Simulating Human Decision-Making in Stock Trading with ChatGPT

Introduction

Traditional automated share-market investment tools tend to focus on specific aspects of a share, such as fundamental financial analysis, technical analysis or sentiment analysis.

It is left to the human stock broker to make an assessment, based on all available information, what the best course of action is vis-a-vis a particular share.

Let’s see if we can approximate the work of the human stock broker, by simulating the workings of the human mind.

Psychoanalysis of a Stock Broker

In traditional stock trading, brokers rely on an interplay of experience, rational analysis, caution and emotional intuition to make decisions. From a psychoanalytical perspective, this means that the broker uses different ego states to assess the different aspects of information that are available to them, concerning the financial instrument in question.

(For an introduction to ego state psychoanalysis see A short introduction to Transactional Analysis and Meaningful opinions and Transactional Analysis).

Applying the Ego States in Share Trading

A human stock broker will engage their ego states in assessing a buy / sell / share analysis of a share, as follows:

1. Adult Ego State (Informational):
   • Role: This ego state is purely rational and objective, it processes factual information without emotional or prejudicial influence.
   • Application in Trading: In stock trading, the adult ego state analyses hard data such as current and historical share prices, financial ratios, volume of trading and other quantifiable metrics.
2. Parent Ego State (Cautions and Norms):
   • Role: The Parent state is concerned with rules, norms, and shoulds. It evaluates information based on learned experiences, societal expectations and regulatory frameworks, where applicable.
   • Application in Trading: This ego state guides the trader to consider cautionary aspects of an investment such as avoiding high-risk stocks or advising adherence to proven investment strategies. It also takes into account ethical considerations, regulatory compliance, and historical wisdoms such as market trends and patterns that have shown consistent results over time.
3. Child Ego State (Emotions and Impulses):
   • Role: The Child is the seat of emotions and spontaneity. It’s driven by desires and immediate reactions, which can sometimes lead to innovative or bold decisions.
   • Application in Trading: In the stock market, this ego state will look for signs to be optimistic about a stock. It may also push for taking a chance on a promising but unproven technology stock or reacting to market news with more emotionally charged responses, such as excitement at a potential big win or fear of a loss.

Simulating Ego States in an AI Assistant

The advantage of simulating specific ego states, when deciding on whether or not to buy shares, is that by doing this we can ensure that the AI stock trading assistant doesn’t just follow a set algorithm but engages in a sort of internal dialogue that mirrors human decision-making complexity:

• Balanced Decision Making: Just as a seasoned trader would, the AI assesses its options from multiple perspectives—rational (Adult), cautious (Parent), and optimistic (Child). It merges these perspectives to arrive at a decision that considers both data and intuition.
• Dynamic Adaptation: By continuously updating its database with quantitative and qualitative market information and adjusting its emotional sensitivity based on market sentiment, the AI remains adaptable and responsive to both hard information and sentiment signals.

In other words, by simulating human ego states, an AI trading assistant can achieve a level of decision-making that is nuanced, balanced, and responsive. This approach leverages the precision of technology and the depth of human psychological processes, offering a sophisticated tool that can navigate the complexities of the stock market with an understanding that goes beyond numbers.

OpenAI Assistants

OpenAI provide two ways for users to create custom AI helpers:

• Custom GPT’s: You can create a Custom GPT with no coding skills at all, simply by providing configuration instructions in plain English. You can also supply a Custom GPT with additional information it needs to answer the questions it will be asked, and allow it to code and search the Internet.
• OpenAI Assistants API: This is like a giant Lego set for AI assistants. It requires some coding knowledge, but it gives developers the power to build super-powered helpers for specific tasks. These assistants can analyze data, write sophisticated code (with extra tools) and keep track of conversations.

OpenAI Assistants API is for developers, i.e. it requires coding knowledge to build assistants. It is more powerful and integrates various tools for complex tasks (data analysis, code writing etc.). Also it focuses on conversation flow, i.e. it manages long conversations and keeps track of context.

On the other hand, Custom GPT’s are for everyone, they are easier to use, and don’t require coding. They are designed for simpler tasks, i.e. they are better for basic interactions and content generation. However they have limited conversation management, and may struggle with long or complex conversations.

A Stock Broker OpenAI Assistant

We can configure an OpenAI Assistant to act as a virtual stock broker, who provides sell / hold / buy advice for a specific stock. The OpenAI Assistant used in the following examples is configured to:

1. Gather information: Gather all freely available information for the specified stock.
2. Divide information up among ego states: Determine which information is interesting to the Adult ego state, which information is interesting to the Parent ego state and which information is interesting to the Child ego state.
3. Come to a conclusion: Let the ego states discuss their assessment of the stock among themselves, and come to a final conclusion.

In this way, the Assistant simulates what a human broker would do when presented with various information concerning a stock.

Assessing a Single Stock

The following examples demonstrate the OpenAI Assistant’s ability to provide a reasonable explanation of buy / sell / hold advice for one specific share.

---

IPT.ASX (buy with risk)

Assistant: Upon reviewing the collected information, here is a synthesis of the financial, market sentiment, and technical analysis for Impact Minerals Limited (IPT.ASX):

Conclusion:

After discussing the information among the various ego states, the conclusion for Impact Minerals Limited (IPT.ASX) would be cautiously optimistic. Potential investors or current shareholders may consider a “hold” stance with an option to “buy” if they believe in the company’s long-term potential and are comfortable with the inherent risks of the exploration sector. Those seeking less risk may want to carefully monitor the cash burn and look for signs of successful exploration results before increasing their position. It’s always wise to keep an eye on broader market conditions and sector-specific news that could impact the company’s performance.

---

NASDAQ:AAPL (buy on dips)

Assistant: Upon analyzing the information available for Apple Inc. (NASDAQ: AAPL):

Conclusion:

After considering the information from all ego states, Apple Inc. appears to be a solid company with positive market sentiment and robust financials. Given the current market optimism, strong fundamental factors, and positive technical analysis signals, interested parties may consider this an opportunity to hold or potentially buy on dips if they are confident in the company’s long-term prospects and can accommodate the aforementioned risks. It’s advisable to monitor the upcoming financial reports and market trends closely to refine investment decisions based on the latest data.

---

XNAS:TSLA (do nothing)

Assistant: Drawing information from multiple sources, let’s present an analysis for Tesla, Inc. (NASDAQ: TSLA) within the framework of the three ego states:

With this information in mind, the collective input from the ego states might lead to a more bullish stance, given the recent stock surge, positive market sentiment, and optimism from some analysts. However, the concerns raised regarding product reception and leadership controversies should be weighed carefully. Prior to making a decision to buy, sell, or hold TSLA shares, it would be prudent to perform a detailed analysis of the financials, consider the broader analyst consensus, and examine the technical analysis for current market trends. Additionally, monitoring the company’s responses to any reputational issues and product updates can provide further clarity for an investment decision. If you have specific investment criteria or considerations, please share them, and I can align the analysis more closely with your goals.

---

Assessing Multiple Stocks

The above examples show the ability of the OpenAI Assistant to assess a single stock, now let’s ask the Assistant to choose one investment opportunity from a selection of shares.

User: Of the following shares, which single share should I buy? ASX:QHAL, ASX:ABG, ASX:A4N, ASX:JIN, ASX:MRM

Assistant: Upon conducting a preliminary scan of the latest news for the shares ASX:QHAL, ASX:ABG, ASX:A4N, ASX:JIN, and ASX:MRM, here’s a summary of the information gathered from various financial news sources:

Balancing these considerations, ASX:JIN (Jumbo Interactive Limited) appears to consistently offer solid financial foundation and performance, which the Adult ego state would appreciate, along with the growth potential that the Child ego state finds attractive. The Parent ego state would acknowledge its past performance and profitability, mitigating some concerns regarding risk.

Thus, if it were necessary to make a decision based on the currently available information, ASX:JIN may still be a recommendation. But please note, in real-world scenarios, an actual investment decision should involve more extensive due diligence, including the latest full financial reports and possibly consultations with financial analysts for current expert opinions.

---

Post IPO

Now let’s ask the OpenAI Assistant for recommendations for Reddit, a month after their IPO.

---

User: RDDT:NYSE
Assistant: Based on the gathered information about Reddit Inc. (NYSE: RDDT), let’s consider the data for each ego state:

After considering the information discussed within the different ego states, the conclusion may be as follows:

• Buy: If you believe in the long-term potential of Reddit, particularly its AI prospects, and are willing to take on the risk associated with a newly public company that is not yet profitable but has strong analyst support (as mentioned by Deutsche Bank).
• Sell: If the recent sale of shares by the CEO, the current lack of profitability, and the cautionary stance by several market observers sway you to avoid potential short-term volatility or if you are concerned about the historical performance of meme stocks.
• Hold: If you already own shares but are undecided due to the company’s current performance vs. its longer-term potential, or if you want to wait for more concrete financial data post-IPO and clear trends before making a decision.

Each investor should consider their own investment goals, risk tolerance, and time horizon before making a decision, and it might be beneficial to consult with a financial advisor for personalized advice.

---

In this case, there is no consensus of opinion among the different ego states, since the available information is so varied, therefore there is justifiable room for any possible course of action vis-a-vis these shares.

Summary

By harnessing the principles of Transactional Analysis (TA) in AI-assisted stock trading, we can create a method that does more than process data: instead it simulates the nuanced processes of the human mind. This allows the AI to engage in a type of internal dialogue that emulates the comprehensive decision-making capabilities typical of experienced human traders.

• The Adult Ego State ensures that the AI’s decisions are anchored in rationality, grounding its analysis in the factual evaluation of data such as market trends and financial metrics. This state is responsible for the logical processing of information, offering a robust foundation for sound investment decisions based purely on empirical evidence.
• The Parent Ego State introduces a layer of caution, drawing on historical data and past experiences to provide a context for each decision. This state enforces a risk-aware approach, considering ethical standards and regulatory compliance, which helps safeguard against potential financial pitfalls and enhances the AI’s ability to make prudent investment choices.
• The Child Ego State provides emotional perceptiveness and optimism, enabling the AI to react dynamically to fluctuations in market sentiment and to seize emerging opportunities.

By integrating these ego states, the AI transcends the limitations of traditional algorithms and enters a realm of balanced and intuitive assessment, akin to that of a seasoned human analyst. This multi-faceted evaluation process ensures that decisions are not only data-driven but also tempered with caution and energized by optimistic opportunism.