TheRumpledOne
6,411 posts
msg #100431
- Ignore TheRumpledOne
modified |
4/26/2011 3:22:30 PM
Buy at close on 4/26/11
Try to sell for profit on 4/27/11
RESULTS
Symbol Last pl
TAT 2.61 0.03
TLAB 4.81 0.01
HL 9.49 0.40
LWSN 11.06 0.06
ERTS 20.24 0.03
IAG 20.57 0.77
SWC 21.73 1.34
CEF 25.1 1.47
AN 33.36 0.46
The first 2 didn't pay off but the rest more than made up for it!!
|
mahkoh
1,065 posts
msg #102858
- Ignore mahkoh
modified |
10/7/2011 5:13:21 PM
I have been busy with this filter recently. During my vacation I did not have constant internet and computer access and decided this would be a good time to test a strategy. My idea was to make a short and a long version of the filter and place market on close orders on 10 longs and 10 shorts. This should make me market neutral.
I set the filter so that it looks for an 80 % probabilty of a 0.9% profit over yesterday's close, scanning for stocks over 8$ as you may get some slippage with MOC orders. I choose to sort by OVERX because I want to put a bit more weight to recent data.
If statistics are correct I expect to get stuck with about 4 stocks (20 % of my position) every day that I will likely have to close out with a loss. The question is: Will my 16 daily winners cover for my 4 daily losses? Take into consideration that a 0.8% profit counts as a loser.
As I considered the possibility that I would occasionaly hold a losing position overnight I added TRO's trend indicator and a measurement of the 8 day slope of the 30 MA(q). I let this value weigh in deciding which stocks to play for the next day.
You will have stocks gapping against you (and hopefully stocks gapping in your desired direction as well). I added a "fading the gap" exerpt so that in case of a gap against me I could reset my target to breakeven and have 70 % probability of the gap being filled.
The long filter:
and the short filter:
My vacation lasted 3 weeks and statistics did indeed prove correct: Roughly half of the positions are closed out within minutes after market open, and usually after 1 hour there would be 4 problem cases left in the portfolio.
In general it proved best to close those out between 10:30 and 11:30.
The testing period was with $3.500 per position on paper from sept 5 until sept 23. The first weeks went very well with daily profits averaging $250. Top day was over $800, worst was a $30 loss. One can imagine that I had a very pleasant vacation thinking about where to spend all the money flowing my way when going live with this strategy.
Then in the last three days things started to deteriorate. I got stuck with more losing positions and it became very hard to just break even. Nevertheless I went live when I got home, after all it would not be very realistic to think there would never be a losing streak. I did scale back a bit, taking on between 10 and 14 $2500 positions.
It has not been easy since, in the 3 weeks since going live there have only been a couple of days with tiny profits, I have switched back to paper a couple of times and of course those were the days with decent gains. I am currently back on
paper as I still think that this startegy should work.
So far I haven't used stop losses, but it in practice it turns out that losing positions rarely recover once the loss is beyond
3 to 4%. This seems reasonable, after all 4 out of 5 should be winners. The winners generate 4 x 1% and if you close the loser below 4 % there should be a profit in the long run.
|