To the important stuff first, and then to the discussion:
1) I can do common benchmark scans as usual now, but I will likely not have the neural nets up and running until next week in all likelihood. I have to learn the new set ups on the alpha version 6.0 of NeuroShell Day Trader Professional. That will take a little while, but I need to do that in order to proceed with the cloud computing project.
2) Even under the best of scenarios I may not be trading forex for at least a couple of months as I see how this new system works out statistically. I should have the old machine up by early next week, but I think it will take until March to get the screens up in a network and have NeuroShell Day Trader Professional take advantage of distributed processing on the network.
Now for the discussion part:
For the last year, I have been falling on my butt with equipment related system problems.
Well, the recovery is closer to completion, and I figured I might as well discuss what happened and what I am doing to meet the goals I discussed earlier this year with regard to trading. The last year or so has been good for trading, but has been a nightmare for systems operation. If I am going to fall on my butt, I might as well fall on it in public, hopefully so others will learn from it and not get frustrated.
As most of you who follow my trading either through StockTwits or VinceRowe.com know, I have had a year of nutty stuff going on with a network that was once pretty efficient, but for which the idea of expanding was probably ill-timed. I also lacked expertise in expanding it.
The reason I expanded originally was to:
1) Split systems between stocks and forex. There is a good reason for that. The stock screening I do is quite intensive and works very well. The biggest problem I have is aggregating and accumulating the signals I generate. I have done it via single computer applications, but speed has limited my ability to generate signals on more that 180 stocks a day. I have at least 8 times that number of models, all sequenced by independent time cycle. That means that no two charts are alike, and for that reason, all charts MUST be separate. As you saw in a previous post, I average about $1.90 in profits for each dollar I lose, and the win percentage is about 72.3%. Over time, that is a nice winning arrangement. Some of the systems I have are over three years old (so they include the sell-off) and they produce $8/$1 lost ratios and are close to 80% correct. That is why I want ALL the signals so that I can screen them further.
2) Separate the research and model building (real time analysis for stocks and futures) from the trading platforms. That is what a product called Chaos Hunter will allow me to do, but it, by itself, is a resource hog. My goal over the next few years is to build an effective 240 minute chart swing model for several currency pairs, or for the currency futures (which do not suffer so much from the cost of spreads as forex pairs do).
The problem is, getting to those signals quickly. Until 2007. NeuroShell Trader did not have the capabilities of taking predictions and throwing them into an output algorithm. It works great with trading system algorithms, but NOT with pattern predictions. I find trading PRICE STRUCTURE, and, from my perspective, trading those pattern predictions are far easier and more profitable than trading some purely indicator based system. That is why I demand that I get those predictions as they arrive. In late 2007, NeuroShell Trader was built to do that. Sadly, my brother died at the same time, and it took me more than a year to straighten out our business circumstances, and it took me the better part of 2009 to fix the tax implications of the business we shared. I simply could not get it all done.
As that wore on, my primary screening computers became outgunned by new software and it caused further problems with time. I used to be able to knock out the Bullish Reversal Report before 9AM most mornings and have my trades in all at the same time. I also had the luxury of 2 PCs engaged in screening, one in doing neural net calculations (if not three on heavy days), and one for writing the report.
At the end of 2009, because of all the other crap (including not being able to power screens to three new computers), I was down to ONE PC. It was tough to fix and I fought with vendors over power requirements, sent numerous pieces of equipment back, and in late October decided to punt until I could find resource help. While that happened, though I back up data religiously, the archive drive went into slow cardiac arrest. Carbonite was of no help, so I decided to use Mozy.com. Mozy.com saved the 80% of the archive that I depend on, right before the drive caved in.
At any rate, I did find a local repair franchise. For those traders in Greenville, N-CTRL, about 2 miles from my house on Butler Road. They have done fantastic things, including rebuilding the old six screen platform PC, which I will use to trade forex. That rebuild saved me about $2K for a new machine.
The new PC will be used to preform end of day and real-time trading analysis on one Forex model I purchased and hopefully three or four I will create to trade currency pairs.
Two things I recommend to all traders who use systems and lots of data:
1) Before you do ANYTHING ELSE, back up your data off site. I think Mozy.com is far better than Carbonite.com simply because a) their customer service responds immediately to problems and they care about getting the right results and b) Mozy.com recognizes more file extensions than does Carbonite. By backing things up, I saved a decade of hard work from an inauspicious death.
2) Find expertise in areas that you are weak. I know how to build trading models, but I have LONG FORGOTTEN HOW TO BUILD EFFICIENT COMPUTER NETWORK SYSTEMS.
I live in an area that has more engineers per capita than any other city in the United States. The problem is that there are tons of people who can help large corporate clients with computer problems and little help for small business. For years, I knew enough to build my own equipment. In the late 90s, I started to lose my edge, but I found some really good entrepreneurs who knew what I needed and could build to my specifications. In the early 2000s, most of those folks went back into corporate consulting and closed their independent shops. Only recently, have I found real help in running systems. Why? Because recession has driven many corporate people into that industry.
Not only am I getting help on systems, but I am also getting programming help with the cloud computing project. I have a couple of local people here who have expertise not only in cloud computing but to one degree or another web database development.
That is basically where I am at the present time. Thanks again for your patience.
