Semantic news analysis is how I develop more than half my trading models. I'll have one posted, and running in real-time, within one month. I've got a webpage dedicated to the people and organizations doing this work at: www.marketpsych.com/news_language_analysis.php
Very interesting thoughts about news analysis... How are you dealing with the fact that the majority of all stock related news (i.e. earnings) are disseminated before or after the market close? You won't be able to trade these news.
Have you ever heard of the Futures market? Last time I checked, ES and NQ futures trade like 23 1/4 hours a day on GLOBEX. Check it out, it's almost 4:20PM and this is my 420th Post! Do you what that means?
Sure I heard of the Futures market. I wonder if anyone did successfully combine news algorithm trading and the equities market.
In terms of news algo: I'm not an expert but it's been around for a while. Because most of the Reuters and Bloomberg news txts have a simple txt template for news, I tried developing a system out of it. Obviously, the market reacts to a similar news differently. The impact of the news differs based on the underlying market condition and it became a big task at hand to figure it out. I'm not an expert in liguistics or computer science (and smart) so I didn't get into the rocket science stuff but I did find a few edges while researching and brainstorming to make the news system work. My conclusion became that the newsfeed itself was a form of a trigger with edge, and the significance lies with the setup part. So I took the news trigger out and put in another trigger that worked better. Though, it was interesting how a series of news affects the market gradually. Also, as a hypothesis, how it correlates with changing market condition. From my research, solely, the sematics of the news itself don't matter so much, it's more about quantifying(maybe symoblize is a better word) the news and deriving a "bias" from the series. Again, I'm not an expert or academic. Hopefully, someone else can shed some light to the whole thing.
Well, here's a link for an explanation of AIS for the rest of us: http://en.wikipedia.org/wiki/Artificial_Immune_System Well, it's another entry in my "To-Do List" So much to do... so little time...
I'd skip all the parsing and AI rigamarole and simply jump straight to the equation; "setup part (news) + another trigger" = works better -jmho Rigamarole - a long and complicated and confusing procedure; "all that academic rigmarole was a waste of time" worldreference.com
Nope. That's not it. I'm sure that if you research about static newsfeed, you realize something similar to what I found out. But in my opinion, don't listen or ask me advice. I would wait for acrary's post on this matter. Again... I am not an expert on news algos...