We should ask ChatGPT to build a profitable trading system

[Tradess0610]

i think i read somewhere that hft algos are good for maybe a week and
they have to be tweaked.

This would not surprise me at all. As a hf, it is so important to keep programmers in-house in order to stay on the top of the game.

 

[virtusa]

So I signed up and typed in "give me an example of a profitable trading strategy"

There are many profitable trading strategies, but one popular strategy is trend following. The basic idea behind trend following is to identify a trend in the market and trade in the direction of that trend. Here's an example of how a trend-following strategy might work:

1. Identify the trend: The first step in a trend-following strategy is to identify the direction of the trend. This can be done using technical analysis tools, such as moving averages or trend lines. For example, if the price of a stock is consistently making higher highs and higher lows, then the trend is considered to be up.
   
   
2. Enter the trade: Once the trend has been identified, the trader can enter a trade in the direction of the trend. This might involve buying the stock if the trend is up, or selling it short if the trend is down.
   
   
3. Set a stop-loss: To manage risk, the trader should set a stop-loss order at a predetermined level. This is a price at which the trader will exit the trade if the market moves against them. For example, if the trader is long the stock, they might set a stop-loss just below a key support level.
   
   
4. Monitor the trade: Once the trade has been entered, the trader should monitor it to see if the trend continues. If the trend remains intact, the trader can hold the position and potentially ride the trend for a significant profit. If the trend changes, the trader can exit the position at the stop-loss level.
   
   
5. Take profits: Finally, the trader should have a plan in place to take profits if the trade goes in their favor. This might involve setting a profit target or trailing the stop-loss to lock in profits as the trend continues.

Overall, a trend-following strategy can be profitable if the trader can accurately identify trends and manage risk effectively. However, it's important to note that no trading strategy is foolproof, and traders should always be prepared for the possibility of losses.

This clearly shows that AI is still far away from inventing a profitable trading strategy. And which AI can describe in a very detailed way, with formulas and explanations of the logic that is used. AI only reproduces what it can find on internet.

If I would have knowledge that is not posted on internet, AI can never use that knowledge.

 

[virtusa]

“Which is heavier, a toaster or a pencil?”

To a human, even a small child, the answer is obvious: a toaster.

GPT-3’s response: “A pencil is heavier than a toaster.”

ROFLMAO.

 

[maxinger]

At your service.

 

[virtusa]

At your service.

View attachment 306562

Don't forget that what you show is hindsight trading. Humans can do that too. Just optimize any system till it get profitable.
But the problem is that they only can do that for past. Let AI trade today and the next week in realtime. I bet the results will be much lower.

90% win rate doesn't mean anything. Expectancy can still be very low or even close to zero.

This post shows that AI has no clue about future moves, and that is exactly that we need to trade. Not the past but the future.

"Will the SPX drop tomorrow?"

As an AI language model, I do not have access to information about future market movements or events that could affect the market. Predicting market movements is a highly complex and uncertain task, and even the most experienced investors and traders cannot accurately predict the future movements of the market with certainty. There are many factors that can affect the movement of the market, including economic indicators, global events, geopolitical tensions, and many other unpredictable factors. It's important to conduct thorough research, analyze data, and carefully evaluate risks before making any investment decisions.

 

[maxinger]

Don't forget that what you show is hindsight trading. Humans can do that too. Just optimize any system till it get profitable.
But the problem is that they only can do that for past. Let AI trade today and the next week in realtime. I bet the results will be much lower.

90% win rate doesn't mean anything. Expectancy can still be very low or even close to zero.

This post shows that AI has no clue about future moves, and that is exactly that we need to trade. Not the past but the future.

agree.

Those bots are stupid.
They should aim for say 50% win rate,
not a 90% win rate.

Experienced traders know very high win rate of 90% means negative returns.

 

[Pekelo]

Well, they are more conversationalist, then future predictors, specially when math is involved. I am sure they can predict the weather because they look for forecast online.

That is not to say in the FUTURE they will get better and at the very minimum they can recognize a trend.

 

[Zwaen]

So what if the bots are fed with disinformation (just like the supermarket card which get swapped between users), which comprises 98,36% of the webcontent now ?

Was that not the reason to the public announcement, to put less garbage in?

 

[Cuddles]

probably easier to have AI flood social media w/FUD and FOMO sentiment to move retail

 

[virtusa]

probably easier to have AI flood social media w/FUD and FOMO sentiment to move retail

In crypto, FUD has become a well known crypto term, and it means one of two things:

1. To spread doubt about a particular token or project in an attempt to manipulate prices downward.

2. The general skepticism and cynicism about crypto as an asset class, and any related news/events. Even the rumor of a negative event possibly happening can generate FUD.