Can we expect good results from AI Investment? For the past three months, I provided Chat GPT with price change data from the past year at the same time every day, and based on this data and surrounding information, I had it decide whether to buy/sell/hold. I have summarized the profits when executing ‘buy’ or ‘sell’ according to Chat GPT’s opinions.
The investment amount for each stock was assumed to be 1,000,000 won.
Trading History and Opinions by Date
For the past three months, the investment history and profit rates for coins and stocks, as well as the reasons for ChatGPT’s investment decisions for each date, can be found at the link below.
Results When Investing Based on Daily Investment Opinions
AI Investment opinions were categorized into cases where trades are made daily, trades are made based on a one-week outlook, and trades are made based on a one-month outlook. Here, I summarize the results for cases where trades are made according to daily investment opinions.
Case of 4 Installments of Purchase
The investment profit is based on the assumption of investing 1,000,000 won by purchasing 250,000 won each time a ‘buy’ opinion arises.
[Samsung Electronics] shows a loss of 8,395 won.
[Doosan Enerbility] shows a profit of 132,833 won.
[Bitcoin, BTC] shows a profit of 59,478 won.
Case of Bulk Purchase
The investment profit is based on the assumption of purchasing 1,000,000 won each time a ‘buy’ opinion arises.
[Samsung Electronics] shows a loss of 33,581 won.
[Doosan Enerbility] shows a profit of 238,308 won.
[Bitcoin, BTC] shows a profit of 132,848 won.
Results When Investing Based on Weekly Investment Opinions
AI Investment opinions were categorized into cases where trades are made daily, trades are made based on a one-week outlook, and trades are made based on a one-month outlook. Here, I summarize the results for cases where trades are made according to weekly investment opinions.
Case of 4 Installments of Purchase
The investment profit is based on the assumption of investing 1,000,000 won by purchasing 250,000 won each time a ‘buy’ opinion arises.
[Samsung Electronics] shows a profit of 9,045 won.
[Doosan Enerbility] shows a profit of 139,659 won.
[Bitcoin, BTC] shows a profit of 113,494 won.
Case of Bulk Purchase
The investment profit is based on the assumption of purchasing 1,000,000 won each time a ‘buy’ opinion arises.
[Samsung Electronics] shows a profit of 29,410 won.
[Doosan Enerbility] shows a profit of 194,344 won.
[Bitcoin, BTC] shows a profit of 72,961 won.
Results When Investing Based on Monthly Investment Opinions
AI Investment opinions were categorized into cases where trades are made daily, trades are made based on a one-week outlook, and trades are made based on a one-month outlook. Here, I summarize the results for cases where trades are made according to monthly investment opinions.
Case of 4 Installments of Purchase
The investment profit is based on the assumption of investing 1,000,000 won by purchasing 250,000 won each time a ‘buy’ opinion arises.
[Samsung Electronics] shows no profit or loss. There were hardly any ‘sell’ or ‘buy’ signals.
[Doosan Enerbility] shows a loss of 53,016 won. There were hardly any ‘sell’ or ‘buy’ signals.
[Bitcoin, BTC] shows a profit of 44,958 won.
Case of Bulk Purchase
The investment profit is based on the assumption of purchasing 1,000,000 won each time a ‘buy’ opinion arises.
[Samsung Electronics] still shows no profit or loss. There were hardly any ‘sell’ or ‘buy’ signals.
[Doosan Enerbility] shows a loss of 100,953 won.
[Bitcoin, BTC] shows a profit of 72,269 won.
Conclusion
Simulation Based on AI Opinions
I compared the profits when investing based on AI trading opinions for Samsung Electronics, Doosan Enerbility, and Bitcoin.
The monthly outlook did not yield much, and it seems to be an unreliable opinion. In the case of Bitcoin, it showed a profit rate of 4-7% over three months, but considering that the price of Bitcoin rose from 128,000,000 won to 149,000,000 won, about a 16% increase, it seems insignificant.
Although the weekly outlook also did not yield much, it produced better results than the monthly outlook. For ‘Doosan Enerbility,’ the profit rate was around 13-19%, and for ‘Bitcoin,’ it was 7-11%. However, seeing that ‘Doosan Enerbility’s’ stock price nearly doubled from 27,000 won in March to close to 60,000 won in June, it also seems insignificant.
In the case of trading based on daily outlooks, ‘Samsung Electronics’ recorded a loss of about 3%, while ‘Doosan Enerbility’ showed a profit of 13-23% and ‘Bitcoin’ recorded a profit rate of 5.9-13.2%. Therefore, trading based on daily outlooks yielded better results than monthly or weekly outlooks. However, only ‘Samsung Electronics’ showed slightly better returns compared to stock price fluctuations, while ‘Doosan Enerbility’ and ‘Bitcoin’ showed maximum profits of 23% and 11% respectively when stock prices increased by 100% and 16%.
Based on the above, it seems that seeking trading opinions and investing based on the past year’s price information and market information recognized by AI is not a recommended approach. There may be errors in the method I chose for questioning Chat GPT or the type of information provided, and there may be better methods, but I conclude that I cannot trust it.
I hope this is helpful~
Conducting Other Simulations
Next, I am conducting simulations on investment methods that involve mechanical trading in various price ranges for stocks that have characteristics of prolonged sideways movement or long-term upward trends.
I have high expectations that it will lead to non-loss trading~~.
Updated : June 19, 2025
{{ reviewsTotal }}{{ options.labels.singularReviewCountLabel }}
{{ reviewsTotal }}{{ options.labels.pluralReviewCountLabel }}
