Kevin_in_GA
4,599 posts
msg #104247
- Ignore Kevin_in_GA |
1/6/2012 12:53:35 PM
I've been kicking around several ideas based on others' work here on trading opening gaps. Most people have looked at TRO's work in this area (Milk the Cows, his Fade the Gap statistics filters, etc). If you have not looked at these, you really should.
My preference is to "set it and forget it" since I cannot daytrade - so it made sense for me to look at weekends (the close on Friday to the open on the following Monday). Many stocks will gap up or down only to return back to the previous week's close, albeit temporarily.
So what does the filter below do? It shows you the statistics for the number of times a particular stock has gapped down at the open on Monday from the previous week's close, and also the frequency of which it then went to fill that gap over the next few days.
Look for stocks that frequently gap down and then fill that gap. At the open on Monday you'll see what stocks have gapped down, how frequently that has occurred over the last 100 weeks, and the percentage of times that gap was filled. It also calculates your predicted gain - easy to do since it is just the difference between the open and last week's close.
I have set the absolute numbers of gaps to be at least 30 over the last 100 weeks. I personally would only look seriously at stocks that have filled those gaps at least 85% of the time.
REMEMBER: This filter is only run on Monday at the open. You set a stop limit at the previous week's close and then ignore these until Friday. If they have not closed by then, make sure to close them out before the end of the trading week.
And before a half dozen people post "How do I backtest this?", StockFetcher does not let you set this up and backtest it because it must be run only on a specific day of the week.
Also note that I have not backtested this, but plan to try and write this code into StrataSearch and get some backtest results over the weekend. No promises that this can or will be done.
And here's the short trade equivalent:
Enjoy.
Kevin
|
Nurdish
2 posts
msg #104249
- Ignore Nurdish |
1/6/2012 1:10:32 PM
Kevin - some of your work here is inspirational, I joined after seeing yours + eman's posts. You guys are great.
|
Kevin_in_GA
4,599 posts
msg #104251
- Ignore Kevin_in_GA |
1/6/2012 1:46:03 PM
Much appreciated. I would not act on this one until I have had a chance to thoroughly backtest it. For now, consider it a work-in-progress.
I had also thought of using other triggers (e.g., the previous weeks pivot points as entry and exit). I may come back to that idea a little later, once this one is more fully developed.
|
sohailmithani
192 posts
msg #104252
- Ignore sohailmithani |
1/6/2012 2:34:25 PM
great filter. Thanks for sharing. I tested it a bit using date off set option and results were excellent. Now in my test what I did the following:
1) only took top 3 stocks (top means percent filled over 90 and gapdown100 to be over 30),
2) close between $5 and $100,
3) gain (max) target restricted between 2% and 4%,
4) excluded ETF and
5) excluded those 5 alphabet stocks from the list.
The results on avg showed gain of between $600-800 per week even in 2008 (tested some months). This one looks to be the one I always wanted.
Any thoughts?
|
mahkoh
1,065 posts
msg #104258
- Ignore mahkoh
modified |
1/6/2012 5:31:07 PM
One thing: Stockfetcher doesn't generate Monday's data until we're at least half an hour underway in the new trading week. (And this week I had to wait until 10:30 before it stopped generating data from Friday 2011-12-30)
|
Kevin_in_GA
4,599 posts
msg #104263
- Ignore Kevin_in_GA |
1/6/2012 6:24:51 PM
One thing: Stockfetcher doesn't generate Monday's data until we're at least half an hour underway in the new trading week. (And this week I had to wait until 10:30 before it stopped generating data from Friday 2011-12-30)
+++++++
I know. The delay limits this system trading real time. However, the real value is in the statistical analysis, which can be run over each weekend to generate all stocks that meet a certain criterion (e.g., 90% fill rate, at least 30 gaps down/up, etc).
That list can be used with other real-time systems to trade as described here. After all, the only numbers you need to fill the trade are last week's close and this week's open. No indicator needed.
|
Billirider321
62 posts
msg #104285
- Ignore Billirider321 |
1/8/2012 11:26:44 AM
One risk we need to consider on gap ups is potential takeover candidates. In those cases we will risk having shors on stocks which will be a big looser. There may be only very few candidates like this but we will always have that stress.
We can always research during the weekend and eleminate as much as possible. Second option is to take position in puts.
we can also look at big cap companies who are not take over candidates. But ther chance of gap fill will decrease. Guys this is realy a awesome idea(as most of the idea's of Kevin are). Lets try to improve with mutual coordination.
|
shillllihs
6,047 posts
msg #104286
- Ignore shillllihs |
1/8/2012 4:17:16 PM
I know what you're doing kevin, and i like it. Steering the sheeple. thanx for the loophole.
|
wkloss
231 posts
msg #104287
- Ignore wkloss |
1/8/2012 5:10:32 PM
Kevin,
Any thoughts on testing your balanced long/short approach on this one?
Bill
|
mahkoh
1,065 posts
msg #104292
- Ignore mahkoh |
1/9/2012 3:46:08 AM
Another issue I have come across in he past with filters that refer to the opening price: Most stocks have a substantial spread during the first minutes of trading, often over 2 %. So, while there is a very slim chance of actually getting the opening price these values are used in the filter calculations to come up with candidates.
|