R Programming for Traders, Helps small traders compete with Hedge Funds

[jj90]

Hello,

Why not just skip all the programming and manual back test a trading idea for about 300 trades based historical data and review the performance metrics and start trading it live?

What advantage will programming a strategy have for me? Why should I spend +10000 hours learning to program? After I'm done learning, what trading ideas to test?

You see where I'm going with this. Explain the advantages programming a trading strategy rather then me just trading it straight up after some manual back testing? What is R anyway?

What if your trading strat is HFT-ish? Or you are doing basket trading? Or multi legged combos? You gonna sit there and click as fast as you can?

Not to be facetious, if your strat doesn't need heavy numeric or computing analysis by all means do it be hand. I've certainly backtested stuff in excel.

BTW, coding is the new literacy. Being able to read/write code is like being able to read and write in the Middle Ages.

 

[SimpleMeLike]

What if your trading strat is HFT-ish? Or you are doing basket trading? Or multi legged combos? You gonna sit there and click as fast as you can?

Not to be facetious, if your strat doesn't need heavy numeric or computing analysis by all means do it be hand. I've certainly backtested stuff in excel.

BTW, coding is the new literacy. Being able to read/write code is like being able to read and write in the Middle Ages.

Thanks jj90 for response,

My strategy is not HFT-ish, maybe 2 to 3 trades a day on average. Its not with numerical mathematically computations. It is simple, if price get above this horizontal line, go long if R:R=>1, vice versa for short. I use one contract.

I manual back test in my spare time. Gives me new ideas as well.

I wouldn't mind programming the method to fully develop a automated system and properly back test it, but I don't use indicators to define trend of the moment. I use my eyes.

Any suggestions?

 

[SimpleMeLike]

Though this looks sarcastic, let me try to answer. R is a programming language. If proven system works it is good to trade manually and if it can be automated thru program more good. You can be at vacation/retirement thru out the year just swiping the cards based on the money earned by the program! Manual back testing will take lots of time and effort. Programming is not that difficult.

Thanks kent,

Sorry if its seems funny, but I was curious on R programming language, I only heard of C programming language with NinjaTrader.

Can resistance and supports be programmed?

 

[kent]

I thought ninja had C++. Anyway, Support/Resistance is the one I believe very iffy and depends on what one sees as support /resistance. If it can be finitely defined, then it can be programmed. But IMHO, S/R is very subjective and it depends on who sees what on the screen/chart. Anything that can be finitely defined/calculated should be able to program..

 

[SimpleMeLike]

Thanks kent,

I would like to expand to coding and automated trading, but I am afraid my experience with trading support and resistance (with no indicators on the screen) is limiting me to one trading strategy that I have invested much time in.

Support and resistances I would like to program use are daily pivots and previous day high, close, open, low.

I was purchasing support and resistance levels from a service for a monthly fee, but I don't use them as much because I don't know how they get these levels. Of course they want share, so I have no way of back testing besides forward testing.

Thanks,

 

[jj90]

Thanks jj90 for response,

My strategy is not HFT-ish, maybe 2 to 3 trades a day on average. Its not with numerical mathematically computations. It is simple, if price get above this horizontal line, go long if R:R=>1, vice versa for short. I use one contract.

I manual back test in my spare time. Gives me new ideas as well.

I wouldn't mind programming the method to fully develop a automated system and properly back test it, but I don't use indicators to define trend of the moment. I use my eyes.

Any suggestions?

I would suggest you take a intro programming class at a local college to learn the basics. Your entry method is definitely codable but for you yourself to do it without know basic data structures, logic, it will be more hassle if you coded it up vs excel.

 

[Murray Ruggiero]

You can program, support resistance levels , momentum , pivots with standard trading languages , like EasyLanguage or TradersStudio Basic, what you can't do is apply machine learning and advance modeling. For example use SVM to do pattern identification for more advanced patterns. Time series forecasting with Arima/Garch or using neural networks. Identify trading regimes with Hidden Markov Models. This is why R and Python are of value. It makes doing these advance things something a normal system developer can attempt.

 

[SimpleMeLike]

I would suggest you take a intro programming class at a local college to learn the basics. Your entry method is definitely codable but for you yourself to do it without know basic data structures, logic, it will be more hassle if you coded it up vs excel.

Thank you jj90,

Yes, I only want to trade resistance and supports in the direction of price moving on the 2 minutes time chart.

Something like a breakout of support or resistance.

Buttttttt here is my biggest concern. I define trend of the moment (meaning I look at 2 min chart, and say price moving up, time for me to wait for resistance to break and get long).

My eyes tell me the trend. I thought about using some EMAS combination to tell me trend for if I need to properly develop an automation around this type of trading.

Will this be a challenge for me for programming?

 

[jj90]

@SimpleMeLike technically speaking, EMAs should calculate based on the completion of a bar/candle/whatever, so it will always lag. That may or may not impact your execution.

if you don't know basic data structures, nested loops, and etc, yeah it's going to be difficult to code it up. Go take the intro programming class.

Alternatively, you could pay a programmer to do so.

 

[SimpleMeLike]

You can program, support resistance levels , momentum , pivots with standard trading languages , like EasyLanguage or TradersStudio Basic, what you can't do is apply machine learning and advance modeling. For example use SVM to do pattern identification for more advanced patterns. Time series forecasting with Arima/Garch or using neural networks. Identify trading regimes with Hidden Markov Models. This is why R and Python are of value. It makes doing these advance things something a normal system developer can attempt.

Thanks Murray,

Well, I want statistical confidence for every trade I take. The Only problem/concern I have right now is that for the strategy I now paper trade, i have limited (about 300 trades) amount of back testing data to confidence say this strategy will make money in the future and all have to do is click buy and sale like robot. No emotional attachment is what I want via manual or automated trading. The 300 trades show good performance metrics.

But is 300 trades enough for me to trade with confidence.

One of the reasons me and few others shy away from automated trading is because noone talks in simple terms. All this advance modeling stuff. I can barely make money manually trading a support and resistance line, all this machine learning stuff.

I reckon all the advance trading people work at Goldman Sachs.

Just the other day I was listen to a webnair with an experienced algorithm system developer talking and he said it normally takes programming and backtesting of 200 trading ideas before finding 1 system to trade live.

Then he said that 1 you find, may stop making money in X amount years.

I'm just a simple minded Texas boy, but when he said that, it made 0 sense to me. And I have an engineer degree, but I enjoy trading as well.

Please explain, why spend time develop system that may not work and was tested for the past 10 years and 2008 to 2009 market crash is within that 10 years?