Trade another day through Systematic Trading | Forex Factory

Options

Go to
Permalinks
Print Thread

Similar Threads

Another Day, Another Chair, Another Day in the Office 203 replies

Systematic trading 1,576 replies

Nonparametric Association Measures in Systematic Trading 57 replies

A Systematic Approach to Markets 68 replies

Systematic Trading vs. Discretionary Trading 14 replies

Attachments: Trade another day through Systematic Trading
Exit Attachments

Trade another day through Systematic Trading

  • Poll

  • What type of trader are you?
  •  
  •  
  •  
  •  
  •  
  • #1
  • First Post: Edited Apr 30, 2019 1:23pm Apr 24, 2019 5:32am | Edited Apr 30, 2019 1:23pm
Hello FF comrades, this is my first trading journal thread.

It is a hybrid of trading discussion, trading systems, and journal.

I trade 100% fully systematic. You are welcome to contribute with the general etiquette of FF.

You are welcome to contribute, share your thoughts, ask questions, and discuss related subjects.

Enjoy!

Success Factors in Automated Trading:
=====================================
In order to be successful in automated trading; you will need to have/learn few skills, and need to use tools to assist you in your endeavor. The following are some of those:

1. Programming: absolutely necessary. If you don't know how to program; I advice you to do discretionary trading. You will be wasting your time.
It is not possible to hire programmers to do the job. There are tons of stuff to code; in all aspects of the trading lifecycle: research, pre-trade analysis, execution, post-trade analysis, portfolio construction, optimization, data mining, debugging, bug fixes, data quality

MQ4 / MT4 is a good start.

However; Object Oriented programming is the best. I personally use Java. I know C, C++, C#, but find Java to be the best; because it is *COST EFFECTIVE*
It is much easier to write a program, debug it, deploy it in Java, then in other languages.

2. Data, 100% accurate price data. Errors in the data will cause havocs. In fact majority of my time is spent in making sure the data is correct, and has minimal errors. I even employ error detection on data.
Divide by zero (experience by Vee) is a good example.

3. Trading software. MT4 is still doable, however it is not scalable; and has few latency issues during peak times.

4. Statistical software for research and analysis.
It is always better to use a tool that is available in the market.
I personally use R, and interface to R from java api. All statistical analysis (correlation, stationary tests, anova tests, graphics) are all done there.
Excel is still doable.

5. Methodology for trading lifecycle: market observation, getting trading ideas, testing the edge, writing a strategy, testing it, optimizting it, forward testing, live trading, retiring a strategy when it dies.

Automated trading is an ongoing process. It does not ends. You keep creating strategies and deploying them to live trading, and removing dead ones as you go.

(more to add later)
  • #2
  • Apr 24, 2019 5:36am Apr 24, 2019 5:36am
Why systematic trading, why automate 100% of the trading system?

An Automated strategy; tries to cover all aspects of trading: the edge, risk controls, discipline, entry and exits, market regimes, etc.

The most important of all; is the risk control and discipline; rules to abide by.

I found this youtube video of Lega Braga (hedge fund manager) a very good narration of Systematic trading vs Discretionary trading:

Inserted Video
  • #3
  • Apr 24, 2019 6:02am Apr 24, 2019 6:02am
Great mate!

Subscribed!
  • #4
  • Apr 24, 2019 8:51am Apr 24, 2019 8:51am
I've tried to develop a fully automated A.I.
The project failed miserably.
It made too many mistakes and in the end cost too much time.

I've analyzed deeply what went wrong.
The only good thing was that it surprisingly revealed something that i would call now a real unexpected "edge".

So the approach is still the same: my trading stays purely discretionary but changed (due to this experiment) dramatically in a positive way.
  • #5
  • Apr 24, 2019 9:37am Apr 24, 2019 9:37am
Awesome, looking forward to the exchange of ideas.

Subscribed!
  • #6
  • Apr 24, 2019 5:24pm Apr 24, 2019 5:24pm
Quoting The-Flipper
I've tried to develop a fully automated A.I. The project failed miserably. It made too many mistakes and in the end cost too much time. I've analyzed deeply what went wrong. The only good thing was that it surprisingly revealed something that i would call now a real unexpected "edge". So the approach is still the same: my trading stays purely discretionary but changed (due to this experiment) dramatically in a positive way.
I had a similar experience with AI. it is not worth the efforts. AI is a complex method of looking at data mining.
  • #7
  • Apr 25, 2019 4:54pm Apr 25, 2019 4:54pm
Quoting alphadude
Why systematic trading, why automate 100% of the trading system? An Automated strategy; tries to cover all aspects of trading: the edge, risk controls, discipline, entry and exits, market regimes, etc. The most important of all; is the risk control and discipline; rules to abide by. I found this youtube video of Lega Braga (hedge fund manager) a very good narration of Systematic trading vs Discretionary trading: https://www.youtube.com/watch?v=SVlXQi406vA
few days later, and the only thing i'm able to glean from this thread is that this woman is the hotter version of linda raschke..(sexy accent)

im hoping for some tips/tricks to the trade..don't let this thread die already man
  • #8
  • Apr 25, 2019 5:26pm Apr 25, 2019 5:26pm
Let's rock and roll with innovative IDEAS. It have tried many and they all have flaws in my eyes stopping me from going live so what I am really hoping to kick this thread off is to define a set of success criteria for any algorithmic strategy. what are the key essential ingredients required for any hands-free strategy to work long term?
Staying in my lane...
  • #9
  • Apr 25, 2019 8:01pm Apr 25, 2019 8:01pm
coming soon comrades. plenty of ideas.

listen to that video above few times, it has the principles we need for systematic trading. put on your seat belts.
  • #10
  • Apr 30, 2019 10:14am Apr 30, 2019 10:14am
Quoting alphadude
{quote} I had a similar experience with AI. it is not worth the efforts. AI is a complex method of looking at data mining.
I have a very good trading method in mind and the general concept is pretty much "easy" to grasp.
The problem is, you need extremely skilled programmers and coders to make it work, because it's not built on all the average silly indicators.

But then your're exposed and have revealed the method, that's a huge no-go.

In the end, trading manually achieves the same results, but ofc misses some or more opportunities.
  • #11
  • Apr 30, 2019 10:27am Apr 30, 2019 10:27am
Quoting The-Flipper
{quote} I have a very good trading method in mind and the general concept is pretty much "easy" to grasp. The problem is, you need extremely skilled programmers and coders to make it work, because it's not built on all the average silly indicators. But then your're exposed and have revealed the method, that's a huge no-go. In the end, trading manually achieves the same results, but ofc misses some or more opportunities.
Good point on hiring developers to share your trade secrets. I tried 3 times before I gave up, rolled my sleeves and learnt MQL in 2015. Anyone can literally learn how to code by just study working examples. That is how I did it and how I am doing it now with Sierra charts in Futures market. Where there is will, there is ALWAYS a way brother!

Trading manually is all good if your skills are transferable to your kids. What happens if (God Forbid!) you fall sick/disabled/die or cannot trade for whatever reason? Will your trading income continue or stop? If it stops and if you have not made immense wealth then what have we as traders achieved beyond a life time of grief and misery to make it from trading? Nothing! We as traders add no value to society if we have 1) not made money 2) if we have made money and not contributed to improve the world around us (and those who are under-privileged and very far away from us westerners
Staying in my lane...
  • #12
  • Edited 5:03pm Apr 30, 2019 12:56pm | Edited 5:03pm
Quoting The-Flipper
{quote} I have a very good trading method in mind and the general concept is pretty much "easy" to grasp. The problem is, you need extremely skilled programmers and coders to make it work, because it's not built on all the average silly indicators. But then your're exposed and have revealed the method, that's a huge no-go. In the end, trading manually achieves the same results, but ofc misses some or more opportunities.
if you want to do automated trading; it is better to learn programming yourself. It will be very hard for any trader to hire a programmer; because it will not end there.

you need whole suite of pre-trade, execution, and post trade analysis to achieve success in automation.

you can still use Excel to do research and analysis. It's cumbersome; but doable.

quant hedge funds does not hire anyone who does not know how to code.

regarding this thread; the idea is that we share trading ideas (not necessarily the secret sauces). We test the edges, share the process, learn about new stuff, etc.

you still have to do the hard work to get there.
  • #13
  • Apr 30, 2019 1:23pm Apr 30, 2019 1:23pm
I have udpated few guidelines on automated trading in the first post.
  • #14
  • Apr 30, 2019 2:30pm Apr 30, 2019 2:30pm
Quoting alphadude
2. Data, 100% accurate price data fErrors in the data will cause havocs. In fact majority of my time is spent in making sure the data is correct, and has minimal errors. I even employ error detection on data. Divide by zero (experience by Vee) is a good example. ....
good list... my problem from yesterday with divide by zero is not a data issue. By design, any ZZ indicator will not return valid values during the formation of last leg. I have found a way to circumvent that for most ticks to reduce the probability by increasing precision in my calculations to 8 decimal digits (hence Sierra clocks in tick based testing mode) ... so it's really a logic or calculation problem and not an issue with raw market data. Just thought of giving a clarification.

Regarding retiring strategies, how many have you retired so far after running it in live and for how long were they running and what types of strategies were they ? Hope you can share the specific of your own personal journey too :-
Staying in my lane...
  • #15
  • Edited 3:11pm Apr 30, 2019 2:46pm | Edited 3:11pm
This interview with Kevin Davey is related to this thread ideas. This audio interview is worth a thousand of pictures!

Inserted Video


Inserted Video
  • #16
  • Apr 30, 2019 3:03pm Apr 30, 2019 3:03pm
Quoting VEEFX
{quote} good list... my problem from yesterday with divide by zero is not a data issue. By design, any ZZ indicator will not return valid values during the formation of last leg. I have found a way to circumvent that for most ticks to reduce the probability by increasing precision in my calculations to 8 decimal digits (hence Sierra clocks in tick based testing mode) ... so it's really a logic or calculation problem and not an issue with raw market data. Just thought of giving a clarification. Regarding retiring strategies, how many have you retired...
hi V,

I can't recall exactly how many strategies I retire; but quite a bit. On average; perhaps one strategy is dead every 3 months; or so.
Most dead strategies was traded over a year or more.
Some are not dead 100%; but need improving.

Think of it as Theasurous ship; over time; they replaced every single spare part in the ship. Is the ship still the same ship; or completely different.

Last year; I have shared my journey over an 11 months period; reaching 100% return on the Verbal Dierreia challenge. There are few tidbits there if you revisit the posts from the beginning.

https://www.forexfactory.com/showthr...8#post11670458

During that experiment; I have run few strategies; and in fact; added few ones on the way; and removed dead ones on the way. That performance is the net result of the process.

My advice; is that you make an objective; to find one simple trading strategy that works; and is profitable. trade it live. Focus on this strat until you get it profitable few months.

Then you will automatically progress from there; by researching more trading ideas.
  • #17
  • Apr 30, 2019 3:35pm Apr 30, 2019 3:35pm
Quoting alphadude
{quote} hi V, I can't recall exactly how many strategies I retire; but quite a bit. On average; perhaps one strategy is dead every 3 months; or so. Most dead strategies was traded over a year or more. Some are not dead 100%; but need improving. Think of it as Theasurous ship; over time; they replaced every single spare part in the ship. Is the ship still the same ship; or completely different. Last year; I have shared my journey over an 11 months period; reaching 100% return on the Verbal Dierreia challenge. There are few tidbits there if you revisit...
Thanks. I would love to do a case study on one of your failed strategies if you don't mind sharing the backtesting and live results for a month or so... just to check the variances of which trading stats started to decay and at what rate to trigger a decommission. To me, if a strategy does not survive for more than 3 months, it is NOT worth going the automation route... unless you are trading a basket of strategies across markets. I would be devastated to see my algo fail (let alone fail after 3 months lol) Perhaps I have yet to experience a failed algo. My last one with FX was simply disconnected by unregulated broker which made me give up on spot fx.
Staying in my lane...
  • #18
  • Apr 30, 2019 4:58pm Apr 30, 2019 4:58pm
There are some "AI" platforms out there where you can build a strategy from over or/with 1000 different indicators.

The AI tries to find the absolute best and most profitable algo, and ofc shows the backtest.

I don't know how valid this is, his name is David from "Build Alpha" , he's also on twitter and youtube and ofc the platform is not for free.
  • #19
  • Apr 30, 2019 5:12pm Apr 30, 2019 5:12pm
Quoting The-Flipper
There are some "AI" platforms out there where you can build a strategy from over or/with 1000 different indicators. The AI tries to find the absolute best and most profitable algo, and ofc shows the backtest. I don't know how valid this is, his name is David from "Build Alpha" , he's also on twitter and youtube and ofc the platform is not for free.
indeed; there are few of those AI generators.

I find them useful to get some trading ideas. but to trade their EAs; it will drain your account. They have data mining bias.

Those guys who developed those tools; hit a brick wall at the end of the day. There is no way out; other than hard work; and understanding the market.

The other way to look at it; if it makes money; why sell it.
Attached Image
  • #20
  • Apr 30, 2019 5:20pm Apr 30, 2019 5:20pm
Quoting alphadude
{quote} indeed; there are few of those AI generators. I find them useful to get some trading ideas. but to trade their EAs; it will drain your account. They have data mining bias. Those guys who developed those tools; hit a brick wall at the end of the day. There is no way out; other than hard work; and understanding the market. The other way to look at it; if it makes money; why sell it. {image}
Exactly.
0 traders viewing now