posted
Some of you may remember a thread that I had started about a year ago where I posted results from a program that I wrote to pull Form-4 data from the internet. At the time, that was the entirety of the process. I pulled the data and posted a list of stocks that had a lot of recent Form-4 action.
The thread and program kind of fell by the wayside when the site that I had used to get the data went to a subscription basis, and my code stopped working.
Well, over the last few weeks I have written a new program that imports Form-4 data from the internet. A Form-4 filing is required when an insider (i.e., CEO, CFO, holder of 10% or more of shares, etc.) buys or sells shares. In addition, I wrote a program to pull historical data, such as price and volume. Using the two programs, I was able to set up a series of analytical steps using technical analysis of historical data in conjunction with the Form-4 data, and create a stock screening process that I believe has the potential to provide excellent returns.
Backtesting this theory on data from 2007 through September, my program generated 19 “buy signals”. Looking at the charts after the buy signals occurred, it was obvious to me that this method held great potential.
The next step in my analysis will be to test different sell rules to maximize returns. However, at this point, in order to generate preliminary results, I set a simple 5% trailing stoploss (based on daily closing prices). Plotting these sales on graphs showed that a lot of potential gains are left on the table, and that even higher returns should be attainable with more sophisticated sell rules.
Given that temporary limitation, these results are subject to upward modification in the near future.
As of now, 15 of the 19 “buys” generated so far this year would have sold. Those 15 buys resulted in the following gains or losses: 10.4%, 16.9%, 3.3%, 4.8%, 12.3%, 7.6, (1.0%), (4.2%), 6.7%, (2.8%), (2.3%), 8.5%, 3.4%, (6.9%), 25.7%.
While the average return of 5.5% may not seem especially impressive, the stocks were held for an average of just 20 calendar days (or about 15 trading days).
In addition to the 15 previously described, there were 4 buy signals that have not yet triggered the “sell rule” (i.e., there has yet to be a day where the close was 5% lower than the highest close since purchase). Those 4 stocks seem to be slow, consistent gainers, with current returns of 33.3%, 7.4%, 10.4%, and 56.8% (for an average return of 27.0% and an average current holding time of 49 calendar days).
Combining these two groups, my analysis shows an average return of 10% per trade, with an average holding time of 26 calendar days.
Note that the 10% monthly return can not be directly extrapolated to a 213% annual gain (1.10^12), due to the irregularity with which these instances occur, and the fact that one should not put 100% of holdings into any individual stock.
In order to estimate the type of return that could be accomplished over an extended period of time, I organized the “buy signals” in date order, and tested two different methodologies for allocation of funds. The first methodology is to simply put 10% of your initial holdings (i.e., $10,000 of an initial $100,000 account balance) into each instance of a “buy signal”.
This method produced a 21.3% return for the 9 months of 2007.
The second method was a more aggressive approach, putting a larger amount into each, with the risk of not having any available capital if a string of “buy signals” occur in a short period of time.
This method did prove to be too aggressive to take advantage of all “buy signals”. However, even though no funds were available to purchase the security that went up over 50%, this method still returned approximately 32% for the 9 months of 2007.
In analyzing the data, it is apparent to me that the month of August (with 11 “buy signals”) will prove to be an outlier, versus the normal 0-3 buy signals in a month. My supposition is that the drastic market decline in early August provided incentive for an inordinate number of insider purchases, which will most likely not happen again in the near future. Excluding August “buy signals” from the data still returned approximately 10% per trade, so while the number of signals was abnormal, the expected return was not compromised by this anomalous month.
In summary, my method produced a range of results from about 21-32% in the 9 months of 2007, versus about 9% for the S&P500, and 12% for the DJIA.
This number will increase, possibly significantly, when a more appropriate “sell rule” is established, and going forward, the likelihood of missing significant gains should be limited.
Out of courtesy, please limit discussion in this thread to the idea itself, or discussion of any stocks that have recently generated "buy signals".
Obviously I can't divulge the methodology, but note that the process is done on NYSE and NASDAQ stocks, so you can be comfortable that there will be no "frontloading" or "pump and dump" risk associated here.
Please let me know if you have any questions or comments, and hopefully this will turn out to be a useful thread for us to all make some consistent gains on some "real" stocks. I know that would be a big change for me, but something I look forward to.
-------------------- I'm back! Posts: 5431 | From: West Chester, PA | Registered: Jan 2006
| IP: Logged |
IT IS A LOT....Depends on how much you are playing with....
In pennies we don't want to risk alot so if we use $500 5% wouldn't be a lot but if you are doing this on Big Board stocks than it would be safe to put $5,000 in wach one, 5% would be $250...& $250 in 2 weeks is great I think!! You just keep letting it snowball & WOW!!!
That's what I so now Swing Trading BB stocks...on average I keep them 2 weeks & make on average about 5-7%......the past 3 weeks I have made $1200....so 5% adds up!!!
I can't wait to do your program....
I'll be watching & participating on this thread...
Good Luck Buddy...
-------------------- ..just remember....Family is EVERYTHING!! Posts: 3944 | From: Rochester, NY | Registered: Sep 2006
| IP: Logged |
posted
PCola, thanks for the continued contribution to the site. I, for one am starting to get a little tired of losing money playing the pennies. This may be the next phase I have been looking for. Unfortunately I dont have the time to watch L2's every minute of the day to figure out when to buy and sell so this added weapon to the arsenal could be a turning point. One question. Does the "mining" take into account form 4s that show where insiders are issued shares instead of buying or is that one of the filter points that you as the information receiver have to filter? Just wondering. Thanks, TMAN...
-------------------- In the end, trust only yourself when trading stocks. Posts: 1993 | From: Texas | Registered: Feb 2006
| IP: Logged |
posted
Nice... PCola can you share some of the other stocks that you watched during the trial run of this program. A few winners and losers. I would like to check them out if that is possible. Thanks, TMAN...
-------------------- In the end, trust only yourself when trading stocks. Posts: 1993 | From: Texas | Registered: Feb 2006
| IP: Logged |
quote:Originally posted by tmanfromtexas: Nice... PCola can you share some of the other stocks that you watched during the trial run of this program. A few winners and losers. I would like to check them out if that is possible. Thanks, TMAN...
PCola I second this request
-------------------- Be Careful Of The Toes We Step On Today, They Could Be Attached To The Butt We Have To Kiss Tomorrow Posts: 4721 | From: Elk Grove ( Sacramento )CA USA | Registered: Jun 2004
| IP: Logged |
posted
Two good ones were RMDX which I would have bought at the open on 8/20 for about $1.85 (currently at 2.95, and would not have sold yet since it would not have hit my 5% stoploss), and BEXP, which I would have bought on 8/14 for about $4.50 (don't have the actual numbers in front og me, but looking at the graph I probably would have sold in the last day or two in the $6.00 range).
One bad one was JAH, which I would have bought on 9/5 at about $32.50 and stopped out around $30.00 a couple of days later. It bounced back up to almost $35.00, so had I held, I still could have made about 7%.
Another bad one was NRP, which I would have bought at just under $32 and stopped out a few days later at about $29. It bounced back to about $35 as well.
Actually, I'm glad you asked that question, because (as I said in the write-up) I simplified my test with a 5% stoploss, but this excercise showed that maybe I shouldn't put the stoploss in right away, so I won't sell out in the first few days to allow myself a little wiggle room.
Just one of the many things I still need to test, I guess.
Just checked two of the other losers, and one I missed out on 30% and the other like 15%. So I feel confident that my estimates are significantly understated, I just have to create a better sell rule.
Another reason that I need to create a better sell rule is IOM, which I would have bought at 2.40 and sold at 2.70 a month or so later for a nice 12% gain, but was 5.75 the other day, representing another 125% that I missed out on.
Edit: just noticed that the IOM would have been bought in 8/06, so forget that one, it wasn't in my 2007 analysis, but you get the idea, I'm sure.
-------------------- I'm back! Posts: 5431 | From: West Chester, PA | Registered: Jan 2006
| IP: Logged |
posted
Thanks PCola. If you get any hits you would like to share let us know if you dont mind. I plan to paper trade a few at first to see how I do. I see that the stop losses can hamper any gains but they sure are good when one goes straight down the crapper. TMAN...
-------------------- In the end, trust only yourself when trading stocks. Posts: 1993 | From: Texas | Registered: Feb 2006
| IP: Logged |
posted
Yeah Tman, that's the way to go, and I plan on posting them all here as they pop up. I need to spend some time making the process more efficient, because as of now it takes like 14 hours
It's set up to pull all the data from scratch, but now that I have the historical data I just need to add 1 day's data each day. I'll get to that part of the proess soon though, and it should only take a few minutes at that point.
-------------------- I'm back! Posts: 5431 | From: West Chester, PA | Registered: Jan 2006
| IP: Logged |
posted
The whole opint of me doing this is to get a strategy that doesn't require me to watch every tick during the day... Yet, of course, I'm sitting here refreshing my account every 5 minutes... Some day I'll quit obsessing...
But for now, hit an HOD a few minutes ago of 28.64, which is just under 5% gain. Adjusting the trailing stoploss should basically mean that worst case scenario from here is break even.
-------------------- I'm back! Posts: 5431 | From: West Chester, PA | Registered: Jan 2006
| IP: Logged |
posted
Thanks man. I'm currently running some tests adjusting the sell rules, I'll report back with what I find. One thing I'm doing is starting with a higher stoploss and grading it down over time (i.e., have a 10% stoploss aftre day 1, 9% after day 2, etc down to 5% after day 5). This may eliminate cases where there is a decent initial drop that would take out your stoploss, only to see it recover and end up being a good one.
Another idea is to look at each peak and adjust he stoploss from there. Like if the peak is up 10% have the stoploss at 5%. But if it peaks at +40%, increase the stoploss to 10% to give it more room to pullback, in case it runs again.
I think I have like 10 million ideas that I need to test...
-------------------- I'm back! Posts: 5431 | From: West Chester, PA | Registered: Jan 2006
| IP: Logged |
UBBFriend: Email this page to someone!
![[Big Grin]](biggrin.gif)
![[Smile]](smile.gif)