Useless analysis from uselessanalysis.com
Over 10 years ago, I started a website, uselessanalysis.com.
It served as a way for me and a handful of friends to answer questions that we’d always wanted to answer, but that no one else had ever bothered to. Questions that didn’t really matter, at least not in a mass appeal or business value kind of way, but questions that absolutely did matter once you started thinking about them.
Most of the content I contributed to the site was sports-related, but I ventured out into music, video games, TV shows, and even politics over a run that spanned from March 2013 through October 2018.
That site was where I first started running my NFL and NBA power rankings. It’s where I first started picking my video game of the year. And it’s a source of a handful of posts that I republished on this site.
It was fun. I learned a ton. And oh, my, did I do some truly useless analysis. And while I’ve already ported most of my favorite content from that site over to this one over the past few years, I thought it’d be fun today to call back to my favorite pieces that were lost when I shut that site down.
Without further ado, I present you the five most useless analyses from uselessanalysis.com.
5. Correctly predicting Russia’s gold medal count in the 2014 Winter Olympics (February 2014)
Even though I don’t really have any interest in the Olympics, and even though I have literally no experience in analyzing data from the Olympics, I decided to get into the prognostication game for the 2014 Sochi games.
I used the following highly scientific process:
I found the % of total gold medals that Russia won in prior Winter Olympics in the past five events: 18% (1994), 13% (1998), 6% (2002), 10% (2006), and 3% (2010)
I conservatively estimated a 2014 forecast to be around 5% of total golds, assuming that 2010 was a uniquely “down year” but accounting for a general downward trend
I then noted that in 2002 and 2010, Russia had to travel halfway around the world to Salt Lake City and Vancouver, respectively, so I pulled the forecast up to 10% of total golds to adjust out a travel penalty for Russia’s poor performance in 2002 and 2010
Next, I looked at prior hosts to understand the typical lift for a host country compared to the prior and following Olympics, and I found that it varied between a 1% and 8% lift in % of golds won
I took up Russia’s forecast by an additional conservative 3%, finalizing my forecast at 13% of all gold medals taken by Russia
There were 98 events scheduled in 2014, so 13% of 98 events = 13 gold medals
I nailed it. Wow. I can’t believe that worked.
4. Analyzing how much Apu from The Simpsons overcharges at the Kwik-E-Mart (April 2014)
This all came from the episode Homer and Apu in which Apu has this exchange with his customers:
Man 1: I need one 29-cent stamp.
Apu: That's $1.85.
Man 2: I'll have $2.00 worth of gas, please.
Apu: $4.20.
Martin: How much is your penny candy?
Apu: Surprisingly expensive!
So how much is it, really? They never told us!
Well, this one’s easy and we can just apply a linear regression analysis to project how much Martin would need to pay for penny candy.
Including rounding, the answer is $1.47. Pretty pricey.
So Apu’s charging much, much more than any credit card company does per transaction. And that’s not even the biggest problem with Apu, amirite?
3. Analyzing whether there really is a “Madden curse” in the NFL (August 2013)
A decade or so, there was a lot of intense talk about the “Madden curse”.
The concept is similar to the Sports Illustrated cover jinx: a player who dons the cover of the annual Madden NFL video game is likely to have a down season, and thus, it’s bad for him and his teammates to be featured.
Since a player will only be featured on the cover if they have a great season, many fans chalked this up to simple “regression to the mean”. It’s tough to top a great season, even if you’re a great player.
I wanted to dig deeper, so I did. I looked at all 14 of the Madden NFL games that had released to date and I looked at how each cover athlete’s team performed in the year that their star was on the cover.
Only two of these teams won a playoff game - the 2006 Seahawks and the 2009 Cardinals - and only the 2007 Titans improved their performance in the following year. So at the time, you could definitely argue that there was a Madden curse - and it not only impacted the player, but his whole team as well!
Patrick Mahomes ultimately broke the “curse” after winning the Super Bowl in February 2020, and no one talks about the Madden curse anymore. Oh well, fun times.
2. A team of Michael Jordans takes on the Dream Team (December 2013)
I had just gotten a brand new PS4 and NBA 2K14, and I needed to see what this new game and system was capable. I thought I’d test a few wacky scenarios in the game, and one of them was:
What would happen if Isiah Thomas were put on the Dream Team?
Well, we know what would happen. Michael Jordan wouldn’t have joined. But I took it a step further:
I don't think that's where the story ends. You see, Michael Jordan would be so angry that he'd make it his mission to prove that the Dream Team made a mistake in choosing Isiah over him.
Since he has a competitive disorder and a penchant for making the impossible possible, I submit that Jordan would find a way to clone himself in order to field a full team of Jordans to enter the 1992 Olympics (for his home country, Nike) and beat the Dream Team.
It’s a stupid premise, but it was actually pretty easy to set up. That version of NBA 2K (and every version that’s released since) has historical teams that you can modify, so I was able to reasonably recreate both the Dream Team - sans Charles Barkley - and a team of nearly all-Jordans.
I then watched these teams play a full game, and Team Jordan won it. It wasn’t quite Fumble Dimension level craziness, but it sure was useless.
1. I watch Street Fighter II play itself to determine the best CPU-controlled fighter (January 2016)
Yeah, this one is going to be hard to top.
As a kid, my dad would hide my Super Nintendo controllers if I played too much. Which I did, always.
For a while, we’d play the game of cat and mouse. I’d find the controllers and sneak in a gaming session. He’d find out and would hide them somewhere else. Eventually, the hiding spots got too good, and I was at a loss. I couldn’t, like, not play Super Nintendo, right?
So I’d watch the games in demo mode. And a lot of the time, the demo that I’d watch was Street Fighter II. Look, it was better than nothing, okay?!
After watching enough of this back in the day, I became fixated on a simple question, “who is the computer best with?”
Decades later, I was going to find out. I decided to watch the computer play itself 60 times, and then I’d run analysis to see which character performed the best.
It wasn’t a perfect setup. I couldn’t tell the game which matchups to show, so it was up to chance. Some fighters would be featured more than others, and I couldn’t see every possible matchup play out. Sometimes the game would put a character against themselves, giving me no information at all. Beyond that, it was also hard to tell at times who was winning before the demo timer ran out, so I had to allow for ties.
Still, it was enough. I tracked the results in a spreadsheet. I ran an optimization algorithm that accounted for strength of schedule. And I concluded that Guile was the best Street Fighter II character, at least when controlled by the computer.
I don’t know that I’ll ever do anything this useless again. Definitely the highlight of uselessanalysis.com.
That said, this still isn’t the nerdiest thing I’ve ever done. More on that some other time.