Market swings for composites during the day are REGULARLY 4%-5%. You can swing trade the S&P 500!!! A lot of the moves, especially since last week, have been on average to low volume.. ~ 85% of the average. I wonder if it is the investment houses who are moving the market like this, or is it the joe trader? Or has the joe trader cashed out his/her IRA and is trading with this money to gain back what was lost?
More later.
Not many updates recently. You'll see that most of the blogs focusing on day trading have stopped posting. Most likely it is because everyone is losing money and that's keeping us pretty quiet.
Working on "the model" has been very profitable for me. That's because I'm still spending more time on the code "in the harbor" than out in the choppy market waters fishing for good trades. I can be thankful that I haven't lost money yet, although I can't say that I've made any money either. Yet, with a market like this, anyone fully in cash can say they beat the market 50% this year... Crazy.
More later,
The Man.
So the market is reaching a fever pitch, and investors are starting to pull out in major ways. Here's where the leverage and the low cash reserves of the mutual funds will thoroughly decimate the stock market. Investors pull out of the mutual funds. Mutual funds, without sufficient cash reserves, will have to pull out as well, against their professional opinion. Funny how we invest in mutual funds and take the 1% to 3% hit in fees for precisely this advice. Unfortunately, this advice cannot be taken when everyone tries to pull out. The good news out of this is that the mutual fund companies will have a lot of cash on reserves again once the market turns up. This will give them more control over their investments, and ultimately a better chance to beat it.
Markets down 450 pts. Markets up 250 pts. Markets down 450 pts. And remember, computer trading stops after 200 pts. People tried very hard to get down to this value. If computers ran the market (like in '87), the market would have already crashed 4 to 5 times already.
This is probably the beginning of "The Big One," which was prophesied by many a bear market proponent since 2000.
On the local side, The Algorithm is serving it's purpose:
Giving me lots to do and keeping me from trading all of this time! Cash is King right now, and the time to invest is sooner than we think.
Speaking of The Algorithm, lots of improvements have occured. The most important improvement is the development of an automated stock scanner that looks for stocks that fit the model well. Already a few stocks have been identified that work better than the ones that the model was developed on, and it is hoped that the datamining operation will yield great fruit. Now we must focus on whether or not macro-scale events (like a stock market collapse) will affect the predictability of the stocks using this model. Most likely the answer is as sure as a microsoft jiggle.
More later...
No updates from me or The Machine since June! Part of the problem was my ISP's mail server blocked The Machine's postings, and the other problem was I have been working for two months overhauling The Machine's code. The first half was beneficial, as I added a robust expert-based neural net decision logic.
The other half was not so useful. Tried to incorporate volatility back into the code, but it onlyserved to weaken the overall solution across all of my vectors. The only good thing was I was so engrossed in this code that I didn't waste my time or money trading. Which could be argued as being the whole point of this exercise.
The best I can do is to reduce the loss on a stock by 50% (buy and hold is ~ 60% loss, and The Machine is ~30% loss over the course of a year), Of course, if you know the stock is going to plumet, then just "sell in May of 2007 and stay away."
Hindsight is 20/20, unless you forget to put on your spectacles in the morning.
More later.
No updates from me or The Machine since June! Part of the problem was my ISP's mail server blocked The Machine's postings, and the other problem was I have been working for two months overhauling The Machine's code. The first half was beneficial, as I added a robust expert-based neural net decision logic.
The other half was not so useful. Tried to incorporate volatility back into the code, but it onlyserved to weaken the overall solution across all of my vectors. The only good thing was I was so engrossed in this code that I didn't waste my time or money trading. Which could be argued as being the whole point of this exercise.
The best I can do is to reduce the loss on a stock by 50% (buy and hold is ~ 60% loss, and The Machine is ~30% loss over the course of a year), Of course, if you know the stock is going to plumet, then just "sell in May of 2007 and stay away."
Hindsight is 20/20, unless you forget to put on your spectacles in the morning.
More later.
Just read my last post. What a bunch of big words stung together just to say: "The model is broken, and I'm almost finished fixing it."
Well, I do feel it is almost done. I now no longer need very much computing power anymore. With psyco optimizations, it takes only about 30 megs of memory to run, and I've downgraded The Machine to an old PII.
To do, the regular postings is down due to a problem with the smtp server.
More later.
Still working on the code. It looks like there are still problems with returning inconsistent answers. Without consistent answers, there is no hope to optimize the algorithm for the best answer.
Anyway, I've narrowed it down to the issue of trying to use stochastic, analog problem solving method on a problem with a very digital output (buy, sell, wait).
In order to solve this problem, I have to rig the code by judiciously invoking some rounding and also by setting up a predetermined random seed for the algorithm. Otherwise, the computed solution will always be just slightly different, which it turns out is enough to give inconsistent answers, and impair the optimization runs, which require definite answers.
Alas.. The problem of using genetic algorithms. It gives "pretty good answers mos of the time", but not absoute answers all of the time!
More later.
