08 September, 2011

Where’ve You Been, Chris?

Some of you have noticed that I’ve virtually disappeared from blogging and Twitter over the last few weeks. In fact, I’ve been asked about it more than once.

So, here’s an explanation.

Other than my family and my work, there are 5 passions in my life. In no particular order, they are: 1) sports (especially basketball and football), computers, mathematics, politics, and reading. Despite my vacation in July, I am still quite a bit behind in my reading, so I’ve been doing a bit more of that lately. That’s the first and simplest reason for my silence.

However, that’s not the biggest reason. I’ve been moderately interested in sports ratings, such as the Sagarin ratings for years. In fact, I wrote a very simple one myself for NFL football back when I was about 16. Research into these things has been an on again off again hobby of mine since then. About 4 years ago, I actually got serious about it. I wrote my own rating system for college football, men’s college basketball and the NFL. It’s actually a pretty good system and I use it to identify which teams are better or worse than their records make them appear to be. I can use it to project winners and losers as well. And this combines three of my passions: sports, computers, and mathematics.

It’s a pretty sophisticated system, and I’ve added to it every year. Unfortunately, the foundations for it are not strong. My first attempt used an Excel spreadsheet. That version quickly morphed into a little Windows app with an Access database. Sadly, I had to use a different Access database for every sport, and adding additional sports (other than probably women’s college basketball) would have been difficult. The Access database still shows its Excel roots, and has problems because of it. The app requires constant babysitting, especially during basketball season, just to make sure that it’s not missing games or including the same games twice, or has the wrong home teams, etc.

In other words, the app’s shaky foundations made future enhancements increasingly more difficult. Also, from the very beginning I’ve wanted a better way to get these ratings and projections published, and the web seemed the best way to do that. But turning this into a web based application, with it’s shaky foundation and all, would have likely made for an unpleasant and unreliable user experience.

Therefore, it’s been a dream of mine for the last couple of years to rewrite the app from the ground up. And that’s what has consumed quite a bit of my time over the last few weeks (actually, over the last few months, but much more since mid-August). I needed to create a much more robust database (using SQL Server this time) that could handle multiple sports, and even sporting events with more than two opponents. I also had it in the back of my mind that I could make it do some political projections. In fact, that’s how Nate Silver of FiveThirtyEight.com got his start. He did baseball ratings and expanded into politics. This would add one more passion to the list as well. Smile

Well, the good news is all of that work for sports is essentially done. There’s still work that needs to go into the website, but the app itself is working and I ran it on college football this week. Most of the calculations require two games of data to produce any results, and really need five or six weeks of data to produce reliable ones, so I have a little time left to finish up the website. It’ll be called www.cjbratings.com. Don’t go there now. I own the domain, but there’s no site yet (I have one on my local PC to play with, but it isn’t ready for prime time).

The better news is that yesterday I had an epiphany on the political side and I now believe I can produce political ratings as well. There will be another website devoted just to that. I haven’t decided on a name yet. But I’ll be able to produce projections on House and Senate control, Governors, and electoral college projections. In time I’ll even be able to do some local stuff, state legislatures and what not. The first things that will appear will likely be the GOP primaries next year. I should have it online with at least the minimal functionality necessary for that by around the turn of the year.

I’ll reference all of that on this blog, but the calculated data will be on its own website. One of the things that Nate Silver does that has always ticked me off is that he combines political commentary with numerical analysis. Yes, I do plenty of political commentary myself, but I go to his site for the numerical analysis, not the commentary (he’s done less of this since he joined the New York Times). It has always irked me having to wade through all of that to get to the numbers. I’d like people to feel comfortable looking at my numbers whether or not they subscribe to similar political beliefs to mine.

There’s a little more effort involved here to get the websites up and running, but almost all of the heavy lifting is done. I’ll be more vocal again in the near future. Probably starting next week. I have my annual 9/11 blogging coming up on Sunday. It’s still not finished, so I need to do some more work on that tonight and tomorrow night.

No comments:

Post a Comment