Saturday, February 24, 2018


Historically, Curling hasn't really been the USA's best sport. The men won a bronze in 2006, but otherwise it's been bupkus. More recently, USA #Curling finished last in the 2010 olympics and next to last in 2014.

But then came 2018. On a magical night in Pyeongchang, four american heroes curled for gold, and cemented themselves in the hearts and minds of America forever.

After a 2-4 start, USA curling was a longshot (2%) to even make the playoffs, and and even longer shot (>200:1) to to win the gold. They needed 4 consecutive wins just to get to the gold medal game! They had to beat 1) 3 time reigning gold medalists Canada, 2) Switzerland, 3) Scotland (where curling was invented) and then 4) Canada again just to get to the gold medal match. Their second win over Canada, a second miracle on ice, came 38 years to the day after the first miracle on ice.

The gold medal match was a tense affair. The teams traded leads all game. But then this happened. Go USA! 

Friday, December 29, 2017

Ohio State's Loss to Iowa was Disqualifying

TL;DR: The best college football team in the country does not lose to an unranked team by 31 points.

I was inspired to do this analysis by an article in the economist, about events of signature significance. The idea is that some events are so rare, they carry a substantial amount of information or predictive power, even with a sample size one.

On November 4th of this year, Ohio State travelled to Iowa City, Iowa, and got blown out by the Iowa Hawkeyes. They lost by 31 points (55-24) to a team that wasn't ranked the entire year.

This loss was an event of signature significance.

  • Iowa was unranked when they beat Ohio State
  • Iowa ended the season unranked
  • Iowa beat Ohio State by 31

Of course there is data. Let's start with past national champions.

Past National Champions
I looked at College Football consensus national champions going back to 1978. I think this is overly inclusive. The game of football has changed so much in the last 40 years, that even games from the 1990s look to me as if they're in slow motion. But, I wanted to air on the side of inclusion. I found that no national champion had a loss this bad. No national champion had a loss close to this bad. Let's take a look.

To start, 25 National Champions went undefeated; count them safely in the bucket of "did not have a loss as bad as Ohio State."

Moving on, 14 National Champions had one loss, but it was to an opponent who was either ranked on the day of the game, ranked at the end of the year, or both.

Here is a list of all such losses. For example, in 1978, Alabama won a national title, but lost to USC, who was ranked #7 on game day and finished the year at #2. Bama lost that game by 10. Most of these opponents were also highly ranked, none of these losses were by more than 31, and only a couple of these losses were close. Let's pause on that point.

No national champion (since 1978) has lost a game by 31 or more, to any caliber opponent.

Our last group of 1-loss title winners had a loss to a team who was un-ranked (both on game day and at the end of the year), but none of these losses were nearly as bad as Ohio State's. 

No national champion has lost to an un-ranked team by more than 14.

Ironically, the worst lost here is Ohio State's from 3 years ago.

All Ranked Teams
I don't think it's harsh to compare Ohio State to past national champions. The CFP for teams who could be national champions! But in the interest of being fair and thorough, I expanded my analysis to include all ranked teams.

Since 1978, ranked teams have played un-ranked teams 8,186 times. Out of all those games, the un-ranked team has won by 31 or more just 26 times, a rate of 0.32%. It's only happened 6 times in the last 10 years. Here's the complete list of those games. None of these teams finished higher than 14th.

In the last 35 years (since 1982), no team with a loss as bad as Ohio State's has finished the season better than 14th.

Going back more than 35 years doesn't change the story much. Even going back to 1978, only two teams with a loss this bad ended up in the top 10. Texas (who finished #2) and Washington (who finished #10) both did it 36 years ago in 1981.

I don't mean to say Ohio State isn't a good team (they are). I'm not saying that Ohio State won't win tonight (I have them favored). I'm just saying they aren't the best team in the country.

The best team in the country doesn't lose to unranked Iowa by 31, not even once.

Saturday, November 25, 2017

Week 13 College Football

Apologies for the late update. Short work week + thanksgiving + family etc.

I won't be able to update the sidebar content this week, but here's the schedule / watchability.

Wednesday, November 15, 2017

College Football Week 12

May not have the splash of last week, but every week with college football to watch is a good week.

Dashboards are updated.

Monday, November 6, 2017

College Football Week 11

Well that certainly shook things up a bit! Losses by Penn State, Ohio State, Oklahoma State, and Virginia Tech knocked those teams effectively out of the CFP.

Dashboards have been updated.

  • Georgia @ Aubutn
    • The model has this game as 50/50! (FWIW, oddsmakers rate UGA as just a 2.5 point favorite, so the model and oddsmakers are fairly close on this one)
    • Interesting that Georgia is 85% to make the CFP if they win here. While a win gives them the inside track to the SEC title game, I only have them as 31% to beat Bama on a neutral field. That means the CFP model is giving serious credit to the possibility that Georgia wins out, loses to Bama in the SEC Title Game, and STILL makes the CFP. That really doesn't feel that wrong to me.
  • ND @ Miami (Fl)
    • Nearly a CFP Elimination game for both teams. Note that Miami is undefeated, and ND has a loss, but their CFP chances are similar. I like that. Miami's schedule has been fairly weak and they've tried to lose most of their games. Notre Dame has been rolling, except for a 1 point loss to Georgia. Their resumes feel similar to me.
  • TCU @ OU
    • Effectively an elimination game. This makes sense, both teams have 1 loss and a second hurts their resume and their chances of making the big-12 title game
  • OkState @ Iowa State
    • Just marvel at Iowa State making top 10 watchability week after week.
  • Washington @ Stanford
    • Since this game is on a Friday, maybe someone will actually watch Washington play.
  • Iowa @ Wisconsin
    • Here's to Iowa just rolling through and causing CHAOS. 
  • WVU @ K State
  • NC State @ BC
  • Michigan State @ Ohio State
    • What is there to say about these games?

Friday, November 3, 2017

College Football Week 10

I apologize for the delay this week, I've been travelling for work. Luckily I have many hours at the airport while my flight is delayed (thanks Seattle snow!) to get everything updated.

  • OU vs. OK State
    • What a good looking game. With one loss each team is playing to keep their CFP (and likely big 12 title hopes) alive. The model thinks highly of both teams and has them super evenly matched. 
    • Ha, I checked to see if this was the highest rated game of the year (it's 3rd) and saw that Bama-FSU was rated 95 watchability back when we all expected FSU to be good
  • VTech @ Miami
    • On one hand I do understand why the committee ranked miami so low despite their undefeated record. Miami has really put up a good effort to lose every game. On the other hand, I'm not sure if there are actual scenarios where undefeated Miami (Fl) gets left out of the CFP.
  • Texas - TCU
    • TCU's CFP hopes hanging by a thread after last week's humiliating loss to juggernaut Iowa State
  • Iowa State - WVU
    • This might be a spot where the committee gets unpredictable. Imagine there's some chaos and a few losses and Iowa State wins out and wins the big-12 title game. Obviously with 2 meh losses the model doesn't like their chance, but I dunno...
  • Clemson - NC State
    • I still can't get past 4. Clemson. C'mon committee
  • Auburn/TAM, KState/Ttech,, Stanford/WSU
    • Just fun looking games despite the total lack of CFP implications
  • Penn State - Michigan State & Wisconsin Indiana
    • I feel like like we're getting one B1G team in the CFP, and that will be the B1G winner. Wisconsin basically has the inside track, Penn State needs help. My model feels differently about Penn State though. With a loss to Ohio State they need big help to get into the B1G title game, but their loss was such a quality loss (1 point, on the road, against an elite team), the CFP model thinks they should hardly be punsiehd at all. If Penn State wins out I see their CFP prospects as likely being the subject of heated debate.
    • Editorial comment: Joe knew.

Monday, October 23, 2017

College Football Week 9

Lots of great games this week. Dashboards updated, schedule below.

  • Penn State @ Ohio State  
    • Huge CFP implications. A win by Penn State would put them in excellent shape. After this week their only semi-tough game is a road trip to Michigan State, so a win by Penn State tomorrow puts them in the driver's seat in the B1G and the CFP. On the other hand, Ohio State would jump right back into contention with a win. 
  • OkState @ WVU 
    • OkState needs a win to stay in serious CFP contention
  • NC State @ Notre Dame
    • Both NC State (what??) and Notre Dame need a win to have a shot at the CFP. Just pause and imagine the world where NC State beats Notre Dame then Clemson
  • TCU @ Iowa State
    • As one of the undefeateds, TCU has wiggle room should they lose. But with games the following weeks against Texas and OU, TCU really wants to keep their wiggle room
    • Iowa State won't be playing in the CFP, but good on them for having a good year!
  • Georgia Tech @ Clemson
    • Clemson is also in win or say bye to the CFP mode
  • Florida @ Georgia
    • Georgia's modeled chance to make the CFP exceeds their modeled chance to win the SEC. So there's some chance we get two SEC teams. That makes sense, but it's interesting nonetheles that Georgia is more likely to make the CFP than win their conference. Thanks Bama