Do any of you guys use news service inputs into your system? I'm currently manually using theflyonthewall.com. It's generally lagged a few minutes from when the true news is released, but it's not bad. Are there any services that are more API friendly and within the price range of a home-brew trader?
I cannot answer with regards API. But I do know about trading news. Quite fascinated by it. Here's one of many private youtube videos I have in my collection - for self-education, the very occasional share, and just for the sake of historical records. http://www.youtube.com/watch?v=zB_SyY5KFaI The point I want to make with this is how quickly the market responds to breaking news - faster than most people realise. The text-based news feed displayed there is the Interactive Broker News Platform, streaming DOW JONES GLOBAL MARKETS NEWS (That is their premium text based service as far as I know, around $250/month, depending who you ask). I tried some other platforms, but my conclusions were that you are depending on 3 actual sources of news, really: Dow Jones (2 stars out of 5) Reuters terminal (5 stars out of 5, but $15k/year?) Bloomberg terminal (5 starts out of 5, but $15k/year?) Dow Jones can be access via third parties such as the Interactive Broker News Platform, but Reuters and Bloomberg breaking news can only be accessed via their respective terminals. For Reuters, that was previously not the case, but is today (I am told). So in conclusion, I think you really need Reuters + Bloomberg to profit. If I had access to a Bloomberg terminal, and had the patience to stare at the breaking news all day long, there is free money to be made if you are ready to click buy/sell in the correct market within 3 seconds. 3 or 4 seconds is all it takes for any of several SQUAWK Boxes to repeat the news, and then every prop trader and his dog is all over it. Dunno if any help to you. Just my thoughts there. Typically downbeat/defeatist/realistic(?) as is the case on trading forums I'm afraid. Goodluck
Not interested in profit. I'm interested in limiting damage. I tend to make money when the market is operating "as usual" and not when there is news blowing things apart. I'd rather have the algo turn off than detect some false signal on account of news.
I hear you. My belief is that unless you can invest several thousand $/year on a tip-top news service (i.e. BBG or Reuters), then whatever news service you are using will activate the "algo kill switch" after the damage is done. If you are baby sitting the algo then maybe you could get a couple of Squawk boxes and manually pull the plug when one or both of the Squawk guys get excited. Just throwing an idea out there: maybe instead of relying on a useless lagging news service, maybe the algo activates the "kill switch" when volumes across multiple markets start to spike. This assumes, of course, you are trading some broad products like Euro/USD, S&P, Oil etc. If it's individual stocks - don't expect that will work at all.
P.S. A good automated solution to this would help me. As things stand, I have just accepted my automated strategy will take a beating occasionally when breaking news occurs during overnight session. Only thing I'm putting in place is to kill the algo when it's lost too much money for me to stomach in one night. Hardly ideal.
The occasional beating can be disastrous, though. For me, a "beating" event can cost 4-5 days (possibly even more) of earnings. It's true the algorithm can detect abnormal volume, but the real issue is when you have a passive element to your strategies and are not HFT. Or, for that matter, if you carry a position almost always have no hedge and just rely on execution. And if you're trading stocks, what abnormal volume is starts to depend on the stock. That makes quantifying abnormal events a little bit trickier, since you now need some sort of reference data or tracking data to figure out whether you're seeing some really abnormal stuff.