupthink for the shifting sands of contextualized comparisons of collected data points. Let us remember The Vest and stay true to the silly superstitions that ever sustain this mighty Buckeye Nation on our journey through the lowest valleys and over the highest mountains. Through bounty and famine, Hayes and Cooper, Bauserman and Stroud. Remember what it means to be a Luddite and what it cost those idiots to pave the way we now follow.
      Last edited: Oct 28, 2022
    5. DaddyBigBucks
    6. dragurd
      I think that was a lot of words to use just to have an excuse to say rigid as many times as possible
    7. DaddyBigBucks
      Completely unfair

      It also allowed me to call Penn State flaccid a couple of times
    8. pnuts34
    9. LordJeffBuck
      To set the record straight....

      I am responsible for 0% of the content and maybe 5% of the style. I was strictly an editor, not a co-author. DBB deserves all the credit (and a GPA).
    10. Bestbuck36
      We bow to your excellence!
      LordJeffBuck likes this.
    11. DaddyBigBucks

      Your Differential Writing Quality (DWQ) for that piece is at least a 2.0, so you obviously deserve at least half of the credit.

      QED bitch
    12. shouright
      Now account for what will happen as a function of turnovers.
    13. DaddyBigBucks
      If I could do that AND a half a hundred other things, then I might bet real money on sports.

      I don't.

      I've done this kind of analysis for years, and the most consistent lesson I've learned from it, is that I am beyond blessed that gambling holds no allure for me whatsoever.

      If I were to start gambling, what I would probably do is crank up the old DSA-for-all-of-FBS spreadsheets and then bet on every game where it looks like Vegas has made the biggest mistake, but bet in the OPPOSITE direction of what the spreadsheets tell me. I bet that would be by far the best "system" that I could come up with. It might even make me money.

      I'll stick with engineering.

      The thing that the sharps have that the rest of us don't is NOT a superior numerical model. It's contacts. The guys who actually make money at this consistently are guys who have contacts all over a given conference. So when a quarterback's girlfriend gets flown out to Maui by a professional athlete and the guys who THINK they're sharp find out about it before the game, the REALLY sharp guys knew that he was dating that kind of woman weeks ago and were the reason the line moved BEFORE the almost-sharps found out where the girl was and long before the rest of us were wondering why the kid threw 4 interceptions.

      That's just one overly specific example of a thousand things that can wreak havoc with any numerically-based prediction. If you don't have contacts, stop donating your money to people who do.
    14. charlohiottean

Share This Page