Description of Algorithm 1 and historical results. As of December 17, 2007.
The first algorithm has been tested for all NYSE and NASDAQ stocks since Jan, 2007 and has now produced 31 buy signals through December 17, 2007.
I've tried various sell rules, and they seem to be much simpler to do "on paper" than in practice, so until I get better at reading charts and finding good exit points, I will be simplifying things to just shoot for a 5% gain per trade.
Of the 31 buy signals, 30 have gone up at least 5% within 10 trading days. The 31st could have been sold at the end of the 10th day for just under a 5% loss.
If you take 1.05^30*.95 you get over 300% gain. that's obviously unrealistic for a lot of reasons, including that there may be multiple "buy signals" in a short time period, so you'd either have to spread your money out or miss out on some. Also, there's always commission, and the fact that you can't really put all of your money into one stock.
Since I started actually trading these at the beginning of September(albeit with different rules), here are the results (in hindsight) so far:
Date Symbol Open 5% higher When Hit Date Buy? High 5% Higher When Hit Date Buy? Beginning Balance Ending Balance Cumulative Gain
09/12/07 RDN 18.61 19.54 Same Day 09/12/07 Yes 19.67 20.65 4th Day 09/17/07 Yes 5,000.00 5,235.00 4.7%
09/20/07 SHOO 18.25 19.16 3rd Day 09/24/07 Yes 18.88 19.82 Didn't, sold after 5 days for 1.4% 09/28/07 Yes 5,235.00 5,296.82 5.9%
10/22/07 CWTR 7.85 8.24 Same Day 10/22/07 Yes 8.43 8.85 4th Day 10/25/07 Yes 5,296.82 5,546.37 10.9%
11/08/07 PFCB 27.38 28.75 3rd Day 11/12/07 Yes 28.57 30.00 Didn't, sold after 5 days for 0.6% 11/14/07 Yes 5,546.37 5,568.98 11.4%
11/12/07 CCRT 12.70 13.34 3rd Day 11/14/07 No 13.00 13.65 3rd Day 11/14/07 No 5,568.98 5,558.98 11.2%
11/15/07 ARII 15.14 15.90 2nd Day 11/16/07 Yes 15.14 15.90 2nd Day 11/16/07 Yes 5,558.98 5,821.37 16.4%
11/20/07 FUR 5.12 5.38 2nd Day 11/21/07 Yes 5.24 5.50 Didn't, sold after 5 days for -0.7% 11/27/07 Yes 5,821.37 5,771.44 15.4%
11/27/07 TRGL 6.17 6.48 3rd Day 11/29/07 Yes 6.17 6.48 3rd Day 11/29/07 No 5,771.44 5,761.44 15.2%
11/27/07 XRM 5.60 5.88 3rd Day 11/29/07 No 5.60 5.88 3rd Day 11/29/07 No 5,761.44 5,751.44 15.0%
11/28/07 ACW 7.27 7.63 3rd Day 11/30/07 No 7.40 7.77 3rd Day 11/30/07 Yes 5,751.44 6,023.26 20.5%
12/03/07 BECN 9.19 9.65 5th Day 12/07/07 Yes 9.19 9.65 5th Day 12/07/07 Yes 6,023.26 6,308.40 26.2%
12/04/07 CENT 5.81 6.10 4th Day 12/07/07 No 5.85 6.14 5th Day 12/10/07 No 6,308.40 6,298.40 26.0%
12/10/07 ASGN 6.83 7.17 2nd Day 12/11/07 Yes 7.10 7.46 3rd Day 12/12/07 Yes 6,298.40 6,597.02 31.9%
I can't get the image to work, so for now I just put a hard to read pasted version of the spreadsheet.
Date: Obviously this is the purchase date
Symbol: Another tough one here...
Open: Price at the open that day, which is what I assume to buy in at.
5% higher: this is just 1.05 x the open price
When Hit: this is when the 5% gain would have happened, or if it did not, what i would have sold for at the close of the 5th day.
Date: The sell date
Buy: The buy column is simply if I would have bought or not. For simplicity, I decided to just forego any buy signals that occurred when I was already holding a stock. In this example there were 4 out of the 14 buy signals that I would have skipped, simply because I already had a position in another stock.
High: This is the high of the purchase day. I did this to try to bracket the results a bit, since open price is often not an accurate portrayal of what you could "really" get in at.
5% Higher: This is 1.05 x the high
When Hit: Same concept as previously, but based on the high, not open price.
Date: the sell date under this method
Buy: Same as above
Beginning Balance: For tracking purposes, I am starting with $5000
Ending balance: This is the balance after this particular trade. In order to account for the fact that you can't really put exactly $5,000 into a stock, just because it never works out that way, I estimated that you can put 98% of the beginning balance into each stock. In other words, for the first one (RDN), you probably wouldn't get 268.6728 shares, you may get 265 shares at $18.61, or $4931 worth. I then calculate my gain (5% in this case) and then subtract $10 commission ($5 each on the buy and sell) and add back in the part of my beginning balance that I didn't use on that trade. That amount is my ending balance. If I explained it properly, the ending balance on the RDN row would be 5000*.98 = 4900, 4900 * 1.05 = 5145, 5145 - 10 = 5135, 5135 + (5000-4900) = 5235. Hopefully that's not too convoluted...
The main idea with this new setup is that it may not be realistic to think you will be able to get in at the open price, but it's certainly realistic to say that the upper bound on what you'd pay would be no more than the day's high. I think the reality will lie somewhere in between.
So in order to get an approximate return, we use the upper bound of our gain, which would be 1.05^(# of trades under the "open" method), and the lower bound would be the amount suggested by the ending balance (which is calculated based on the "high" calculation vs the "open" calculation.)
So in this case the best case scenario shows a return of 1.05^10-1 = 62.8%, and the "worst case" shows a return of 31.9%.
Either way, not too bad for 2 and a half months.
