Kevin_in_GA
4,599 posts
msg #112371
- Ignore Kevin_in_GA |
3/22/2013 9:20:05 AM
@Xglassm:
Each backtest was run independently, with you exiting the trade if an EXIT signal triggered. If a BUY signal also triggered, the backtest sold you out at the open of the next day and simultaneously bought back in. In essence you just paid the extra commission. The frequency of trades is therefore higher but the results should be the same if you ignore the commissions.
|
jackmack
334 posts
msg #112384
- Ignore jackmack |
3/22/2013 2:49:54 PM
Kevin
I know this is a purely mechanical system but I was thinking of ways to prevent substantial draw downs when the market acts up from time to time and was just looking at price when (if) below the EMA(20) AND the EMA(50).
If a buy signal was given and a position entered then price moved through both EMA's without an exit signal given
and positions then exited before an exit signal was given would this improve the draw down and prevent possible greater losses?
Just curios.
Thank you
Cheers
|
Kevin_in_GA
4,599 posts
msg #112388
- Ignore Kevin_in_GA |
3/22/2013 4:14:08 PM
Why not back test it and let us know?
|
Xglassm
5 posts
msg #112393
- Ignore Xglassm |
3/22/2013 10:47:39 PM
Hey Kevin, sorry I don't think I was very clear with my question. I ultimately wanted to know if a filter's win rate applies to every buy signal or only the first buy triggered after an exit. For example, if one were to enter today based on the SMI filter, would they still have a 91.4% chance of winning?
|
Kevin_in_GA
4,599 posts
msg #112409
- Ignore Kevin_in_GA |
3/24/2013 9:25:16 PM
Pretty close, but the absolute number of trades is lower so the performance will never be exactly the same.
This was pretty easy to do in Stratasearch - I just added the requirement that the specific divergence NOT be active when the exit triggers. The results for the period 1/2/2007 until 3/14/2013 are:
Number of trades = 59
Win percentage = 89.83%
Average return per trade = 2.02%
Average days held = 11
Annualized Sharpe ratio = 1.838
Pretty impressive.
|
jackmack
334 posts
msg #112421
- Ignore jackmack |
3/25/2013 12:19:47 PM
Kevin
I meant NO disrespect in my previous question about prices moving below the MA's.
I wish I knew how to back test that but that is above my ability level as I am still trying to figure many of the particulars
of just getting a filter to work.
Thank you
|
Kevin_in_GA
4,599 posts
msg #112424
- Ignore Kevin_in_GA |
3/25/2013 12:47:07 PM
Jackmack: None taken. The comment was to gently chide you into expanding your skills. The SF code for both entries and exits is in the first post on this thread - all you would need to do is add in your thoughts on MAs, then backtest using SF. Give it a shot.
|
Xglassm
5 posts
msg #112464
- Ignore Xglassm |
3/26/2013 9:52:37 PM
Interesting stuff thanks! I might be being a big dummy here, but I was wondering if there is a way to get more trades by buying every time a filter returns a “buy” signal, regardless of whether that filter is already in a trade. Then exiting all of a filter’s positions once it receives an exit. I tried to calculate that win rate by turning “allow multiple positions per symbol” on in advanced options, but unfortunately I’m unable to backtest a large amount of data so I’m having trouble seeing how it chalks up to the current “buy only one position per filter and hold until exit” win rates on the front page.
|
Kevin_in_GA
4,599 posts
msg #112466
- Ignore Kevin_in_GA |
3/26/2013 10:59:02 PM
My SS system triggered an exit today for all 10 trades currently in play, and a new BUY for nine of them. Based on the current thinking one would exit only the CMF divergence trade, at a profit of 1.45%.
The SF filter code does not show an exit. This could be because it has to be based on SPY, where SS can use indices in the signaling. That's why I posted this info, in case folks choose to act.
|
jackmack
334 posts
msg #112490
- Ignore jackmack
modified |
3/28/2013 9:55:49 AM
I know it is not to be traded during market hours but I am seeing exit signal now and first time in a while.
Cheers
Check that - and this is why it is not to be run during market hours - exit signal now off ;-)
|