DOW 2h TREND CHASER v4 MM
- This topic has 26 replies, 4 voices, and was last updated 4 years ago by
.
-
Vonasi, you are a veritable fount of cool ideas – thanks for that!
1 user thanked author for this post.
@GraHal I added Paul’s mods to the trailing stop and ran the WF; there’s good news and not so good. It’s rock steady, only 3 losing trades … but not nearly so profitable. For sure I would run it with MM and cumulate orders.
1 user thanked author for this post.
The good news in that last test is that last test is that the OOS line is at the same angle as the IS line. The WF ratio at 78% also looks positive. The bad news is that not all OOS tests were profitable. Another observation is that 83 trades is not very many to give any great degree of confidence in the results.
Some other observations are that during the data that you are testing on the DOW has been pretty much in a bull rally the whole time with only a few sideways movements and one minor dip. It is very easy to create a strategy that makes money on data that is always going up. How would it cope in a market crash when price is behaving in a very different manner. Perhaps you should consider a very loose fuse stop to shut the strategy down once a loss that you are uncomfortable with has occurred. Yes you lose money and quite a lot of money but not as much as you possibly could do if the market carries on down.
There is also every possibility that by using a trailing stop that you have curve fitted the trailing stop values to the data. You need to test with values either side of your ideal settings and graph the results in Excel or similar to see if the strategy is on a trailing stop cliff edge. You should try to do this with any other variables too to try to prove whether it is a curve fit or not. Once again only 83 trades is your enemy when doing this with this strategy.
1 user thanked author for this post.
how to decide which to go with?
Yeah there is no black and white answer in a way! 🙁
The golden rule is to use the set of values for the most recent IS period. This kinda fits in with our re-optimisation MO.
However, I do not always use the most recent IS values … if I judge from the equity curve that there was an exceptional drop, rise or even a prolonged flat period over that most recent IS period.
Reason: WF values would be tuned to an exceptional event period (which could end very soon!).Conversely, if the most recent OOS period (the period for which the profit / loss figure is shown is exceptional due to sharp rise, drop or flat then I might use the most recent WF values.
Reason: The exceptional OOS period may be the reason for low profit or loss, but the corresponding IS period (over which the WF values were formed) was not exceptional.And having said above … I probably flick about from one set of rules to another and more in between! 🙂
WF primarily is to see if … in order to show a profit over OOS periods … does this require big variation in values for variables.
If Yes, then probably the Auto-System is not going to be a big winner,
Also the WF Efficiency Ratio needs to be > 50% ish (but with small variation in variable values between IS periods)!2 users thanked author for this post.
after establishing robustness using a 5 period WF – then do a normal optimisation (not using WF) over the most recent 30k bars or 50k bars?
Yes, that sounds good as a method (I have to have a method or my brain just stalls from having too many possible ways forward). And more in keeping with out experiments in short timeframe optimization… cheers!
1 user thanked author for this post.
Find exclusive trading pro-tools on 
