Newbie needs help automating strategy.

I think you should heed the advice given to you to work on your own psychological issues instead of pursuing automated trading. What you are experiencing is what many experience, you may be able to predict moves with great accuracy but unless you learn to manage your own emotions and risk appropriately you will not be profitable.

Many people think that automated trading is some kind of silver bullet but it is just as complicated, if not more complicated, than manual trading. There are tons of tradeoffs to consider and you not only need to find a highly skilled programmer but also one who is ethical and not interested in stealing your idea. Programming is a very hot commodity right now and this project will be expensive, assuming that your method can even be translated to a program.

I’m not sure what your account size is but this will likely not be worthwhile unless you plan to trade six figures. If doing it on your own still doesn’t interest you, you should consider the advice above of finding a business partner who can execute your plan based on your given rules in a profit split fashion. This may be tough to hear but trading is very complicated no matter how you plan to do it and there is no substitute for discipline and focus.
 
Mike34 said:
In the hard to believe category, I'm exceptional at finding support and resistance and bad at trading them. I have ADD and compulsive behavior disorder and it prohibits me from exercising financial discipline and patience. I'm just looking to automate orders around support and resistance that also factor in some basic market internals into the decision making process. It's a little more complicated than that, but not by much, at least to start. I have zero programming background or knowledge and don't have any real interest in doing this myself. From my research, that's still in the early stages, it sounds like IB would be the best platform to use if you are placing 20-50 smaller trades a day due to their commission structure and technological capabilities. Does that sound right? Also, where do I look to hire somebody? I'm currently on TOS, but I imagine if I have my trading automated on IB, I cannot use the same ago on different platforms.

1- Is IB the best place to automate your trading?

2- Where can I find programmers?

Thanks in advance. From what I've read, people on this forum seem really helpful.
More...

Been down your path long ago, I use to have OCD and other acronyms, then started to go for hypnosis once a month for a year, I still go 1-4 times for reminders, for me, it helped much as manual trading tested my unsteady nerves back then, I still don't care much for manual training as it is very boring after many years of doing so, but if you don't do it-you will lose it cause of age. The problems you will encounter dealing with programmers if you do not know how to program some is constantly going back to programmer cause you programming changes as volatility changes or speed of the market changes. What was a steady market one week has turned tight variance into sloppy price action. Contact a Dean at nearby Community college and ask who his three brightest students are as you want something programmed for financial systems, students need money and they usually most eager to work on it, plus they often will charge for much less, you could have a nice working relation till you have a few different programs going for various studies/models. In course of me working with other programmers, you are always going back for corrections, they are programmers and you have to have the ability to think in programming thoughts and why so tough to get anything done the first time around.

I now have a trusted partner, what I do now is develop a model, then back test over 15 years of data, then trade one lots for 3-6 months so as to work out all the bugs of the system, then send it to my buddy for automation, I am 99% automation but I have others monitor it during the day. What I thought would be easy to do like Head & Shoulders pattern based on angular velocity has turned into a 3 month nightmare of sorts but we getting there. The more defined refinements, math calculations of physics, less back and forth has occurred.

These folks have nifty software. http://www.nebadawn.com/ and free trial I think.
 
Handle123 said:
I have others monitor it during the day.
More...
How did you set that up? Is it someone 100% dedicated? Do they have the ability/knowledge how to stop, start, and possibly close positions if necessary?
 
OP, the hard part about automating a strategy is not the actual decision-making process of when to enter a trade (unless it's some complex chart pattern). It's the infrastructure around that to manage the trade, handle any problems with contract overruns (i.e., out-of-sequence cancel requests and buy/sell orders due to speed of the market, typically solved by a state machine) and any disconnect/reconnect problems which may occur.

There are commercial solutions which can take you part of the way there, providing you the infrastructure I mentioned above, where you are left to focus on just the entry/exit logic. THE PROBLEM with these commercial solutions is that when analysis needs to shift a to tick-by-tick basis for decision-making rather than bar close decisions, their software logic (the part you never have to code) can eat up your CPU to the point that you cannot stay in-sync with the market.

Think of every tick coming into your PC as a real-time loop where you need to analyze it with what you already have for data and your software needs to fully analyze that before the next tick comes in. You know how in games they'll report "frame overruns" to you? Same kinda thing. Your software wasn't fast enough to analyze the real-time info so you got "blown out of a frame" where, in the trading case, it's new data coming in but your software is not ready for the new analysis because your last loop through real-time was too slow.

This is why you see guys on these forums hardly ever recommending something built on top of a charting package. They'll usually code in C++ with no unnecessary overhead between the autotrading software and the real-time data coming in. It's probably okay for strictly bar close decision-making but tends to break down if you want to also look at the tick level in intraday autotrading. Some vendors will try to compensate by doing something called "tick throttling" where the tick-by-tick logic is skipped and the ticks are looked at in like 200-500 "tick chunks" which could be considered as synthetic bar closes inside an actual bar close of like a 1-5 min timeframe.

Anyway, since you don't program, I'm going over your head on some things here but maybe I've opened your eyes a little to the realities of what you want to do.
 
SteveH said:
OP, the hard part about automating a strategy is not the actual decision-making process of when to enter a trade (unless it's some complex chart pattern). It's the infrastructure around that to manage the trade, handle any problems with contract overruns (i.e., out-of-sequence cancel requests and buy/sell orders due to speed of the market, typically solved by a state machine) and any disconnect/reconnect problems which may occur.

There are commercial solutions which can take you part of the way there, providing you the infrastructure I mentioned above, where you are left to focus on just the entry/exit logic. THE PROBLEM with these commercial solutions is that when analysis needs to shift a to tick-by-tick basis for decision-making rather than bar close decisions, their software logic (the part you never have to code) can eat up your CPU to the point that you cannot stay in-sync with the market.

Think of every tick coming into your PC as a real-time loop where you need to analyze it with what you already have for data and your software needs to fully analyze that before the next tick comes in. You know how in games they'll report "frame overruns" to you? Same kinda thing. Your software wasn't fast enough to analyze the real-time info so you got "blown out of a frame" where, in the trading case, it's new data coming in but your software is not ready for the new analysis because your last loop through real-time was too slow.

This is why you see guys on these forums hardly ever recommending something built on top of a charting package. They'll usually code in C++ with no unnecessary overhead between the autotrading software and the real-time data coming in. It's probably okay for strictly bar close decision-making but tends to break down if you want to also look at the tick level in intraday autotrading. Some vendors will try to compensate by doing something called "tick throttling" where the tick-by-tick logic is skipped and the ticks are looked at in like 200-500 "tick chunks" which could be considered as synthetic bar closes inside an actual bar close of like a 1-5 min timeframe.

Anyway, since you don't program, I'm going over your head on some things here but maybe I've opened your eyes a little to the realities of what you want to do.
More...
Good comments. Assuming a commercial platform offers a trial, performance can easily be tested via writing dumb strategies that will simulate actual strategies in terms of data processing. Another thing people don't consider is how is the ATS (automated trading system) going to be monitored and the user notified of any issues. If the user is forced to watch the ATS run live for monitoring then that is a major fail and defeats the purpose of running automated.
 
fan27 said:
If the user is forced to watch the ATS run live for monitoring then that is a major fail and defeats the purpose of running automated.
More...
Respectfully, I disagree. You very well know how many problems can and do come up with automation. The lower the time frame, the closer you must watch. And when do things start breaking? At the worst f@#g possible time! (Like big red days that overwhelm resources)

What automation does allow you to do is to run multiple strategies (or same strategy on multiple symbols) simultaneously. That is huge as most manual traders can only handle managing three to five at a time.
 
userque said:
1 - Probably

2 -
https://www.upwork.com/o/profiles/browse/?q="interactive brokers"&user_pref=1
https://www.guru.com/d/freelancers/q/interactive-brokers/

You'll find, if you haven't already, that the hardest part about internet chat forums is ... getting direct answers to direct questions.

Good Luck!
More...
Thanks man. Nobody had answered that question directly. You wouldn't happen to have any feedback on all those posts from India, would you? There is a universe where someone from India has enough knowledge to automate a strategy for you at a discounted price. For what's it worth, my level one needs for automation are pretty simple.
 
qlai said:
Respectfully, I disagree. You very well know how many problems can and do come up with automation. The lower the time frame, the closer you must watch. And when do things start breaking? At the worst f@#g possible time! (Like big red days that overwhelm resources)

What automation does allow you to do is to run multiple strategies (or same strategy on multiple symbols) simultaneously. That is huge as most manual traders can only handle managing three to five at a time.
More...

So that is exactly what I need to start. I've basically eliminated the idea of setting up anything that has a dozen plus lines on each chart and expect an algo to follow a 100 plus different symbols in part because of the feedback I'm getting on here. My starter idea is to really lock in on 2 or 3 lines on each symbol that offer the greatest odds IMO and have an algo follow a 100 plus different symbols. To your point, more then I can follow on my own. Technically, it would only be when a line was hit that the algo would activate and only when the the line held (working on defining just that) that an order would be placed. I'm a self admitted newbie to this world, but it feels to simple to be hard to do. On a crazy morning, I can't imagine it having to process more then 5-10 orders in any given minute and that may be on high end. Thoughts?
 
qlai said:
Respectfully, I disagree. You very well know how many problems can and do come up with automation. The lower the time frame, the closer you must watch. And when do things start breaking? At the worst f@#g possible time! (Like big red days that overwhelm resources)

What automation does allow you to do is to run multiple strategies (or same strategy on multiple symbols) simultaneously. That is huge as most manual traders can only handle managing three to five at a time.
More...
Why should I watch when a monitoring system can watch 10 times better than me? All my strategies write strategy execution and health check events which in turn get added to AWS CloudWatch. I also have my ATS log data going directly to CloudWatch. I have AWS CloudWatch integrated with my Slack account where I can configure Alarms in CloudWatch and post those alarms to any number of Slack channels. I can control notifications in Slack at a channel level and have real time visibility into my ATS via the Slack app on my phone. This is how modern DevOps works where engineers are not watching a system run but instead are notified of system issues.
 
You must log in or register to reply here.
Back
Top