Systematic Swing trading / backtest

[sysswing]

Hello,

I am working as a quant in the financial markets, I have a background in computer science, big data, machine learning, artificial intelligence.

I have a personal systematic swing trading project. I'm automating and backtesting a strategy based on a cup and handle pattern, i.e. buy on resistance breakout, on a daily basis for now.

I'm making good progress, I have advanced detection of beautiful patterns and the first results are interesting. See image of a trade made automatically, signal: S, entry: ▲, exit: ▼, stoploss in red, rebound on resistance: blue stars.

But unfortunately, I haven't followed any course/training and I have no trading experience on this strategy, so I'm looking for people who are interested in this topic. I still have a lot of things to refine: money management, entry of position, exit of position, management of stops. On each of these points there are plenty of ways to proceed.

So, if you are interested in this pattern type, or want to verify a strategy, I can verify it with a backtest over 40 years, and even then on different markets and different timeframes.

Example of points that I will soon address / refine:

1. What do you thing about selling half the position when you reach R1?
2. How many open trades in parallel?
3. What if all the capital is invested and a new opportunity comes up?

[dmedin]

On 07/10/2020 at 20:08, sysswing said:

I have a background in computer science, big data, machine learning, artificial intelligence.

 

In that case I would advise getting a real job

[Joeeeeyy]

Why don't you try to automate it in RealPro Time. It is free for a Demo account

[Bopperz]

Hi Syswing,

This is good work, I have tried writing code that picks up visual patterns before and it was very hard! Once you have this bit sorted, the other parts should follow fairly easily.

As you have a background in computer science, you know how to write you code in "interchangeable blocks" etc etc. So I suggest keeping it REALLY simple, then updating as you go along. Every part of the trading strategy is important, but imo you will struggle to turn a 60% loss into a winning strategy by tweaking the other parts. So make sure your core idea (entry and exit) is a winner.

1) Exit point: You will want to try a variety of exit points, and maybe a combination based on level and time. For a first attempt, set your sell point at 2x your risk. You will need to do as much work on your exits as you do on your entry.

2) This is linked to total capital usage. Go with 4-6 open trades. If you are following several markets, consider only taking the strongest signals.

3) Good rule of thumb, MAX 2% equity at risk per trade, with 6% at any one time. As trades earn money, consider moving the stop loss up to lock in the PnL, which allows you to open more trades. For a first attempt, just have three trades at 2% equity, and build it up from there.

[jlz]

On 07/10/2020 at 20:08, sysswing said:

Hello,

I am working as a quant in the financial markets, I have a background in computer science, big data, machine learning, artificial intelligence.

I have a personal systematic swing trading project. I'm automating and backtesting a strategy based on a cup and handle pattern, i.e. buy on resistance breakout, on a daily basis for now.

 

A quant that knows about the 3 hardest topics within computer science tries to use a cup and handle pattern. Really?

This is how a quant would ask a question:

https://quant.stackexchange.com/questions/58604/has-a-closed-form-formula-for-the-collateral-choice-option-been-found

 

[dmedin]

50 minutes ago, jlz said:

A quant that knows about the 3 hardest topics within computer science tries to use a cup and handle pattern. Really?

This is how a quant would ask a question:

https://quant.stackexchange.com/questions/58604/has-a-closed-form-formula-for-the-collateral-choice-option-been-found

 

 

 

 

Why?

[Bopperz]

1 hour ago, jlz said:

A quant that knows about the 3 hardest topics within computer science tries to use a cup and handle pattern. Really?

This is how a quant would ask a question:

https://quant.stackexchange.com/questions/58604/has-a-closed-form-formula-for-the-collateral-choice-option-been-found

 

Maybe he/she is not actually a quant. A quant would probably know the answer to the questions they had asked. And don't even get started about the compliance aspects! 

[Kodiak]

"“Why don’t we have enough teachers of math and science in the public schools?  One answer is, well, if they knew the subject well, they’d also know enough to work for Google or Goldman Sachs or God knows where.”"

 

https://millyuns.com/quotes-simons/