HR Geeks

Years ago I thought it would be neat to try to predict something. Not in a psychic fashion per-se, but predict events based on current events, mainly trying to forecast stock prices. I had an interesting idea, sure I am aware that a hojillion or so other methods of modeling the stock market are out there now, and probably one identical to my initial concept. This concept being, trying to use a news service, such as Google News, to predict the stocks. In brief, take the headlines from a news RSS feed and then associate certain words in the headlines (e.g. killing, gun shots, hairy Hobbit feet) to values of certain stocks. Would certain current events, represented by news headlines, actually be a reasonable predictor of stock values? Well, I don’t really know.

So, I started working on an application that would help me do this, like I said, I am sure this is not a unique idea. Anyways, for one reason or another, I really do not recall, I decided to try to forecast the state of the world. Kinda like the Department of Homeland Security’s “Threat Level” thermometer. I care about the state of the world; however, predicting the stocks accurately would be mega-cool in my opinion. So, I came up with Slayercast. Being a fan of one of the greatest bands of all time, I figured, hey… they know what’s up! Let me see if associating Slayer lyrics to the news headlines would give me a “forecast” relating to the state of the world. Like the great Terrance McKenna’s 1970′s derived tool Timewave Zero, but with a nice metal-injection! After all, Slayer does sing about negative subjects, which are often portrayed by media. Why? I suppose we are parasitic beings that enjoy reading about scary-negative stories in the news, well at least the media seems to think that we enjoy reading the like (my assumption). The Slayerizer tool is rather simple, it combs the top ten headlines from Google News, and generates a ratio for each headline, a ratio of Slayer lyrics to non-Slayer lyrics. Thus, a value of 100% would mean that the Slayercast value is 100% for that headline. The ratio of Slayer-to-non-Slayer words is averaged for all ten headlines for a given hour. I cron’d the Slayerizer on users.757.org and captured a Slayercast for about each hour between the dates of August 19, 2009 till some time on November 12, 2009. I say “about” because for some hours, my utility returned early without providing data.

Caveats:

• This tool is not perfect and did not successfully grab all data all of the time.
• Not all hours were captured.
• Articles, like the words ‘a’ ‘an’ ‘the’ were not removed from the lyrics, hey if Slayer uses them, it’s fair game!
• This is not to be some thesis or great scientific endeavor, it’s just a goofy-project!

So what now? I sat on the data for a while, but never really did too much with it. Well, back to the stock concept. I never really linked the Slayercast to world events. But we can’t have data sitting around doing nothing! It must be exercised! Back to the stock world. I decided to see if Slayer knew what was up with the defense industry. I make the assumption that a good view of the defense industry can be represented by a stock exchange-traded fund (ETF). Basically, an ETF is just a bunch of stocks pertaining to a slice of the overall stock market, such as the technology sector, or in this case the defense industry sector. Since Slayer sings about death and stuff, and I hypothesize that the defense industry might provide a data relation, the Slyayercast value might be a reasonable tool to forecast the sector of the stock market pertaining to the defense industry.

The results? Can Slayer lyrics be used to forecast the stock values of the defense industry? I used the PowerShares Aerospace and Defense ETF to associate against the Slayercast value for a given day. I assume this is a reasonable “view” into the performance of the defense industry, from a market perspective. Hey, if the defense industry is rocking, what does that mean? That is a thought experiment for the reader.

With the data in hand I had to relate the Slayercast value against something! How about a stock value! Simply, I took the difference between the opening and closing values for the Slayercast for each day. 12AM Slayercast value subtracted from the 11PM Slayercast value. I did the same for the stock value of that given day, I took the difference between the opening and closing value for the ETF. One discrepancy is that the stock market open/close values are not 12AM to 11PM, but from 9AM to 4PM each day. The correlation between opening Slayercast values and opening stock market values is not terribly direct. Might I add, that some hours/days, when the Slayerizer borked (messed-up) or the market was closed, should not be in this graph. And, even for some headlines the Slayerizer might have messed-up, and that data might be reflected in this graph.

Now, interpret at your own jest:

Shall we sit down and have some story time? So here’s the deal. I have had this fascination about being able to predict future events given the current state of a system. Seemingly random events, I have a hard time believing are truly existent. Of course, I do favor the Heisenberg Uncertainty Principle, but Markov processes can be interesting. The latter being a process that is independent of its previous state. In other-words, you transition from your current state (present) to a future state because of a seemingly random event. In other, other, words, a stochastic process. Interpret the Slayercast values with your own discretion.

Data Sources:

• Slayerize Stuff: http://users.757.org/~enferex/slayerize/
• Slayer lyrics from (I think): Dark Lyrics
• Stock data from: Google Finance for PPA
• News data from: Google News

-Matt (enferex)