We generated our latest stock recommendation using OpenAI's new flagship model, GPT-4o.

On May 13th, 2024, OpenAI released its latest flagship model, GPT-4o. We decided to use this model, along with a variation of The GPT Investor's most successful stock recommendation strategy to date, to generate our latest stock recommendation.

In this essay, we will describe the technology, methodology, and results.

GPT-4o's standout feature is integrated multimodal capabilities, allowing it to process and generate text, audio, and visual inputs and outputs within a single neural network. The model also excels in reasoning tasks, achieving top scores in benchmarks like the Massive Multitask Language Understanding (MMLU), which tests world knowledge and problem-solving abilities across diverse subjects.

In our experiments at The GPT Investor, we've found that newer models substantially outperform older models in stock recommendations. For example, GPT-4's return was approximately 156.86% better than GPT-3.5's return. We expect this trend to continue as new models become smarter.

Methodology:

First, we used a ChatGPT Plus account to ask GPT-4o to describe a hypothetical stock called "X," which has all the attributes of the most investable stock on the NYSE and NASDAQ. A variation of this same prompt had previously produced our most successful stock recommendation. The main difference between the old prompt and this new prompt was that we removed a large portion of the prompt to eliminate any potential bias. The idea was to give GPT-4o more liberty in determining what attributes are important for "the most investable stock," rather than specifying what attributes we thought were important. See both the new and the old prompts below.