Hi, I'm posting on this forum because I've developed a swing trading system that actually works and I can show consistent results (at least 25% profit per year) for the last 14 years. I know this sounds too good to be true but I can prove it and can show automatically generated spreadsheets covering the last 14 years. I'm certain that I don't have any bugs in my code and I haven't optimised my system on the data I am testing. I am a very experienced programmer and have good expertise in statistical modelling. I'm interested in improving this system even more and hearing input from others. My system can automatically email stock candidates that it identifies so if anyone would like to receive these for free then please drop me an email. The system I have developed has taken about 5 years of trial and error to develop and is actually relatively simple in the way it works. The one strange and counter intuitive thing that I have discovered, which I can back up with solid evidence, is that the system always makes a bigger profit if I use a time duration stoploss instead of a threshold stoploss. So for example, if I buy a stock on day 1, I would expect it to reach its target price within 25 days. If it doesn't reach this target within 25 days then I sell anyway regardless. This strategy always gives me better profit that setting a lower limit threshold where I sell if the stock falls below a certain value. What are your views on this? It makes me feel uncomfortable but 14 years of test data backs it up! I've also tested lots of technical techniques from the many books I have on the subject and I can confirm most of them don't work over an extended period. This includes candle patterns, moving averages, stochastics etc. Again, I can back this up with solid evidence. I look forward to your comments and hopefully an interesting discussion! Marek.
What do you want us to discuss? You didn't really give us a whole lot of detail and didn't really post any questions for discussion...
Seems like a waste of an automated trading system to arbituarily just close a position after a certain period of time. Certainly closing a position after 25 days is not going to make the trade more profitable than holding it for 30 days or 35 days, unless the stock just happened to go down after 25 days. If you said you close non-performers after a certain period of time, so that you can reinvest the funds into something that is moving, I could go along with that. I know when I used to daytrade, I would jump into a position when the momentum was up and the second it stopped going up (within 3 minutes) I would get out. But the idea of opening a position and sitting on it for a month and then arbituarily closing it doesn't make sense to me. Ed
Ed / Corey Thanks for your replies. I'll give more detail. The main point of this posting was to invite people to try this system for themselves and discuss the results. Just drop me an email and I'll start sending you the stock picks. The system I've got gives me a target sell price. On average (but not always) this target is hit within 10 days at which point I submit a trailstoploss. I have also found that statistically over many years data that if the target is not hit within 25 days then its best to sell. Sure you could make this time period 20 or 30 days but after testing over several years of data I've found this to be a good time period. So this is not an arbitrary rule its a rule derived from the statistics of lots of data. The system looks for J-Hook patterns, i.e. where the stock has climed steeply, pulled back and is starting to move up again. This may seem very simple but I've found it to be very powerful. I define the jhook shape based on the 4 day moving average and I find the tuning poins of the peaks and troughs and make sure the gradients between the peaks and troughs are steep enough, again based on statistical evidence. The buy point is when the stock just starts to turn and the sell target is when the stock hits the previous peak. This is a modest target but it hits it in roughly 75% of cases within 25 days. Marek