Kevin_in_GA
4,599 posts
msg #101237
- Ignore Kevin_in_GA |
6/13/2011 6:36:29 PM
Follow up - 2 exits triggered today:
LO - entry at 97.00, closed trade today at 101.57 (+4.71%)
R - entry at 50.71, closed trade today at 50.90 (+0.37%)
Replace these two with CHK and IP at the open price tomorrow.
|
duke56468
683 posts
msg #101239
- Ignore duke56468 |
6/13/2011 8:39:42 PM
four or anyone....do you know this answer? In the backtest if I leave the "selection method" blank will SF choose the stocks in the order of the filter selection or do I have to write in a selection method??
|
Kevin_in_GA
4,599 posts
msg #101240
- Ignore Kevin_in_GA |
6/13/2011 9:16:47 PM
I use the following:
SET{PRICERATIO, CLOSE / IND(^spx,CLOSE)}
SET{RATIOMA16, CMA(PRICERATIO,16)}
SET{RATIOSTD16, CSTDDEV(PRICERATIO,16)}
SET{DIFF16, PRICERATIO - RATIOMA16}
SET{ZSCORE16, DIFF16 / RATIOSTD16}
SET{THRESHOLD16, RATIOSTD16 * 2}
select by zscore16 ascending
This insures that the stocks with the lowest z-score are chosen ahead of any others.
|
fortyfour
189 posts
msg #101243
- Ignore fortyfour
modified |
6/14/2011 2:51:54 AM
Tradeoff........ Maybe this (or any other ?) "market weakness" test wouldnt kill the performace
by too much.......reduce drawdowns......maintain a decent sharpe ratio and allow
more individuals (with lower risk tolerances) to trade the system at all times.
There is an area where the market is weak (walking lower bb's) yet leaders have a long way to fall to ma(200)..
that may produce many consecutive losses? or ... Maybe not. I know it hurts equity but I dont know how well
it reduces drawdown...
PS: Try this::: user Stockfetchers backtester for Kevins filter from 5/15/2008 - 7/15/2008 ( Downtrend where SP500 lost
about 4%) . The system made about $3000 on an initial 100K during that 2 month period. That is indeeed surprising !
Make a filter of just "symlist(spy)" and use HISTORICAL to look at that period before the meltdown.
I
|
shainadir
25 posts
msg #101246
- Ignore shainadir |
6/14/2011 1:48:10 PM
Kevin,
Great work by the way.
Not sure if this is considered curve fitting, but using the highest correlation instead of lowest z-score in my backtests on stockfetcher have been producing slightly better equity results. I use the statement, select by corr(^spx,10,Close) descending" for the selection method. Can anyone test this on Stratasearch? Thanks.
|
Kevin_in_GA
4,599 posts
msg #101247
- Ignore Kevin_in_GA |
6/14/2011 4:24:25 PM
shainadir:
I can check this tonight. Odds are that the differences will be small over the 10 year period tested.
I cannot use StrataSearch to put together a pair trade for these stocks (with SH as the other half). It's not really set up to do that type of analysis. Obviously that strategy would be useful, but would probably result in a lower overall profit since much of the time the market went up (slowly) versus down (rapidly).
|
novacane32000
331 posts
msg #101260
- Ignore novacane32000 |
6/15/2011 7:27:09 PM
Does anyone know what the time delay is on the SF filters?
At 4:40 pm the Zscore filter showed 6 or 7 stocks. Now at 7:20pm it only shows 3 ??
Same with the exit filter. None at 4:40pm ,at 7:20 pm it shows one.
|
duke56468
683 posts
msg #101261
- Ignore duke56468 |
6/15/2011 8:17:29 PM
- Ignore novacane32000 6/15/2011 7:27:09 PM
Does anyone know what the time delay is on the SF filters?
++++++++++++++++++++++++++++++++++++++++
SF offers 2 e-mail times 6PM and 9PM(to allow for corrections and final data)
|
saico
59 posts
msg #101300
- Ignore saico |
6/20/2011 3:00:39 PM
Hi Kevin,
first of all thanks for the great work! I wonder if its possible to imply to the exit filter a code that gives an exit signal after the 20 days maximum holding time passed by?
saico
|
Kevin_in_GA
4,599 posts
msg #101302
- Ignore Kevin_in_GA |
6/20/2011 7:47:18 PM
Saico:
Unfortunately this cannot be done in SF. You would have to track this manually.
Kevin
|