Dump it Here

[Gringotts Bank]

Gringotts - could you advise the URL for this? I can't see it mentioned anywhere....

I've been looking through emails and bookmaks and can't find it. Sorry. The email was from futures.io, so Mike may know.

 

[MovingAverage]

So quiet around here these days...I guess all the breakout traders are busy processing their sell orders?

 

[DaveTrade]

So quiet around here these days...I guess all the breakout traders are busy processing their sell orders?

Most markets are going down right now, this is one big advantage of trading options, you can trade long and short, just saying. I suppose a lot of long-only traders are in house keeping mode at the moment.

 

[MovingAverage]

Most markets are going down right now, this is one big advantage of trading options, you can trade long and short, just saying. I suppose a lot of long-only traders are in house keeping mode at the moment.

Nothing wrong with being long (if your timeframes are short enough)...I think it's more whether you're a breakout trader with longer hold times...those less seasoned would be questioning their strategies right now. Or maybe not as some will say ignorance is bliss.

 

[farmerge]

So quiet around here these days...I guess all the breakout traders are busy processing their sell orders?

Hmm I was working all day untill 6.30 only got to 37 deg, but managed to keep an eye (when the sweat wasn't running into them) on the market. Happy to have got some PLS.

 

[MovingAverage]

Asked ChatGPT to give me AFL code for an RSI based swing system and this is what it gave me. Far from perfect but a reasonable starting point.

 

[DaveTrade]

@MovingAverage I created a template with these signals to have a look. Here is a chart of $SPX with these signals;

 

[MovingAverage]

@MovingAverage I created a template with these signals to have a look. Here is a chart of $SPX with these signals;

View attachment 153802

Nice, thanks for that. Was planning to dump the ChatGPT code into AB (as is with no changes) and see how it performs with some backtesting across the all ords--will post up the results.

 

[Captain_Chaza]

@MovingAverage I created a template with these signals to have a look. Here is a chart of $SPX with these signals;

View attachment 153802

Am I miss reading the charts
ie: Every trade was a loss?

 

[DaveTrade]

@MovingAverage I haven't put the stop-loss on the chart because I'm not sure what it should be, your code was
param("Stop Loss %", 2.5, 0.5, 10, 0.5) but not sure how to read this.

Is the the Stop Loss 2.5% ? and What is the exit ?

 

[Richard Dale]

Perhaps consult the manual: http://www.amibroker.com/guide/afl/param.html

 

[Gringotts Bank]

Nice, thanks for that. Was planning to dump the ChatGPT code into AB (as is with no changes) and see how it performs with some backtesting across the all ords--will post up the results.

Can you ask it to add/subtract some rules that would increase profitbility?

 

[MovingAverage]

Can you ask it to add/subtract some rules that would increase profitbility?

Funny you should mention that.
I've been playing around with ChatGPT this arvo some more. The code I included above was a generally enquiry--I simply asked it to give me AFL code for an RSI swing based system and provided no other specific requirement. Howver, I gave it more specific requests such as:
a) the RSI swing system that only trades stocks with certain daily turnover;
b) the RSI swing system that would give me a specific sharp ratio;
c) the RSI swing system that had a specific drawdown;
d) the RSI swing system that on average generated 5% profit per trade;
e) the RSI swing system that had on average a 30% exposure rate;
f) the RSI swing system that provides 15% CAGR since 2018.
When I gave ChatGPT the more specific requests it actually generated refinements to the original AFL code it gave me originally (which I posted above). The refinements look pretty credible but need some work obviously. Not suggesting it is perfect out of the box code but it is not bad and as a starting point for someone familiar with AFL it seems like it could save you a little time. I'm in the middle of backtesting all the code it has given me to see how good it is. Will be posting results in next few days.

 

[MovingAverage]

@MovingAverage I haven't put the stop-loss on the chart because I'm not sure what it should be, your code was
param("Stop Loss %", 2.5, 0.5, 10, 0.5) but not sure how to read this.

Is the the Stop Loss 2.5% ? and What is the exit ?

yes it is 2.5%. you need to enable the AB's stop loss in the settings for this to work. 2.5% is a bit tight for a lot of stocks on the ASX

 

[MovingAverage]

Am I miss reading the charts
ie: Every trade was a loss?

No you're not miss-reading, but Dave is just charting $SPX...not exactly good example for ChatCPT's swing code.

 

[DaveTrade]

Funny you should mention that.
I've been playing around with ChatGPT this arvo some more. The code I included above was a generally enquiry--I simply asked it to give me AFL code for an RSI swing based system and provided no other specific requirement. Howver, I gave it more specific requests such as:
a) the RSI swing system that only trades stocks with certain daily turnover;
b) the RSI swing system that would give me a specific sharp ratio;
c) the RSI swing system that had a specific drawdown;
d) the RSI swing system that on average generated 5% profit per trade;
e) the RSI swing system that had on average a 30% exposure rate;
f) the RSI swing system that provides 15% CAGR since 2018.
When I gave ChatGPT the more specific requests it actually generated refinements to the original AFL code it gave me originally (which I posted above). The refinements look pretty credible but need some work obviously. Not suggesting it is perfect out of the box code but it is not bad and as a starting point for someone familiar with AFL it seems like it could save you a little time. I'm in the middle of backtesting all the code it has given me to see how good it is. Will be posting results in next few days.

@MovingAverage be careful of only coding things that would be known before entering a trade or your backtest result will be useless. For instance if you code for 5% profit then the cart is before the horse because the profit made is not know until after the trade is complete, so you would need to code for factors may lead to profit.

 

[MovingAverage]

@MovingAverage be careful of only coding things that would be known before entering a trade or your backtest result will be useless. For instance if you code for 5% profit then the cart is before the horse because the profit made is not know until after the trade is complete, so you would need to code for factors may lead to profit.

I was more interested in seeing how ChatGPT would respond to the request. In response ChatGPT included a few lines of AFL that were a profit exit. Same way ChatGPT included a stop loss.
Not sure I follow your logic on "coding stuff before entering a trade". There is absolutely nothing wrong with exiting a position if your price target is hit and this will not make backtest test results useless. It is the same logic as a predefines stop loss--exit if stock price dips below $x. Profit exit is the same just in reverse--exit if stock price closes above $y. Coding this into AFL will not impact validity of backtests. Maybe I am misinterpreting your comments.

 

[DaveTrade]

I was more interested in seeing how ChatGPT would respond to the request. In response ChatGPT included a few lines of AFL that were a profit exit. Same way ChatGPT included a stop loss.
Not sure I follow your logic on "coding stuff before entering a trade". There is absolutely nothing wrong with exiting a position if your price target is hit and this will not make backtest test results useless. It is the same logic as a predefines stop loss--exit if stock price dips below $x. Profit exit is the same just in reverse--exit if stock price closes above $y. Coding this into AFL will not impact validity of backtests. Maybe I am misinterpreting your comments.

Yes I agree with what you are saying here, I was only warning against using results in a backtest.

 

[Newt]

Funny you should mention that.
I've been playing around with ChatGPT this arvo some more. The code I included above was a generally enquiry--I simply asked it to give me AFL code for an RSI swing based system and provided no other specific requirement. Howver, I gave it more specific requests such as:
a) the RSI swing system that only trades stocks with certain daily turnover;
b) the RSI swing system that would give me a specific sharp ratio;
c) the RSI swing system that had a specific drawdown;
d) the RSI swing system that on average generated 5% profit per trade;
e) the RSI swing system that had on average a 30% exposure rate;
f) the RSI swing system that provides 15% CAGR since 2018.
When I gave ChatGPT the more specific requests it actually generated refinements to the original AFL code it gave me originally (which I posted above). The refinements look pretty credible but need some work obviously. Not suggesting it is perfect out of the box code but it is not bad and as a starting point for someone familiar with AFL it seems like it could save you a little time. I'm in the middle of backtesting all the code it has given me to see how good it is. Will be posting results in next few days.

Holy Frack!!

 

[Gringotts Bank]

I was more interested in seeing how ChatGPT would respond to the request. In response ChatGPT included a few lines of AFL that were a profit exit. Same way ChatGPT included a stop loss.
Not sure I follow your logic on "coding stuff before entering a trade". There is absolutely nothing wrong with exiting a position if your price target is hit and this will not make backtest test results useless. It is the same logic as a predefines stop loss--exit if stock price dips below $x. Profit exit is the same just in reverse--exit if stock price closes above $y. Coding this into AFL will not impact validity of backtests. Maybe I am misinterpreting your comments.

Maybe ask DAN to play the part of Jim Simons, and create a system from scratch.