Home › Forums › Trading System Mentor Course Community › Psychology › System Alterations
- This topic is empty.
-
AuthorPosts
-
November 27, 2017 at 6:01 am #101728ScottMcNabParticipant
How does everyone go letting their systems run ?…I find myself always going back and trying a new idea or minor improvement…but by doing so I am never really getting or going to get a long term live track record.
Does anyone else find it hard to leave the system alone once it is live ? Perhaps it would be better to save up the changes all together and introduce them once a year as a new system and then let them run alongside each other for another 3-6 months ?
November 27, 2017 at 7:59 am #108063SaidBitarMember
i totally know what you are talking about. I used to modify the systems all the time then i reached one stage that i can not know if the system is working or not or the improvements are adding any value. then i decided unless the modification are really important i do not do them so if the modifications are going to add a bit to CAR or decrease a bit in MDD i do not do it.November 27, 2017 at 8:41 am #108067JulianCohenParticipantSaid Bitar wrote:
i totally know what you are talking about. I used to modify the systems all the time then i reached one stage that i can not know if the system is working or not or the improvements are adding any value. then i decided unless the modification are really important i do not do them so if the modifications are going to add a bit to CAR or decrease a bit in MDD i do not do it.I’m kind of the same. I just made some small adjustments to one of my systems but I only did it because the selection bias was at 98% so I don’t see a big change to the system as a whole. It added about about 1% CAGR so it was an incremental change. However I know exactly what you mean.
November 27, 2017 at 8:46 am #108068ScottMcNabParticipantSaid Bitar wrote:
i totally know what you are talking about. I used to modify the systems all the time then i reached one stage that i can not know if the system is working or not or the improvements are adding any value.Exactly….Hit nail on the head….I need leave them alone…I know I wont stop tinkering so maybe a rule where changes introduced once a year ?
November 27, 2017 at 4:00 pm #108064LEONARDZIRParticipantWhen I started making significant money I stopped tinkering.
November 30, 2017 at 12:59 pm #108065RobGilesMemberSo I finished the mentoring stage understanding that for short term MR style systems, its prudent to optimise key parameters once every 6 months with a 3 year lookback and longer term momentum style systems once a year looking back over about 5 years. In both cases you’re looking for statistical flat spots in the test results. Is this what the more experienced guys in this forum are doing without considering it “tinkering”?
November 30, 2017 at 8:12 pm #108107ScottMcNabParticipantHi Rob,
I remember Nick discussing this in the group call a couple of months back as I asked him a similar question. I went and tested that night (hard to believe with my tendency to tinkering but until then I had not optimized) and it didn’t seem to have much impact on the MOC MRV I was using. For my rotn momo systems I use a slightly different approach and have several versions with different parameters for each (US market). Its an extra 10-15 minutes work each month but seemed to smooth out those bumpy periods a bit in backtesting.February 13, 2018 at 9:52 am #108066ScottMcNabParticipant“Beginners cycle mistake x+1”
System envy
..may just be me but found myself spending (wasting) time going back and looking at why my system didn’t perform as well as other peoples when really all that matters is that live results are tracking backtest…if they are then the cagr long term should be fine and relative under-performance is likely to be temporary and to be a reflection of different systems suiting slightly different market conditionsFebruary 13, 2018 at 9:56 am #108420JulianCohenParticipantScott McNab wrote:“Beginners cycle mistake x+1”System envy
..may just be me but found myself spending (wasting) time going back and looking at why my system didn’t perform as well as other peoples when really all that matters is that live results are tracking backtest…if they are then the cagr long term should be fine and relative under-performance is likely to be temporary and to be a reflection of different systems suiting slightly different market conditionsI did the same, but I found something that materially affected the results, and so I did make a change.
-
AuthorPosts
- You must be logged in to reply to this topic.