Artificial Intelligence for Trading: Insights from My Experience and Results

[trend2009]

I am sorry but measuring profitablity over one week for a long only strategy has zero value in determining the profitability of a strategy. I have long-short strategies that worked for 14 years generating millions of dollars that stopped working from one day to the next.

You mean the strategy (I assume it is pair trading) stopped working since then? When did it stop working?

 

[Maverick2608]

You mean the strategy (I assume it is pair trading) stopped working since then? When did it stop working?

It was not pairs trading. It stopped working February 2022 after more than 100.000 trades.

 

[Maverick2608]

"I have pre-trained the AI model according to my strategy..."

It is my understanding that you cannot train Chat.GBT or Gemini. I assume you cannot train Meta's AI either? What do you mean by pre-training?

 

[murray t turtle]

Measuring the 'intelligence' of AI isn’t about how smart it is like a human, but how well it performs specific tasks. For example, in trading, we look at how accurately AI can predict market movements compared to traditional methods. There isn’t a strict threshold for what makes something AI. In my case, I use AI to do jobs that usually require human thinking, like understanding language or making decisions from data.

%%
1] LOOKS like it works fine with language + other tasks;
I noticed it does an excellent job in replacing some union workers[machine learning + machine tasks]
I've noticed trading can be much more complex than union work.
Don't know if they're union + don't care on this example;
2] SAVE has been so long below 200day moving average\ see if it underperforms, long term??
SAVE did great today,+ this week ,so its not a prediction.
- 50% ] As far as buy low + sell hi, that can work well;
but SAVE dropped about 50% in one week+ down\a lot from that level.
Of course super manager Peter Lynch did well with one airline, a historically weak sector , full of unions ;
but SAVE is rated about last in customer service.[JD Power ratings]
Thanks , hope this helps, it helps me.

 

[exo]

"I have pre-trained the AI model according to my strategy..."

It is my understanding that you cannot train Chat.GBT or Gemini. I assume you cannot train Meta's AI either? What do you mean by pre-training?

You can technically "pretrain" (i.e start from scratch) open source models like Meta's Llama, but would take $$$Ms spent on compute (assuming same model/vocabulary).

Otherwise, you could also initialize a model with their published weights (which they already pretrained and spent $$$Ms to compress 'world' knowledge), and "finetune" a model based on your specific task. So instead of tuning 7billion+ parameters, you are only tuning 100k+.

That being said, this post is snake oil

 

[murray t turtle]

It was not pairs trading. It stopped working February 2022 after more than 100.000 trades.

%%
A good one like an apple tree \ may not do anything for 12 months,= loss counting prop taxes ;
then start back up again with fruit/profit

 

[maciejz]

As a person who is reasonably knowledgeable in AI/ML, let me break down why the OP is full of shit. Why do I deem it worth my time to do this? Because posts like this piss me off as they are obvious scams and give people who are ACTUALLY working on AI/ML a bad wrap.

Currently, I use an artificial intelligence model based on Meta's Llama-2, and as a "second opinion," I use OpenAI's GPT API.

I have pre-trained the AI model according to my strategy, and it operates directly through Interactive Brokers via an API. The core functions of the model include:
Joaquín.

OK, thanks for giving us a bit of a background, we will see whether what you write next jives with this background.

1. Real-time market monitoring
The model receives real-time updates on prices, ratios, and news and processes them according to the strategy to detect buying or selling opportunities and execute the corresponding orders. This allows me to detect and manage new positions or close existing ones, although there are additional steps involved to properly qualify a trade.

No it doesn't. That's just not how ChatGPT or any of the LLMs you mentioned above work. They are conversation style models that don't really engage in long running tasks such as "real-time market monitoring." Now, perhaps you have code that is NOT AI code and use that to "monitor" the market, but that's not what you're telling people here; you are stating that "your AI" is doing that, and you and I both know that it's not!

2. Monitoring portfolio exposure
The model continuously processes and calculates the portfolio’s exposure and risk, as well as the results obtained. This allows determining the appropriate size of a new position based on the risk and current portfolio composition and to manage open positions and their performance. Instead of using Take Profit or Stop Loss, the AI acts as a trailing stop, capturing gains or cutting losses based on the current context (e.g., holding a position upon positive news or closing it on negative news).

Yup, same as above, either you don't understand how LLM models work or you do and think that nobody else does. In either case, you're clearly trying to mislead people about what you may or many not have.

Specifically, ChatGPT is NOT "monitoring" your portfolio exposure. You may have some code doing that, but that code has nothing in common with AI (as you claim).

3. Calculating statistics, ratios, and prices
The AI constantly calculates market ratios and statistics. I apply various mathematical formulas to technically support the above fundamentals and precisely determine buying or selling prices, for example, measuring average trading volume over different periods to calculate the VWAP, estimating the average historical price variation of the asset, its relation to the indices, etc.

Again, nope! What does this even mean -- "the AI constantly calculates market ratios ..." ??? Do you mean that you have code that gets data from IBKR and your code calculates ratios? Sure, I believe that. But why claim that your very pedestrian code is some AI? What's your goal here? The fact that you even wrote these claims makes me think that you have very little experience with: writing code, AI/ML, and investing/trading.

4. Training the model based on results
In parallel to the main model, I have two other AI models operating in demo mode. The first one learns in real-time about the positions taken by the main model and has the freedom to modify the strategy and make decisions. Training this model simultaneously allows me to leverage the AI’s ability to improve based on experience and then, after refining the results, to implement enhancements in my main strategy. The second model simply implements what is learned to test in a simulated scenario whether the new strategy yields better results than my principal strategy.

This is "plausible" although you have not provided any details that would let me evaluate whether you even have a clue about how any of what you claim would be done. Specifically, HOW are you training these two additional models? There is pretty much only one correct answer to this question given the scenario you have provided and you made no mention of it. There is also the question of why are you training these side models? Other than it sounding cool, is there a reason? The reason you provided in the above paragraph again just points out your lack of understanding of AI/ML.

Although the background and operations are quite complex, in practice, it boils down to determining if any fundamental event could impact the price of a stock, understanding the entire history of such events and their effects to gauge the potential impact they had, have, or might have. It also involves calculating formulas according to my strategy and the current exposure of the portfolio to decide whether to buy or sell stocks, at what price to do so, how much, and when to take profits or cut losses.

So, let me guess, you've read a book or two on investing, heard about ChatGPT, and decided that this kind of post is your ticket, huh? What you've written here sounds great. Unfortunately, what you claim is not possible with the technologies you specifically claim to be using. There are so many problems with this paragraph that I thought about skipping it. But, I will address a few of them.

First, what is a "fundamental event"? YOU must have "trained" the model on fundamental events, right? So, you should be able to answer this in mind numbing detail; I suspect you can't provide any details beyond some general hand waving.

Second, if we gloss over the first point and assume that "the AI" determined the definition of "fundamental event" on it's own during training (not likely), then why does it need to then calculate your formulas to make decisions? The whole point of training is for the AI to find the relationship between your input (the data you feed in) and output (the decisions). In this paragraph, you kinda claim that the AI was trained to find this relationships between the "fundamental events" and forward price (returns), but then claim that the same AI then needs very specific calculations (formulas provided by you) to make trading decisions? I'm gonna call bullshit on this again. It really sounds like you don't really understand how AI/ML works.

It really is starting to sound like you're very new to all these subjects: software, AI/ML, and investing/trading. What you've written sounds good on the surface to a newbie. The problem is that none of it is very consistent. Who determines "fundamental events"? Do you determine then and feed them into the AI? That's plausible but then we get to my two points above. If the AI determines the "fundamental events" on its own, then the "fundamental events" become invisible as they are integrated into the relationship that the AI "learned" between the input and output. In this case, why even mention the term "fundamental event" other than because it sounds good?

My investment strategy
Broadly speaking, my investment strategy encompasses all aspects mentioned in the implementation section (macro and micro fundamentals, price action, statistics), along with risk parameters and exposure suitable for my profile, in addition to some proprietary formulas I've developed over the years through trial and error.

I think I've spotted the kitchen sink in there too Got it, your strategy is the "best" because it takes literally "everything" into account. This is just further evidence for the point of view that you're really new to all of this. Again, sounds good on the surface to a newbie, but is not very consistent. Do you really think that macro events with data released on monthly or quarterly cycles are going to affect what happens in your two day holding period? Yeah, neither do I But it sure sounds good when you tell us that you take all this stuff into account.

Moving on.

Instead of using traditional technical indicators, I directly calculate the relevant variables for my strategy based on historical price data. Rather than using Take Profit or Stop Loss, I measure the real-time impact of news, events, price action, volume, and the order book (and compare it with the historical data) to adjust the exit price. To determine which stocks to buy, I focus on two factors: the current price in relation to historical data and the events that have affected, are affecting, or may affect it positively or negatively. I also calculate the potential return it can generate and whether the risk is suitable for my portfolio.

Up until now, I thought that you merely don't know anything about AI/ML. But, the first sentence here reveals that you don't know anything about technical indicators either I'll give you a hint, technical indicators ARE based on price data.

The rest of the paragraph reads like a laundry list of "features" a newbie would come up with when asked to design a trading system.

In summary, I simply apply the rule of "buy low, sell high." The real key is knowing as accurately as possible when something is cheap and when it is expensive.

You don't say. Shit, we've all been going about this the wrong way

Profitability
Over the last week (from Friday, May 3, 2024, to Friday, May 10, 2024), I achieved a net return of 14.86% on closed positions. Considering the positions that remain open, the return exceeds 23%. This profitability is measured in relation to a capital of $100,000 and not on the actual investment return, which is considerably higher. The S&P 500's performance for the same period was 1.85%.

Portfolio Ratios

• Sharpe Ratio: 17.5
• Sortino Ratio: 255.41
• Standard Deviation: 2.6%
• Downside Deviation: 0.18%

These ratios reflect the exceptional risk-return relationship in managing the portfolio.

I guess we have to add "statistics" to the list of things you're clueless about. I just don't even understand what you're doing here. I mean, how can you calculate these kinds of statistics based on one week worth of results? I just don't get it man, are you really this bad at trying to scam people? At least try to make it look half way legit. What you've posted just screams that you don't know shit about shit.

Undoubtedly, there is still a long way to go, at least in my case. I began real account trading with the model last November, starting cautiously and gradually increasing exposure.

Dude, just take the win. Anyone with a 17.5 SR trading strategy would be spending their time picking out the color of their new Ferrari, selecting the builder of their mega yacht, and touring Caribbean islands to buy. Don't be so modest claiming that there is "still a long way to go". You, my friend, have legitimately made it

Thank you for reading about my experience. I would be delighted to hear your opinions or comments, or to answer any questions (even technical ones) you may have.

I believe I have made my "opinion" quite clear. But, if you'd be so kind, please tell me why you have posted this garbage here? Is the idea to get people to believe you and then send you money that you can scam? Perhaps you are running an psychological experiment to see just how gullible people can be? Or, are you fishing for compliments and people to tell you how smart you are? Is there perhaps another reason?

I can tell you that I find posts like this extremely toxic. You waste everyones time and you misrepresent what AI can and can not do. AI/ML does have applications in investing/trading, but not the way that you suggest and certainly not the way you've described. You are causing harm to the industry because you are causing AI to be equated with scam. Please stop!