Some Blog Site

As I was listening to a pre-game interview on the radio recently, I was interested to hear all the questions coming from the reporter guy.

• So-and-so had a good game last time against this opponent. Talk about him.
• The team is 3 games behind the leader in this division. Talk about the importance of this game.

I’ve written about the issue of sports reporters and the use of “talk about” before.

Since I can’t beat them, however, I will join them. Rather than trying to get reporters to ask real questions, I now suggest replacing them with a robot. Simple natural-language AI that starts each statement with “Talk about” and then adds some keywords relevant to the team and the schedule and the players – that should be indistinguishable from the current situation.

Think of it as a reverse Turing test. If the audio clips are so meaningless/predictable/boring that they could be replaced by a computer and no one would notice, then why hire reporters to ask the questions?

Also, the answers the coaches/managers/players give can be just as bland:

• I thought everyone played well individually today, we just didn’t play as a team
• We had a good game plan, we just didn’t execute well
• I was glad to be able to do my part to help the team get the win today
• We had some trouble early, but everyone pulled together and it showed in the second half

If the AI is good enough, it could replace both sides of the interview and neither reporters nor sports figures would need to be bothered.

The only problem would be the audio itself – getting the computer to sound like the actual person. Maybe have the computer generate the scripts, hire a couple of voice actors and they can read the scripts and produce the interviews for the whole league in one batch.

I’ve also come to view pre-game and post-game interviews as the equivalent of comments sections of news sites: they’re going to have them, but I know they’re a waste of time so I avoid/ignore them.

They were the police, the jury, and the executioner all in one.

They were The Commissioners.

So maybe he’s not the executioner. But he will approve your sentence and then hear your appeal and decide if he agrees with himself or with you.

In the last week, we had the announcement of the 2015 NFL schedule. Now that we know who will play whom and when, we can start predicting wins and losses.

I keep my predictions over at Some Fun Site. View results of previous football seasons.

2014 Summary

Last year, I predicted that

• Cleveland Browns = 4-12
• Detroit Lions = 10-6
• New York Giants = 5-11
• Pittsburgh Steelers = 10-6

How they actually did was

• Cleveland Browns = 7-9
• Detroit Lions = 11-5
• New York Giants = 6-10
• Pittsburgh Steelers = 11-5

Pretty good, except for the Browns.

(more…)

Everyone familiar with timed sports (football, basketball, hockey, etc.) knows that the last minute of the game lasts a lot longer than the first minute of the game.

But how bad is it? And how does it change during the game?

I thought I would put together a chart showing the concept. I don’t have any actual data – I’m just going off my instinct here.

Maybe the chart is skewed toward the worst case, not average. For example, how long – real time – does the last 10 seconds of a close NBA game take? How many plays can occur in the last 15 seconds of an NFL game if a team is trying to rally a win? That is what I was thinking when I chose the Y-axis scale of multiples of game time. If 15 seconds of the game clock takes 2 minutes of my life, that’s a scale of 8x.

One of the more annoying aspects of sports is the delay that is part of the game but shouldn’t be. Example #1: intentional fouling near the end of a basketball game. Example #2: trying to ice the kicker for field goals in a football game. I hope at some point they change the rules to forbid those.

Any recommendations for the Y-axis scale?
Any other proposed changes to the rules to make the ends of games less annoying to the fans?

Okay, okay, it is 2015 at this point, but the results are headlines as 2014 because they match with the 2014 predictions made in 2014 for the 2014 season. Also, the results are not all-haiku, just the predictions were. A more accurate title would be “Results for the All-Haiku Predictions made in 2014”.

Before the bowl games commenced for this past college football season, I made some predictions. Here, for your reading enjoyment, is the tally of those predictions. Note that the results are not in haiku form, in contrast to the predictions.

Results

Here is the list (correct predictions in green, incorrect in red):

Nevada over ULL

Utah State over UTEP

Colorado State over Utah

Air Force over Western Michigan

South Alabama over Bowling Green

Memphis over BYU

Marshall over N. Illinois

San Diego State over Navy

Western Kentucky over Central Michigan

Rice over Fresno State

Louisiana Tech over Illinois

Rutgers over North Carolina

NC State over UCF

Virginia Tech over Cincinnati

Arizona State over Duke

Miami over South Carolina

Boston College over Penn State

USC over Nebraska

West Virginia over Texas A&M

Oklahoma over Clemson

Arkansas over Texas

LSU over Notre Dame

Georgia over Louisville

Stanford over Maryland

TCU over Ole Miss

Boise State over Arizona

Mississippi State over Georgia Tech

Auburn over Wisconsin

Baylor over Michigan State

Missouri over Minnesota

Oregon over Florida State

Alabama over Ohio State

Pittsburgh over Houston

Iowa over Tennessee

Kansas State over UCLA

Washington over Oklahoma State

Florida over East Carolina

Toledo over Arkansas State

Oregon over Alabama

And here are the results of the various forecasting methodologies (see the first year for description of the methodologies) (also, use the word methodologies if you want to sound important; methods would work just as well and is shorter) :

• Some Blog Site picks were 21-18 (similar to last year)
• CBS120 picks were 22-17
• HTW was 22-17 for the official Home Team Wins (HTW)
• HTW was 25-14 for the Geographical Home Team (GHT)
• Isaacson-Tarbell Postulate (ITP) was 22-17 if using HTW
• ITP was 23-16 if using GHT

I won’t analyze the results as much as I did last year, mainly because I had more time and more sleep last year. But it was a good year for all predictors – every method was over 50%. I just need to figure how to better predict outcomes. Especially against the spread.

Because of the disparities between institutions and between conferences, it is tougher to predict bowl games than NFL games. But one thing seems to be pretty consistent: if you don’t want to do a lot of research, just pick the team whose campus is closer to the bowl game.

Thoughts on the season’s results

I bet TCU agrees with me that an 8-team playoff would be an improvement on the current format of 4 teams. But 4 teams is better than the previous format of 2.

Conferences

Since the strength of the conference has something to do with the results, I thought I would tally each conference’s bowl game record for the 2014 (and the first bit of 2015) season.

• AAC: 2-3
• ACC: 4-7
• Big 10: 6-5*
• Big 12: 2-5
• Independent: 2-1
• MAC: 2-3
• MW: 3-4
• PAC12: 6-3*
• SEC: 7-5
• Sun Belt: 1-2
• USA: 4-1

* additional win/loss due to playoff + championship game

So the best conference was Conference USA (they won 83% of their bowl games) and the worst was the Big 12 (at 29%).

Or maybe the Sun Belt is the worst conference because they sent the fewest teams to bowl games again this year.

Perhaps you could say that the SEC was the best because they had 12 teams go to bowls. Or you could say they were just the most popular conference.

Perhaps you could say that the Big Ten was the best because they won the championship. Or maybe the SEC is the best because they won the most bowls (at 7).

Next year: playoffs again! TCU and I will be awaiting the expansion of the format to 8 teams. It might be too early to tell with only one sample year, but 8 seems to be the most reasonable number to whittle down 120 teams to a champion.

It is time once again to update Some Blog Site readers on the results of my Some Fun Site project to create a more accurate football prediction method.

The 2014 NFL season is over, and here are the most accurate methods for predicting regular-season game results (wins-losses):

• MPWLS: 69%
• MPWHFA: 66%
• ITPLS: 66%

MPWHFA is listed first because it was one game better than ITPLS.

This year saw the addition of MPWHFA to the prediction methods. “MPWHFA” stands for More Points Wins with Home Field Advantage and contends that the team that will win is the team that scores more points than its opponents, adding a few points to the home team. It did great, capturing the long-term title among prediction methods. (For the ideas behind the methods, please visit the Some Fun Site page.)

And if you think you have a formula that can predict the winner of an NFL game better than 63.8% of the time, let me know and I’ll add it to the list.

There are those who disagree with the current setup of playoffs for the NFL. If you’re not familiar with the playoff format: the NFL is divided into 2 conferences, and each conference has 4 divisions of 4 teams each. The winner of each division gets into the playoffs, and there are 2 extra spots (wildcard) in each conference for teams with the next best win-loss records.

It sounds like a fairly straightforward arrangement. Division schedules are setup so that there is a clear winner within each division. And if a team is good but happens to be in a division with a slightly better team, it can still get into the playoffs via the wild card spot.

The problem that some people have with this arrangement is this: what if one whole division is good and another whole division is bad?

For example, let’s say the best team in the NFC West has a losing record, at 7-9. But they win the division, so they get into the playoffs. Meanwhile, teams with better records than 7-9, in other divisions, miss the playoffs.

Maybe the NFC West teams had tougher schedules. I think that, either way, those teams will be eliminated along the way.

The alternative promoted by those opposing divisional seeding is to scrap the meaning of divisions and base playoff entrance and seeding solely on overall record.

Going to seeding based only on record doesn’t account for tougher or easier schedules. Plus I don’t like change, so I’m going to say keep the playoff format like it is.

What is the purpose of the playoffs and championship game?

To determine the best team that year?

Ostensibly, yes. In reality, no. It determines who is playing the best at the end of the year. Otherwise, why not just play the season and crown as champion the team with the best record? If there’s a tie, have a playoff.

Why don’t they do that? Because people like having tournaments. And because they like having underdogs who could have a chance.

Those who do want to change it do so because they add to the purpose of the playoffs. They want to find the best team, and they want close, exciting playoff games.

But what if the 4 best teams are in the same division? Would people want to watch them play again after having watched them play each other twice in the regular season already? No. People want to see matchups of different divisions.

Bottom line: If a team didn’t win its division, then it is not the best team in the league. So the current format does fit the purpose of the playoffs.

And Mr. Commissioner, if you’re reading, the one thing you should absolutely not do is expand the playoffs. Right now, it’s about a third of the teams make the playoffs. That’s a good ratio. If you widen the playoffs to include more teams, you’ll approach the mark of half the teams making the playoffs. That’s ridiculous – playoffs should be significant. Handing them out to anyone who is average or better makes them less relevant.

Plus it would mess up all those “they haven’t made the playoffs since …” stats.