StockFetcher Forums · Filter Exchange · MODIFIED CONNORS RSI(2) FILTER<< 1 ... 16 17 18 19 20 ... 22 >>Post Follow-up
wkloss
230 posts
msg #95817
Ignore wkloss
8/24/2010 7:16:20 PM

Kevin,

Is there any way to backtest the composites? I tried, got mixed results then realized that it was difficult if not impossible to duplicate your entries on Mondays and exits no later than Friday's close. There is also the question of how many stocks you would buy in a day and your maximum number of open positions.

Bill

Kevin_in_GA
4,547 posts
msg #95823
Ignore Kevin_in_GA
8/24/2010 10:33:38 PM

You should be able to backtest the new composite filters I posted - just copy and paste. Set the selection criteria to

"composite above X" (set X at whatever threshold you choose - start at 0.5 and go up 1 unit each time)

I would not limit the number of trades for a backtest - more data is better.

all exits are close above/below MA(10). You can also set a max hold for 10 days or less.

No stop loss or profit stop.

Kevin_in_GA
4,547 posts
msg #95824
Ignore Kevin_in_GA
8/24/2010 10:49:26 PM

Update 8/23:

Rough day today - MDCO and DYN both went the wrong way bigtime. Still beating the market for the week, although trading at a slightly lower loss is not warming my heart all that much.


Long Plays selected from the new filters:

NTY (scoring 3/5) - in at 54.13, closed at 53.99 for a loss of 0.35%
MDCO (now scoring 4/5) - in at 12.26, closed at 11.52 for a loss of 6.12%


Short Plays selected from the new filters:

DYN (scoring 1/5) - in at 4.74, closed at 4.94 for a loss of 4.31%

GRS - in at 7.03, closed at 6.68 for a gain of 4.88%


Net so far for the week - a loss of 1.47% (down $589.49 on $40,000 invested in these 4 stocks, which includes commissions at $8.95 per trade)

Remember - all exits for these are based on a close above/below the MA(10).


Other plays flagged this week:

Monday

ARNA (2/5) - Long Play

PCBC (2/5) - Short Play (watch out on this one because of the stock split next week)


Tuesday

DECK (3/5) - Long Play

wkloss
230 posts
msg #95828
Ignore wkloss
8/25/2010 1:12:55 AM



DYN is not a failure of your signal. It is a takeover play by Blackstone Group. You should not expect it to fall in price.

This is a very minor flaw in your strategy and unfortunately not one to be concerned about since you do not have time to research every trade and you only have a very small amount at risk on each one. Some of the losing trades in the backtesting are probably special situations like this one

Kevin_in_GA
4,547 posts
msg #95829
Ignore Kevin_in_GA
8/25/2010 6:35:37 AM

Agreed. Since I didn't finish these filters until late Sunday night, I would normally have had time to look at these and might have flagged this one as risky.

Kevin_in_GA
4,547 posts
msg #95839
Ignore Kevin_in_GA
8/25/2010 9:11:14 PM

Update 8/25:

Better day today - no new long plays, one new short play (CWTR).


Long Plays selected from the new filters:

NTY (scoring 3/5) - in at 54.13, closed at 54.15 for a loss of 0.05% after subtracting commission
This closed above its MA(10) - 54.10. Trade will be closed at the open tomorrow.

MDCO (now scoring 4/5) - in at 12.26, closed at 11.91 for a loss of 2.94%


Short Plays selected from the new filters:

DYN (scoring 1/5) - in at 4.74, closed at 4.88 for a loss of 3.04%

GRS - in at 7.03, closed at 6.88 for a gain of 2.04%


Net so far for the week - a loss of 1.00% (down $399.65 on $40,000 invested in these 4 stocks, which includes commissions at $8.95 per trade).

Compare this to SPY - down 1.94% since the open on Monday.


Possible replacement Long trades (no new ones today, but these are from this week):

Monday

ARNA (2/5) - Long Play
Opened Tuesday at 6.69, closed today at 6.82 for a gain of 1.94%. MA(10) is currently at 6.90.

Tuesday

DECK (3/5) - Long Play
Opened today at 44.90, closed at 46.07 for a gain of 2.60% (already closed above its MA(10), so it is no longer a candidate).

I will wait to see the futures before selling or buying anything tomorrow, but if things look good I'll sell NTY and buy ARNA - not much room left there (about 1% before it closes above its MA(10)).

Kevin_in_GA
4,547 posts
msg #95849
Ignore Kevin_in_GA
modified
8/26/2010 10:56:44 AM

NTY sold at the open at 54.15.

I decided to put in a limit order for ARNA at the open price on Tuesday (6.69) and see if I could get it at that price - that way I would not miss out on any of the possible gain by being in 2 days later. So far not filled.

EDIT: ARNA limit order filled at 6.69. Let's see how this plays out ...

saratur
14 posts
msg #95861
Ignore saratur
8/27/2010 2:44:23 AM

Kevin - great work, thank you for sharing. Very encouraging results.

Questions regarding parameter optimization with StrataSearch (I am not familiar with it):
- does it allow user defined target/fitness function, and/or 'canned' system oriented measures (e.g. Van Tharp's Health, linear correlation etc.) - or net profit only?
- any robustness tests via parameter sensitivity, and/or walk forward optimizations?
... and if yes, did you get to do such tests with these filters?





Kevin_in_GA
4,547 posts
msg #95862
Ignore Kevin_in_GA
8/27/2010 8:38:38 AM

saratur:

StrataSearch can use fitness criteria to select trading systems - many are built-in already. You can also select key elements that are important for you (e.g., K-score, low drawdown, high Sharpe ratio) and weight them as you see fit. The result will "guide" the filter selection process and gives you a fitness score through which the results can be filtered.

A fully functional download is available - try it for 30 days. Honestly, that is enough time to build and test lots of possible systems, since the runs are quite fast even against thousands of stocks or decades of data (they let you go back as far as 1980 for testing).

I started by using their "one click" strategies, which do everything for you, but ultimately wanted to translate these into SF code. Some of their criteria and indicators are not part of SF, so I ultimately decided to look at single indicators (CMO, TSI, Bollinger bands) and combine them into the current system.

I did some robustness testing on these settings, but not as much as I would have liked. I also need to make sure that the filters are connected to the current state of the broader markets - I am re-running them looking at three states (market roc is above 20, below -20, and then everything in between). Different filter settings can then be used that are most effective for a given trading environment. Easy to translate to SF once the results are in.

Expect a new set of filters in a week or so - constantly trying to improve them, but will settle on whatever this last set of filters is for the rest of this year.


saratur
14 posts
msg #95880
Ignore saratur
8/29/2010 4:15:35 PM

It seems StrataSearch is pretty powerful. I never had the bandwidth to try it, your work encourages me to try make the time.

"Good" optimization fitness function(s) and parameter robustness checks are very important. My experience in optimization is that systems with parameters tuned to maximum net profit in a backtest quite frequently crumble, or perform significantly worse, in forward testing.

StockFetcher Forums · Filter Exchange · MODIFIED CONNORS RSI(2) FILTER<< 1 ... 16 17 18 19 20 ... 22 >>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