wkloss
231 posts
msg #110902
- Ignore wkloss |
2/2/2013 10:09:29 PM
evo34
A few thoughts to consider:
Weren't industries like you want to exclude included in the excellent test results?
Aren't a lot of "troubled" companies included in the test results? Isn't the underlying concept of this strategy that a lot of oversold companies have fallen beyond a reasonable point?
I'm not trying to say you are right or wrong. I'm suggesting that you might need to review the list of trades to verify your idea. If you do and it improves performance, then you have improved a statistical edge. Otherwise, you are taking a mechanical system and turnng it into a discretionary system. I hope your results are better than mine, but everytime I go discretionary, I increase my lifetime tax deductions.
Bill
|
evo34
82 posts
msg #110905
- Ignore evo34 |
2/3/2013 1:32:37 AM
Yeah, that's the whole idea. You need to confirm that excluding certain industries will not degrade you past test results. Based on my testing on other platforms (with other systems), excluding medical/biotech usually improves your reward/risk ratio. My point is that in SF, you are not able to test the hypothesis, which is unfortunate.
As for troubled companies, I haven't suggested to avoid them. I suggested avoiding stocks which are about to experience known high-volatility events that your system probably has no ability to predict. Again, in an ideal world, you would be able to test this; although, getting accurate EPS date data into a backtesting tool is rather difficult. But unless you have reason to believe that your system results are driven by such outlier cases (upcoming earnings or drug trials), I believe it is advisable to exclude them simply bc the variance of your returns for such trades goes through the roof.
It's very different if you have a fundamentals-based or event-based system, of course. There, you are relying on big events and so certainly could not try to exclude them. But when the majority of your tested trades are not subject to major news events, it usually will not hurt to avoid the stocks you think will experience them very soon.
|
miko
68 posts
msg #110917
- Ignore miko |
2/4/2013 12:20:21 AM
An alternative to excluding (because of higher volatility, eg earnings/biotechs) is to reduce the position size for those trades only. It seems human nature to be all or nothing, black or white, but there is plenty of space in between. Of course that guarantees that you won't be right (if it loses, you shouldn't have traded it at all, if it wins, you should have traded full size).
|
novacane32000
331 posts
msg #110919
- Ignore novacane32000 |
2/4/2013 7:58:37 AM
Does someone have the settings used to backtest this system on SF?
|
compound_gains
225 posts
msg #110925
- Ignore compound_gains |
2/4/2013 8:49:15 AM
That being said, I think you'll find if you back check that "medical - biomedical" stocks account for a lot of the entries with this filter...it's simply their volatility...so you've changed the whole system if you exclude them.
====================================================================================
I was wrong about the prevalence of Medical - Biomedical stocks in the population of stocks triggered by the system that's being used here. In 2012 there were some 378 stocks triggered using the set-up on page 19 (msg #110375) and a trigger where the next day's low is 6% or more below the previous day's close. Among these, 15% (n=56) were in the broad Medical sector and only 10% (n=38) were in the industry Medical - Biomedical. If someone was interested enough you could go back and generate a reasonable estimate of what the return would have been on those 38 Medical - Biomedical stocks, or any other industry for that matter.
|
Kevin_in_GA
4,599 posts
msg #110958
- Ignore Kevin_in_GA |
2/4/2013 1:44:41 PM
New trade triggered this morning.
HLF - limit entry at 32.97, currently at 34.09 (up 3.4%).
|
Kevin_in_GA
4,599 posts
msg #110961
- Ignore Kevin_in_GA |
2/4/2013 2:41:42 PM
And one more:
KERX - limit entry at 6.68, currently at 6.76 (up 1.2%).
|
gmg733
788 posts
msg #110966
- Ignore gmg733
modified |
2/4/2013 3:07:43 PM
Kevin,
Looking at the page 19 filter, you have a 4% pull back candidates and a 6% pull back for the entry. Is this correct?
low more than 4% below close 1 day ago
set{limitentry, close *0.94}
During the WFA, are you optimizing these two variable independently?
Thanks.
|
Marketyoda
11 posts
msg #110989
- Ignore Marketyoda |
2/4/2013 4:54:06 PM
Kevin-
Do you have a max stock price that you will trade? Has there been anything you've seen in backtesting to hint at higher priced stocks not returning as much as other price ranges?
Thanks!
|
Kevin_in_GA
4,599 posts
msg #111005
- Ignore Kevin_in_GA |
2/5/2013 9:47:21 AM
@gmg: Yes, both are optimized in the WFA, once a month.
@marketyoda: No upper limit was used in the backtest. Lower limit was 5 dollars. The key really is liquidity - illiquid stocks will screw you on the exit (not the entry since it is a limit order, but it may not get filled).
|