Confirmation on Strategy and Programming Planning

SimpleMeLike

SimpleMeLike

Hello,

So I spent about 5-7 months manually backtesting and forward testing a strategy that was not very profitable and just did not know what I was doing.

Good news: through the process I gained alot of chart screen time and discovered some other ideas I would like to test. However, I will still manually backtest but not as much.

After a bit of research and reading threads here at ET forums. I now have a new plan going forward.

Any comments and confirmations please on the following plan going forward:

1. Invest in my learning. Bought https://www.amazon.com/Kevin-Davey/e/B00J34ZNPQ Building Winning Algorithmic Trading Systems by Kevin Davey. Plan to buy a few other books recommend by other on ET.

2. Logically write down simple ideas/strategies on paper or Pseudocode etc in plain English. Continue reviewing the charts daily for new ideas.

3. Manual Backtest these ideas for about +50-100 trades and collect the correct data.

4. If the trades are profitable, proceed to step 7, else back to step 6.

5. Find a platform to program and backtest the strategy with historical data. So far I'm leaning towards NinjaTrader. its free trial as well.

6. Invest in my learning. https://www.udemy.com/csharpninjatrading/ along with a paid #C programming learning course and a book. Its been 7 years since #C or MathLab programming, should be fun. At my stage, I do not want to pay someone to code for me. I think its best I learn myself for now first.

7. Program the strategy, backtest it, and review performance.

8. If profitable, paper trade it one contract while reviewing the system daily for flaws.

9. Trade with small real money. 1 contract.

What do you think of my plan going forward? I welcome all feedback and recommendations.
 
Sounds more thought out and structured than most others here on this site who expressed similar goals. I would say go for it. You may want to, however, consider other alternatives as backtesting platform than ninja trader.

SimpleMeLike said:
Hello,

So I spent about 5-7 months manually backtesting and forward testing a strategy that was not very profitable and just did not know what I was doing.

Good news: through the process I gained alot of chart screen time and discovered some other ideas I would like to test. However, I will still manually backtest but not as much.

After a bit of research and reading threads here at ET forums. I now have a new plan going forward.

Any comments and confirmations please on the following plan going forward:

1. Invest in my learning. Bought https://www.amazon.com/Kevin-Davey/e/B00J34ZNPQ Building Winning Algorithmic Trading Systems by Kevin Davey. Plan to buy a few other books recommend by other on ET.

2. Logically write down simple ideas/strategies on paper or Pseudocode etc in plain English. Continue reviewing the charts daily for new ideas.

3. Manual Backtest these ideas for about +50-100 trades and collect the correct data.

4. If the trades are profitable, proceed to step 7, else back to step 6.

5. Find a platform to program and backtest the strategy with historical data. So far I'm leaning towards NinjaTrader. its free trial as well.

6. Invest in my learning. https://www.udemy.com/csharpninjatrading/ along with a paid #C programming learning course and a book. Its been 7 years since #C or MathLab programming, should be fun. At my stage, I do not want to pay someone to code for me. I think its best I learn myself for now first.

7. Program the strategy, backtest it, and review performance.

8. If profitable, paper trade it one contract while reviewing the system daily for flaws.

9. Trade with small real money. 1 contract.

What do you think of my plan going forward? I welcome all feedback and recommendations.
More...
 
Zzzz1 said:
Sounds more thought out and structured than most others here on this site who expressed similar goals. I would say go for it. You may want to, however, consider other alternatives as backtesting platform than ninja trader.
More...

Thank you Zzzz1 for confirmation,

Yes, I am still reviewing backtesting and automation platforms. I really do not know (i am still researching, etc.) how to pick a platform, I am only custom to ThinkorSwim, but they have no automated trading.
 
My advice is to not throw away a system just because it had a few losing months. Look at success on a yearly basis. People who are not properly funded are much more likely to abandon good systems. When I was younger I never had much money and I threw away good systems due to lack of funding.

Also don't get hung up on the system being "robust". Markets vary greatly, Corn Futures look vastly different than Crude Oil. You know at 9:30 AM CST every Wednesday that there will be massive volatility in Crude. The point is that a successful system does not have to work in every market.

Since Support & Resistance is important to you and since those lines are very subjective then you must define a personal set of rules that is concrete and can be coded. If nothing off the shelf does the job for you then learn about Bresenham's Line Algorithm and how to measure variation in X cords & Y cords to accurately draw these lines.

One last thing, don't get hung up on directional accuracy. You will probably never get much better than 50% accuracy as the sample size increases. You are not a fund manager so don't bother yourself in measuring against S&P 500 returns. All that matters is if you make money or not and can stay in the game.
 
Agree with most except the last paragraph. While I am not a fan of benchmarking against stock indexes, one should pick an appropriate benchmark which should function as hurdle rate. When you make 5 dollars net net the first year then you can pad yourself on the back and be happy, but you performed worse than even the lowest paid janitors. Pick a benchmark that you must beat in order to stay with this endeavor.

MoneyMatthew said:
My advice is to not throw away a system just because it had a few losing months. Look at success on a yearly basis. People who are not properly funded are much more likely to abandon good systems. When I was younger I never had much money and I threw away good systems due to lack of funding.

Also don't get hung up on the system being "robust". Markets vary greatly, Corn Futures look vastly different than Crude Oil. You know at 9:30 AM CST every Wednesday that there will be massive volatility in Crude. The point is that a successful system does not have to work in every market.

Since Support & Resistance is important to you and since those lines are very subjective then you must define a personal set of rules that is concrete and can be coded. If nothing off the shelf does the job for you then learn about Bresenham's Line Algorithm and how to measure variation in X cords & Y cords to accurately draw these lines.

One last thing, don't get hung up on directional accuracy. You will probably never get much better than 50% accuracy as the sample size increases. You are not a fund manager so don't bother yourself in measuring against S&P 500 returns. All that matters is if you make money or not and can stay in the game.
More...
 
Zzzz1 said:
Agree with most except the last paragraph. While I am not a fan of benchmarking against stock indexes, one should pick an appropriate benchmark which should function as hurdle rate. When you make 5 dollars net net the first year then you can pad yourself on the back and be happy, but you performed worse than even the lowest paid janitors. Pick a benchmark that you must beat in order to stay with this endeavor.
More...
The reality is that he should set his expectations low. He is a day trader to my knowledge and will not be holding positions overnight. His success will be primarily due to leverage. If I buy the SPY on margin then my "gains in my pocket" are now twice compared to SPY in a cash account.

The only measure of success is if you can pay your bills and live your life by your personal definition of success. If you blow up your account and can't afford to live then it is time to go get a job. A fund manager has a much higher standard of "success".

To the OP, I could probably write a Function for you if I am in a good mood and compile it as a stand-alone .EXE to determine S&R. You can shell out to this .exe and communicate with it through a .txt file. You will need to be a good enough programmer to parse delimiters like a .csv file. This is really basic stuff but if you are a beginner you won't know how to do it. You will need to know really basic string manipulation.

Learn to output OHLC data to a text file and then learn to load that data back into an array. If you are working on a platform that does not provide APIs then learn and use this free scripting language to automate the UI.
https://autohotkey.com/docs/AutoHotkey.htm

Anything can be automated even without an API. This is not the most professional way to approach things and Zzzz1 would probably laugh at these methods but it gets the job done. The goal is to find and build working solutions to fit your limited budget.
 
Man, stay out of P&R, you make much more sense here. :D Good points made. Thumbs up.

MoneyMatthew said:
The reality is that he should set his expectations low. He is a day trader to my knowledge and will not be holding positions overnight. His success will be primarily due to leverage. If I buy the SPY on margin then my "gains in my pocket" are now twice compared to SPY in a cash account.

The only measure of success is if you can pay your bills and live your life by your personal definition of success. If you blow up your account and can't afford to live then it is time to go get a job. A fund manager has a much higher standard of "success".

To the OP, I could probably write a Function for you if I am in a good mood and compile it as a stand-alone .EXE to determine S&R. You can shell out to this .exe and communicate with it through a .txt file. You will need to be a good enough programmer to parse delimiters like a .csv file. This is really basic stuff but if you are a beginner you won't know how to do it. You will need to know really basic string manipulation.

Learn to output OHLC data to a text file and then learn to load that data back into an array. If you are working on a platform that does not provide APIs then learn and use this free scripting language to automate the UI.
https://autohotkey.com/docs/AutoHotkey.htm

Anything can be automated even without an API. This is not the most professional way to approach things and Zzzz1 would probably laugh at these methods but it gets the job done. The goal is to find and build working solutions to fit your limited budget.
More...
 
MoneyMatthew said:
My advice is to not throw away a system just because it had a few losing months. Look at success on a yearly basis. People who are not properly funded are much more likely to abandon good systems. When I was younger I never had much money and I threw away good systems due to lack of funding.

Also don't get hung up on the system being "robust". Markets vary greatly, Corn Futures look vastly different than Crude Oil. You know at 9:30 AM CST every Wednesday that there will be massive volatility in Crude. The point is that a successful system does not have to work in every market.

Since Support & Resistance is important to you and since those lines are very subjective then you must define a personal set of rules that is concrete and can be coded. If nothing off the shelf does the job for you then learn about Bresenham's Line Algorithm and how to measure variation in X cords & Y cords to accurately draw these lines.

One last thing, don't get hung up on directional accuracy. You will probably never get much better than 50% accuracy as the sample size increases. You are not a fund manager so don't bother yourself in measuring against S&P 500 returns. All that matters is if you make money or not and can stay in the game.
More...

Thank you MoneyMatthew,

Really appreciate the response.

I will respond to your comments.

MoneyMatthew said:
My advice is to not throw away a system just because it had a few losing months. Look at success on a yearly basis. People who are not properly funded are much more likely to abandon good systems. When I was younger I never had much money and I threw away good systems due to lack of funding.
More...

I am still discretionary trading this system and documenting the results accordingly. I believe the system will work if I simplified it a bit. Yes, I trade resistance and supports.

Do you have any advice or recommendations, book, article about stats to record when backtesting a strategy? For instance, I was not aware I need to calculate the Maximum Favorable Excursion and Maximum Adverse Excursion for each trade. I just assumed a 2 points stop loss for each trade. For allllll those trades I took, I never calculated max profit could have made or max loss while in trade or max loss before price turns in my favor.

So I need to learn what data to collect while I am backtesting.

So, no I am not going all out fully programming and automating every idea. I am still daily discretionary trading in paper my system and recording notes. The plan above is just my idea path forward.
 
You must log in or register to reply here.
Back
Top