StockFetcher Forums · Filter Exchange · Telling the Future...<< 1 2 3 4 >>Post Follow-up
heyen
124 posts
msg #43756
Ignore heyen
5/16/2006 5:49:46 AM

hscott,

do you use "select by c43 descending" ?
Thats where i select the most market following stocks to enter.
Also, you used a full year for backtesting.
Have you checked the precaution toppers for that time?
Fetcher[
market is nasdaq
avgvol(90) above 25000
set{c1,count(day change 1 day ago above 0,1) + count(ind(qqqq,day change) above 0,1)}
set{c2,count(day change 1 day ago below 0,1) + count(ind(qqqq,day change) below 0,1)}
set{precaution,count(max(c1,c2) equal 2,100)}
precaution above 60
add column precaution
sort by column 5 descending
date offset 12/30/2005
]




heyen
124 posts
msg #43757
Ignore heyen
modified
5/16/2006 6:44:41 AM

I have tried different shorter time frames for the last 1 year:
Performance increasing to date. The most recent 1.5 months are the best, if using the precaution top 4 for March (VRTX, IBCP,SBCF,MFLX).

These must be updated regularly. Either this concept found a temporal anomaly, a strong sector/industry influence on a indifferent market or a space time continuum wormhole distortion.



hscott
34 posts
msg #43758
Ignore hscott
5/16/2006 8:50:41 AM

heyen - No I didn't use "select by c43 descending". But I knew you must have something like that - it just didn't show up in your post.

I haven't run the filter over a selection of time periods - just starting with it and would not use it without fully testing.

The "preferred" stocks that predict the Nasdaq change so what you really need is a phony "stock" that changes with time according to your filter but I have no idea how to mock this up.

Here is something I tried - instead of correlating with Nasdaq, I chose the best performing etf over last 100 days (which is IAU) and correlated with that. It did a little better than with Nasdaq.


hscott
34 posts
msg #43763
Ignore hscott
5/16/2006 11:46:08 AM

heyen - You are a daytrader I guess. But I gererally hold stocks a week or two. So I am interested in what the nasdaq is going to do in the next couple of weeks. So I modified your filters to extend the prediction time.

This filter shows stocks that correlate with nasdaq gain for 10 days hence.

market is nasdaq
avgvol(90) above 25000
set{c1,count(day change 10 days ago above 0,1) + count(ind(qqqq,close) - ind(qqqq,close 10 days ago) above 0,1)}
set{c2,count(day change 10 days ago below 0,1) + count(ind(qqqq,close) - ind(qqqq,close 10 days ago) below 0,1)}
set{precaution,count(max(c1,c2) equal 2,100)}
precaution above 60
add column precaution
and add column c1
and add column c2
sort by column 5 descending

So then I take this filter and construct another (as you did) for trading. Results are pretty good. See attached.

Approach Information
Approach Name: market is nasdaq avgvol(20) above 50000 set{c1,cou...
Test started on 06/30/2004 ended on 03/31/2006, covering 442 days
Filter used:
market is nasdaq
avgvol(20) above 50000
set{c1,count(ind(qqqq,day change) above 0,1) + count(day change above 0,1)}
set{c2,count(ind(qqqq,day change) below 0,1) + count(day change below 0,1)}
set{c3,count(max(c1,c2) equal 2,100)}
add column c3
sort by column 5 descending
set{c4,count(ind(aemlw,day change) above 0,1)+count(ind(wlsc,day change) above 0,1)}
set{c5,count(ind(rock,day change) above 0,1)+count(ind(urgi,day change) above 0,1)}
set{c6,c4+c5}
c6 above 3
c3 above 77

Trade Statistics
There were 46 total stocks entered. Of those, 31 or 67.39% were complete and 15 or 32.61% were open.
Of the 31 completed trades, 27 trades or 87.10%resulted in a net gain.
Your average net change for completed trades was: 7.34%.
The average draw down of your approach was: -5.16%.
The average max profit of your approach was: 9.38%
The Reward/Risk ratio for this approach is: 21.90
Annualized Return on Investment (ROI): 62.41%, the ROI of ^SPX was: 7.80%.

Exit Statistics
Stop Loss was triggered 0 times or 0.00% of the time.
Stop Profit was triggered 11 times or 35.48% of the time.
Trailing Stop Loss was triggered 0 times or 0.00% of the time.
You held for the maximum period of time (250 days) 0 times or 0.00% of the time.
An exit trigger was executed 20 times or 64.52% of the time.

Statistics By Holding Period
 Completed2 day chg5 day chg10 day chg25 day chg40 day chg
Winners:272020243325
Losers:42526221321
Win/Loss Ratio:6.75:10.80:10.77:11.09:12.54:11.19:1
Net Change:7.34%0.23%0.58%1.91%8.26%7.32%

Statistics By Variable: Match Price
 <10<20<30<40<50<60<70<80<90<100
Completed6:07:07:36:1--1:0---
2 day chg4:23:85:95:32:1-1:00:1-0:1
5 day chg4:24:74:105:42:1-0:11:0-0:1
10 day chg6:05:63:116:31:2-1:01:0-1:0
25 day chg6:07:49:57:21:2-1:01:0-1:0
40 day chg6:06:56:86:30:3-0:11:0-0:1

Statistics By Variable: Average Volume
 <10.0M<20.0M<30.0M<40.0M<50.0M<60.0M<70.0M<80.0M<90.0M<100.0M
Completed26:4------1:0--
2 day chg19:24------1:1--
5 day chg18:26------2:0--
10 day chg24:20------0:2--
25 day chg32:12------1:1--
40 day chg24:20------1:1--



heyen
124 posts
msg #43764
Ignore heyen
5/16/2006 12:24:03 PM

hscott! sweet! 87% Win rate - thats a beauty!
What was your profit stop???????

Your trending forecast idea is nice, we would have to use a smoothed directional
indicator comparison with a 1 or 2 week offset. Arghh, we are getting complex here.

Comparing day-to-day correlation over 100 days is one thing, comparing trend correlation of 5 to 10 days is another book.


hscott
34 posts
msg #43766
Ignore hscott
5/16/2006 12:35:46 PM

heyen- I used a 12% profit stop with no stop loss.

Yes I have noted that all these Nasdaq correlation filters give a very high win rate. I am not used to seeing 90% win rates and I wonder if it would work out in real life.

I don't really know what you have in mind with DI - why don't you give it a try.


heyen
124 posts
msg #43769
Ignore heyen
5/16/2006 1:37:09 PM

darn! i didnt realize you had the loosers in open positions.
that dawned at me, when i achieved 100% winners in my backtest.

still not found the 100% perfect crystal ball, but close...... to be continued


hscott
34 posts
msg #43770
Ignore hscott
5/16/2006 1:45:16 PM

"losers in open positions" - yes that is a major flaw in backtesting. Gives you a false sense of success. I don't know of any easy way around it. If you are doing a backtest a way back in the past you can make sure all positions are closed by setting max days to hold. But doing a current backtest, all I know to do is to manually calculate gains and losses using issue price as of the end date of the backtest. Generally this brings down the ROI.


shelupinin
120 posts
msg #43953
Ignore shelupinin
5/20/2006 7:03:24 PM

heyen, I would like to fetch stocks in your filter
Fetcher[market is nasdaq
set{c1,count(day change 1 day ago above 0,1) + count(ind(qqqq,day change) above 0,1)}
set{c2,count(day change 1 day ago below 0,1) + count(ind(qqqq,day change) below 0,1)}
set{precaution,count(max(c1,c2) equal 2,100)}
volume above 5000
and add column precaution
sort by column 5 descending
]

with "precaution" value equals to "precaution" of QQQQ but I don't know how to do that :( Could you help me?
Thanx
Alex


heyen
124 posts
msg #43955
Ignore heyen
5/20/2006 8:52:57 PM

Shelupinin, i might not understand your intention. The "precaution" screen just checks for the statistic occurence in stocks that go up, 1 day before the qqqq goes up or go down, 1 day before the qqqq goes down. The idea was to have an idea in which direction the market will go tomorrow. Just an idea out of curiosity. And i was surprised to find stocks beeing 1 day ahead about 70% of the time. So i dont see advantage of finding stocks equalling that statistic value of the qqqq. I'd rather search for an outstanding deviation way above the average and assume the future from the past.

Now if you really want to see how often the qqqq foretells itself just add:
precaution equal ind(qqqq,precaution)

Fetcher[
market is nasdaq
set{c1,count(day change 1 day ago above 0,1) + count(ind(qqqq,day change) above 0,1)}
set{c2,count(day change 1 day ago below 0,1) + count(ind(qqqq,day change) below 0,1)}
set{precaution,count(max(c1,c2) equal 2,100)}
volume above 5000
and add column precaution
sort by column 5 descending
precaution equal ind(qqqq,precaution)
]



another interesting screen:

Fetcher[
market is nasdaq
set{c1,count(day change 1 day ago above 0,1) + count(ind(qqqq,day change) above 0,1)}
set{c2,count(day change 1 day ago below 0,1) + count(ind(qqqq,day change) below 0,1)}
set{precaution,count(max(c1,c2) equal 2,100)}
volume above 5000
and add column precaution
sort by column 5 descending
date offset 05/18/2006
precaution above 65
]



1 day before the market gained ground again, the top 5 outnumbered 3-to-2 gainers over lossers....


StockFetcher Forums · Filter Exchange · Telling the Future...<< 1 2 3 4 >>Post Follow-up

*** Disclaimer *** StockFetcher.com does not endorse or suggest any of the securities which are returned in any of the searches or filters. They are provided purely for informational and research purposes. StockFetcher.com does not recommend particular securities. StockFetcher.com, Vestyl Software, L.L.C. and involved content providers shall not be liable for any errors or delays in the content, or for any actions taken based on the content.


Copyright 2016 - Vestyl Software L.L.C.Terms of Service | License | Questions or comments? Contact Us
EOD Data sources: DDFPlus & CSI Data Quotes delayed during active market hours. Delay times are at least 15 mins for NASDAQ, 20 mins for NYSE and Amex. Delayed intraday data provided by DDFPlus