My Trading Journey

so I was searching on how to build strategies with ai to pinescript code or mql code and i came across this video.
this man made a EA that turned $10k to 10mil like tf ???????
the only thing that came to my mind was if its this easy then anyone would do it cuz like wtf but yeah i mean it did have a procedure nor he shared the code so ig its kind of good but idk whether it would work in real market conditions but heres what i saw

he used deepseek to build the EA on MQL, but he does have experience in writing code himself ( which i dont have any experience about ) and you know even balance equity curves are good like tf
he bactested it on gold for 1 year ( 2025 ) worth of data and he shared the results
but this is the problem, what he used was a indicator called SAR as his basis which is a mathematical formula so it can be expressed to code easier than english

what did i mean by that? lets say for example you want to say your stratergy wants to target nearby swing high / low liquidity pools , you cannot express it to code directly bc its plain english but if I say within this ATR ( x specified Range ) find highs and lows , which still requires OHLC candle data so thsi is what im having a issue with bc i dont use indicators so its kind of hard for me to make it a strategy as i dont know pinescript and mql as a expereinced coder to build them so i have to use AI but i dont know how to explain it again bc i dont know the code basics


Outcome



and his Sharpe ratio is like 24.78 BRO THATS CRAZY ASH LIKE FROM WHAT I KNOW A SHARPE RATIO GREATER THAN 1 IS GOOD and MORE THAN 3 THAT IS LIKE BEST PERFORMANCE , even hedge funds doesnt have that kind of crazy sharpe ratio ( according to perplexity ) but heres what perplexity said

it could be a error or small data set which does make sense bc he only used 2025 which is like only 6 months only so maybe testing for like 3+ years ( according to perplexity ) it would be more suitable and again this is not real world market conditions so yeah but still i mean its good

But my goal is not to have a perfect or a high winrate EA but to make my work easier by making my strategy into code so I can backtest more effiencently so yeah idk how to do it but i wanna do it so yeah lets see how long it will take , and what it will take to find that data

Best wishes and go ahead

Got a update tn
will post soon

Abdul Momen said:

Best wishes and go ahead

Click to expand...

thanks man

terunoz said:

so I was searching on how to build strategies with ai to pinescript code or mql code and i came across this video.
this man made a EA that turned $10k to 10mil like tf ???????
the only thing that came to my mind was if its this easy then anyone would do it cuz like wtf but yeah i mean it did have a procedure nor he shared the code so ig its kind of good but idk whether it would work in real market conditions but heres what i saw

he used deepseek to build the EA on MQL, but he does have experience in writing code himself ( which i dont have any experience about ) and you know even balance equity curves are good like tf
he bactested it on gold for 1 year ( 2025 ) worth of data and he shared the results
but this is the problem, what he used was a indicator called SAR as his basis which is a mathematical formula so it can be expressed to code easier than english

what did i mean by that? lets say for example you want to say your stratergy wants to target nearby swing high / low liquidity pools , you cannot express it to code directly bc its plain english but if I say within this ATR ( x specified Range ) find highs and lows , which still requires OHLC candle data so thsi is what im having a issue with bc i dont use indicators so its kind of hard for me to make it a strategy as i dont know pinescript and mql as a expereinced coder to build them so i have to use AI but i dont know how to explain it again bc i dont know the code basics

View attachment 455991
Outcome
View attachment 455992


and his Sharpe ratio is like 24.78 BRO THATS CRAZY ASH LIKE FROM WHAT I KNOW A SHARPE RATIO GREATER THAN 1 IS GOOD and MORE THAN 3 THAT IS LIKE BEST PERFORMANCE , even hedge funds doesnt have that kind of crazy sharpe ratio ( according to perplexity ) but heres what perplexity said View attachment 455993

it could be a error or small data set which does make sense bc he only used 2025 which is like only 6 months only so maybe testing for like 3+ years ( according to perplexity ) it would be more suitable and again this is not real world market conditions so yeah but still i mean its good

But my goal is not to have a perfect or a high winrate EA but to make my work easier by making my strategy into code so I can backtest more effiencently so yeah idk how to do it but i wanna do it so yeah lets see how long it will take , and what it will take to find that data

Click to expand...

Try Codex (wiht a new code repository on Github, which is a prerequisite), especially if you're already paying for ChatGPT. If you are serious about coding a complex EA with many subtleties, that will allow the LLMs to read your repository with multiple files of code.
Just add visual lines or some kind of markers for each concept you are modeling to make it easy to test it does what you asked it to.

Also, you'll need years or at least hundreds of trades in order to make some statistics. The good news is that there are many pairs to get that from to test a strategy, just don't overoptimize the parameters, it should still keep true to the strategy.

And you probably need more than a few strategy if you want to get over 60% wins on 1:1 RR, that's pretty hard to achieve by just using an indicator. Keep it simple and good luck with your journey!

eabuilder said:

Try Codex (wiht a new code repository on Github, which is a prerequisite), especially if you're already paying for ChatGPT. If you are serious about coding a complex EA with many subtleties, that will allow the LLMs to read your repository with multiple files of code.
Just add visual lines or some kind of markers for each concept you are modeling to make it easy to test it does what you asked it to.

Also, you'll need years or at least hundreds of trades in order to make some statistics. The good news is that there are many pairs to get that from to test a strategy, just don't overoptimize the parameters, it should still keep true to the strategy.

And you probably need more than a few strategy if you want to get over 60% wins on 1:1 RR, that's pretty hard to achieve by just using an indicator. Keep it simple and good luck with your journey!

Click to expand...

i see , i didnt know abt codex thanks for sharing abt that with me .

thanks

so today i finally started looking at building EAs on pinecode ( tradingview )

I started with this https://www.youtube.com/@michaelionita videos, i did see he sells like a masterclass which i dont like to see but he does share some good information on the videos which is helpful so respect to him for that.

anyways first thing I started with is using just a bare indicator to start building strategies as the video told me

i used Williams Alligator, as michael told abt it bc it would be easy for me to follow
he shared the code in his description also which was helpful but the thing is he used chatgpt but i was wondering what the results are if we use other ai models

so first I will share the stats for the code he shared : (results using chatgpt o3 )



stats for the code i created following his video : ( results using deepseek-r1 )



stats for the code i generated with gemini : ( results using gemini pro 2.5 ) weirdest one so far



so when we take a look at these stats

we can see the chatgppt code is leading with 382,313.66 USD
followed by deepseek 302,854.56 USD and weirdly gemini with $31,025.38 USD

so im not sure why gemini gave that result but i guess maybe it said in the code to use a fixed range for the data but when i checked i didnt find anything like that . im not sure i will have to investigate on it later

this was done with williams alligator indicator , with a dynamic tp and sl based on atr values , about the data time range also im confused bc on top we can see it is oct 5 2009 to july 9 2005
but i m guessing that was not considered as it is deep backtesting but rather the time i set with my strategy

im sleepy, ill talk abt this tmrw
gn

Sounds good. Wish all the best

hey, i’ve been at this a while now, and trust me, patience is everything. started with tiny accounts, learned to cut losses fast, and not chase profits. trading’s a grind, not a get-rich-quick scheme. stick with it and refine your edge daily.

Hey, I've been trading for a while now, but consistency is key. Took me a while to get the hang of it. Started with small funded accounts and slowly built up. It's all about patience and learning from mistakes. Don't expect quick profits, it's a marathon, not a sprint.

I tried it again, with also claude this time and I got these results. I did this to find the best model to continue building strategies




we can see chatgpt performed the best here, followed by gemini and then claude and deepseek

for chatgpt code was done using chatgpt o3
deepseek r1
gemini 2.5 pro
and claude sonnet 4

so other than chatgpt o3 all of them was possible with free tools, but it seems like chatgpt would be a good investment but if we look at the stats again

chatgpt results are good i mean it has he highest total pnl but we can see it also had a very high drawdown of 42.95%
amount of proffitable trades are also high hence winrate

deepseek has the highest drawdown % which is very risky

but lets take a look at gemini

we can see it is the second highest pnl of 1753.51% m with the lowest max equiti drawdown % 17.15% , and highest profit factor

Profit Factor means like for every every $1 lost it gave me $4.083 so like 1:4 RR which is good and the low winrate is reasonable for that case as with the lowest winrate so it is offset automatically

then we can see the sharpe ratio is pretty much equal to chatgpt version just 0.003 different ( im not saying it is good to care less abt numbers even when its small dps but still its reasonable )

and if we look at the sortino ratio it is actually the highest sortino ratio

What is sortino ratio ?
basically like sharpe ratio but sortino focuses on negative returns / downside returns so sharpe ratio considers both positive and negative returns but sortino only considers negative. ( being honest even i dont understand it completely , but just explaining what i understood so maybe one day i will understand it also at the moment lets just care abt the values )

Sortino Ratio Ratings

less than 0 is bad m it means investment return is less than risk-free rate not good at all
0.00 to 1.00 is average
and anything above 1.00 is good and thats what we should target for

so looking at gemini we can see it has the sortino ratio of 2.844 now this is may not be realistic but we can see it performs better

so considering these conditions ill use gemini 2.5 pro to code then also try with chatgpt 4 later but ill go with gemini

i was writing this last night actually but i couldnt finish it , now im disoriented on what should i write so imma finish it for now

and grok heavy is cool and i wanna try it later if i can but damn expensive

anyways thats up for today

see yall again
-terunoz

catarsan said:

Sounds good. Wish all the best

Click to expand...

thanks

wot lmao

i passed the 10k ftmo acc btw , total funded rn is 15k with funded next 5k and ftmo 10k

also i started a journal on x, i will post there more now and post here also
idk who is reading this but if u want to read more u can follow me on x

https://x.com/terunozz

Take care when you test your algos on TradingView. The stats it provides are not accurate. Lot of people complaining about it because overfitting,.
EDIT: gla on your journey

lordtiesto said:

Take care when you test your algos on TradingView. The stats it provides are not accurate. Lot of people complaining about it because overfitting,.
EDIT: gla on your journey

Click to expand...

thanks man, appreciate the tip

Good luck on your journey but trading is very risky as a lot more peoples lose money than make it

Jacobpicotte said:

Good luck on your journey but trading is very risky as a lot more peoples lose money than make it

Click to expand...

yeah well no risk no reward

I started the same way with patterns and candlesticks before refining a real edge.

Hey brother,

I wish you the best of luck in your Trading Journey! I'm hoping to also get back into Trading as well. I do have some experience as a trader, but would like to start from a beginner perspective and re-learn/further indulge myself in the educational aspect of trading. If you have any suggestions on where i could start, I would be highly appreciative. Thanks.

Alpha Ejderhan said:

Hey brother,

I wish you the best of luck in your Trading Journey! I'm hoping to also get back into Trading as well. I do have some experience as a trader, but would like to start from a beginner perspective and re-learn/further indulge myself in the educational aspect of trading. If you have any suggestions on where i could start, I would be highly appreciative. Thanks.

Click to expand...

Check out Tom Houugard, incredible trader and you can just check out his youtube, also read his Best Loser Wins book. Trading is hard, Really hard, especially when you have bills to pay or you have money commitments, It is definitely a long journey and it took me a while to understand that it is all about risk management and simply being lucky enough to hit some big winners along the way. Also Copy trading / Watching the best traders is such an important thing and the fact that you can do it for free in the crypto space is amazing. You can see people growing portfolios and see their entries, strats etc. It's a long way there and at this point I'm convinced everyone has lost their portfolio before learning how to actually trade. Starting with prop firms should be a main priority to learn proper risk management. There will simply be some trades that are FREE and you can profit a lot of money from them, example is $PUMP right now which everyone could have made money on it both shorting it and longing the lows, just look at the chart, add some indicators to it and see how FREE money it actually was. This is why trading is good but at the same time you're gonna burn out hard before you discorver your own perspective. You can't just buy TJR course and be profitable, It takes so much more to enter a trade then some random lines on a chart, you need real time data etc etc. So, I wouldn't start with trading if you need money right NOW.