Systematic FX: Specific / Advanced Topics

Hi,

you may know me from this famous thread here, in which I exposed the actual market mechanics (incl. manipulation) and in which I demonstrated and taught my way of approaching the FX markets manually (albeit overall I only showed ca. 12% of the market mechanics):
https://www.forexfactory.com/showthread.php?t=768090


Within THIS thread however, I would like to do something quite DIFFERENT.

Firstly, I would like to discuss the topic of "SYSTEMATIC FX" ONLY (so NO manual trading topics).

The discussion here should be LIMITED to
- automated
- statistical
- fully mechanical
trading systems ONLY.


Topics discussed here should move around rather ADVANCED topics in that context, such as:

- how to create and implement an efficient, dynamic stop rule
- how to manage a system across different currency pairs (correlations)
- how to manage a portfolio of different systems
- etc.

In contrast to the other thread, the rules here will be rather loose. That means, you are free to post own topics here that fit the bill. I will NOT be very strict in this thread and I will neither work a lot on "keeping the flow" of the thread. STILL, please do not go off-topic too much.

Stay safe,
FF

FF,

Are you planning on disclosing a system with clear cut rules? If not, this might not be the right section and will be moved by the admins. Look into "Trading Discussion or Journals"

Thanks

Hi,

yes, if you want Admins, this thread can be moved elsewhere, it is fine with me.

FF

Interesting topic. I believe first of all the genre of a system has to be selected. Example: Trend Following or Mean Reverting. Correlations can be used to size your positions accordingly. There are a bunch of other items to thin kabout.

I would like to do more here, but unfortunately my time is limited and I cannot do everything at the same time.

But maybe in the future there will be a moment where I can post some interesting content here.

Stay safe,
FF

PS: Feel free to post here already!

Topic 1: When does your auto system die? / What to do...

You have a proven, fully automated system.

It passed the multi-year backtests (tickdata, var. spread, slippage, commissions). Good relationship between return & DD etc.

Robust logic, balanced parameters tested under various conditions etc.

Like the one I posted here: https://www.forexfactory.com/showthread.php?t=810051

Here are some of my experiences / thoughts / tips:


- The FX market changed a lot as compared to 15 years ago. The reason is the SM market manipulation I exposed in my big thread.

That implies that if your BTs show a better performance for recent years, that is a good thing.

- Also, if your system performs not as good on the data from 8 years ago, that is not necessarily a bad thing.

- These days, the SM algos change the "market conditions" (such as ranges etc.) every 3 / 4 months........that interval used to be 12 months, then 6 months...

So they are changing "quicker" these days........

- Your system should have a logic that is rather "timeless". That means, that it trades "sth." that will "always be there".



NOW:


- Still, there is NO GUARANTEE that your auto system will perform well in the next 3 years.

- Even more problematic, it is also possible that despite all the preparation, the system even "completely dies" at some point (meaning that it leads to a DISPROPORTIONATE DrawDown).

- This is the stuff that keeps auto traders / quants awake at night: Where is that "cutoff point"?

In my career, I have come across various approaches. Here some examples:


A) If the equity curve breaches the "account trendline" significantly.
B) If the system experiences 1.5x or 2.0x the max. DD.
C) If parts of the system simply do "not work" anymore.

Stay safe.
FF

This is my cup of tea! Keep it coming FF. For cutoff, I do a tiered %MaxDD multi-prong cutoff structure based on where my equity is in comparison to TOTALDEPOSITS. Initial is 3 rounds of 10% each hard-cutoff which I refer to as the "uncomfortable" phase as I am risking my own capital. Once I reach 30-40% of TOTALDEPOSITS, I have a hard cutoff when account equity = totaldeposits and a floating 3-step cutoff between equity-deposits.

hard cutoff means close-all and stop trading, take a break, re-assess, analyse the market condition and trade stat, what went wrong etc... many don't realize but automated trading is ALSO hard work... perhaps even harder than a trader coming in and making one trade at a time.

Once my automation went live (little bit last year and then in April of this year) I have been working non stop thinking about this shit to the point that is making me wonder at times if it is really worth the time and effort :-(

"many don't realize but automated trading is ALSO hard work... perhaps even harder than a trader coming in and making one trade at a time."

Yes, that is true.

FF

For that reason I am having serious doubts about my automation... speaking frankly. My entries are all automated and diversified across 28 pairs trading round the clock but my exits are partially automated requiring me to monitor and adjust the parameters several times which is also fine. It's the part of "I have my thinking and monitoring cap always on" about trading which I am not liking very much. I thought automation would free me to good better things in life and I ain't liking the round the clock effort.

Plus my broker is skillfully making frequent "Not Connected" episodes to mess with my time-sensitive executions during London while I am sleeping. Pisses me off that this (MT4) is not an environment for anyone to make any decent long term consistent gains :-(

"I thought automation would free me to good better things in life and I ain't liking the round the clock effort."


So if you do not automate the exits (the most crucial part of an EA), then the system is not automated but only semi-auto.

Thus, you can neither properly test it / not be sure whether it will actually work.

Keep that in mind, dear.

But yes, even if you fully automate a strategy and things such as news etc. are accounted for, the trader's brain is still required.

Apart from the fact that research / improvements / the quest for additional systems never ends.

;-)

FF

"Plus my broker is skillfully making frequent "Not Connected" episodes to mess with my time-sensitive executions during London while I am sleeping. Pisses me off that this (MT4) is not an environment for anyone to make any decent long term consistent gains :-("


Yes,

unfortunately they do that apart from other tricks.

FF

Agree.... let me clarify about exits... every aspect is coded in the EA and driven by parameters that I control. there is no manual invention with the trades and I could fully automate it but as a thick head that I am, I embarked my journey to rule out price action or indicators or measurement of price from all aspects of my strategy. I don't measure High or Low or range or distance or strength or correlations... not in my entries and certainly not in my exits. I have eliminated as much variabilities from my strategy as possible and that includes all the 'after-effects' that price causes.

Example: Risk based position sizing is typically done on AccountEquity or AccountBalance or something inbetween.... that will ALWAYS be affected by Price. My solution: I do not use them at all. instead, I actively track PEAKEQUITY in real time and use %risk based on PeakEquity. PeakEquity becomes a constant for almost 90+% of the time which results in consistency... and faster recovery from drawdowns if you know what I mean.

% based PnL Exits is also based on a constant i.e. TOTAL DEPOSITS or PeakEquity and not floating equity or balance. See my thinking? My strategy simply cannot be backtested. I have gone thru atleast 100 iterations in demo/forward testing to make this work.... and still I have not happy with the overall consistency because it is combination of both trend following and mean reversion to make it a market neutral endeavor.. not easy at all requiring some level of intervention from a mind that understands and actively keeps track of overall market sentiment.

EDIT: Sharing this because I know you will understand. My entries or "signature" (Mandlebrot?) are universal in nature constantly generating high volume signals ready for me to take. Each signature could be pure trend following(Breakout) for growth or reversal(bounce) for income or pullback/continuation/consolidation (for scalping/HFT) which I no longer trade.

Well VEEFX you are the only person that I have seen in FF that trades 28 pairs. I have 2 VPS and trade less than half the pairs you trade. I'm sure monitoring 28 pairs is hard work. I don't agree with you that MT4 is not good to make consistent profit. I tried many platforms and, at least for me, MT4 is way better because you have tons of available EAs, indicators, systems, etc. that work on that platform. All the other platforms I tried are very limited. I think your problem is your broker and if it's the case get a better one. I wonder how you can get good restful sleep monitoring your trades. Good luck with your trading and never give up.