Three financial workflows with Grok 3 Mini Beta in NexusTrade
8 min readThis April 2025 tutorial shows three financial workflows: a historical rebalance test, a stock screen saved as a watchlist and a news query. The walkthrough selects Grok 3 Mini Beta. The chart below evaluates Grok-3-Beta, a different model.
The chart reports a 0.747 average SQL-generation score and a 67.5% perfect-score rate for Grok-3-Beta in that evaluation. Those scores do not establish Mini Beta’s result or rank every financial workflow. The tutorial examples below can be inspected separately through their saved conversations.
The purpose here is to show the questions, saved outputs and historical test settings for these three workflows.
Follow one workflow, open its linked conversation and compare the saved output with the question. The first backtest uses January 1 through April 10, 2025.
The following Google SQL article is a separate evaluation. It is background on the testing method, rather than evidence for Grok 3 Mini Beta’s ranking.
I tested the best language models for SQL query generation. Google wins hands down.
Imagine a cheap, fast model that’s more accurate than the expensive ones. That’s Gemini Flash 2
medium.com
How can AI help you make better trading decisions?
AI can help you make better trading decisions by doing the hard work of shifting through market noise.
Using AI tools like NexusTrade, you can:
- Test different investing strategies and see which ones fare well in different market conditions
- Find novel stock buying opportunities, including fundamentally strong stocks are stocks that are oversold
- Stay up-to-date on news for your favorite stocks so you’re always in-the-loop.
Let me show you how this works using the NexusTrade platform.
NexusTrade - No-Code Automated Trading and Research
Perform financial research and deploy algorithmic trading strategies
nexustrade.io
Choose the model used in this historical walkthrough
For this April 2025 walkthrough, I selected Grok 3 Mini Beta in the model settings. The exact model behind each saved transcript needs to be checked in its record.
- Go to nexustrade.io/chat
- Click “Settings” on the right-hand side
- Click the model selector. This tutorial selected Grok 3 Mini Beta; that choice is distinct from the Grok-3-Beta benchmark above. The other models named here were the alternatives described in April 2025.
Don’t see a model that you like? Leave a comment and I’ll add it faster than you can say “NexusTrade rocks!”
Once you have the model setup, you can ask any type of hard financial question. Here are some examples.
Structuring this section
To help readers extract the most value from this section, I’m structuring in the following way:
- Header (hyperlinked with the direct link to the full conversation)
- Question: What exactly did I ask the AI?
- Answer (Screenshot): What did the AI respond?
- Direct link: A direct link to the full output of the AI model
This will allow you to review the AI’s responded without having to stare at this tiny little screenshots. It’ll also allow you to follow along, so YOU (yes, you) can become an algorithmic trader. You can continue from where the conversation left off or copy paste what I said into the chat.
When following along, please note that I have my backtest settings to automatically test from Jan 1st 2025 to April 10th 2025. If your backtest looks different than mine, change the settings.
Now, let’s get started!
How to use AI to test different investing strategies?
Let’s say I wanted to test out different investing strategies and see which ones fare well in the long-term. Here’s how I would do this.
First, I will start with an idea. My idea is “stocks with a high market cap tend to perform better”. Then I will test it.
To start, I’ll first fetch a list of stocks with a high market cap. I’ll choose to do so in the past to avoid lookahead bias.
What are the top 100 stocks by market cap as of Dec 31st 2024?
Then, I’ll create a strategy and test it out this year to see how it performed compared to the broader market.
Create a strategy that rebalances the top 25 stocks by market cap every 3 months at equal weight
Using AI, we were able to test out this complex trading strategy within seconds. We know that at least during this time period, stocks with a higher market cap performed slightly worse than just holding SPY (-7.26% vs -6.72%).
Zooming in on the backtest, we can also see other metrics like our sharpe ratio, sortino ratio, and drawdown. These metrics tell a story of how risky our strategy is compared to the broader market.
Using this same exact methodology,
- We can change the rules to rebalance every week. Or every 3 months
- We can test it out in different time periods like Covid or during presidential elections
- We can test with other metrics like revenue or free cash flow
- We can include more complex conditions, different actions (like “buy” or “sell”) or add filtering logic for the strategy. For example, we may want to exclude the “Magnificent 7” and see how our results perform.
All of this possible from a regular conversation with an AI.
Once we’re done, we can take the trading strategies we’ve created, deploy them live to the market, and even monetize them if they’re successful!
How I Make $600/Month With AI-Powered Algorithmic Trading Strategies (No Coding Required)
No startup capital required. How to make money online in 2025
medium.com
To read the full conversation, check out this link.
How to use AI to find novel stock buying opportunities?
In addition to testing trading strategies, you can use AI to find novel stock buying opportunities. This includes using technical indicators and fundamental indicators to sift through the market.
For example, let’s try to find fundamentally strong stocks that are potentially oversold.
What AI stocks with a market cap above $20B and a fundamental ranking above 2.5 have the lowest RSI?
“Fundamental ranking” is a shortcut for understanding if a company’s business is strong. I generated this by supplying the earnings of every company into a language model and asking the model to evaluate the company fundamentally. By filtering on stocks with a ranking above 2.5, I get all stocks that have AT LEAST a “decent” business.
I used AI to analyze every single US stock. Here's what to look out for in 2025.
I can already feel the vitriol from the anti-AI mafia, ready to jump in the comments to scream at me about "stochastic…
nexustrade.io
The model responds with the following:
Similar to what we did above with the trading strategies, we can then take these stocks and test out different trading ideas or create a new watchlist of stocks.
Add these stocks to a new watchlist
By creating a watchlist with these stocks, we can remember them for later, which can help us inform what stocks we want to buy at a given moment.
Additionally, we also get a daily email on what’s going on with these stocks in the news.
However sometimes, waiting until the end of the day is too long, and we want to know what’s happening with these stocks RIGHT NOW.
We can do that too.
Read the full conversation for this section here.
How to use AI to stay up-to-date in the news?
Using AI, we can automatically find the most relevant news that we care about right now.
For example, I’m an avid NVIDIA trader/long-term holder.
I lost $16,649 on NVIDIA. Here’s why I bought more.
My Robinhood account year-to-date
medium.com
Seeing that NVIDIA was down today, I might want to understand “why” without having to search the news manually.
Now, I can within seconds.
What’s going on in the news with NVIDIA today? Why did it fall?
This isn’t limited to just particular stocks. We can analyze groups of stocks or even search for news based on broader industries.
What’s in the news for AI this week?
News about anything you want, right at your fingertips.
How insane!
Read the full conversation for this section here.
What other things can AI do?
This is just the start of what AI can do to help make you a better trader. Some other things you can ask it includes:
- (After uploading an attachment) “Analyze this watchlist fundamentally”
- “Backtest the strategy you’ve created across the Covid pandemic”
- “What is the difference between technical and fundamental analysis?”
If you need help, the NexusTrade platform even has in-app tutorials to show you step-by-step how to make the most out of the platform.
Trading Tutorials - Learn Algorithmic Trading
Learn algorithmic trading and financial research with our comprehensive tutorials. From basic concepts to advanced…
nexustrade.io
What are the limitations of AI for these tasks?
While AI is powerful for financial analysis, it’s not without limitations. Even the best models like Grok 3 have constraints that traders should be aware of:
- AI cannot predict unexpected market events or “black swan” scenarios that haven’t occurred in historical data
- The accuracy of these models fully depend on the data that’s inputted into it. The old data science adage remains true: “garbage in, garbage out”
- Technical accuracy doesn’t replace human judgment about risk tolerance or investment goals
- AI tools work best as supplements to your trading knowledge, not replacements for fundamental understanding
- Backtesting results, while useful, don’t guarantee future performance in different market conditions
Conclusion
The future of trading and investing is here, and it’s AI-powered. With tools like NexusTrade’s Grok 3 integration, you can quickly test trading strategies, find novel buying opportunities, and stay informed with comprehensive news analysis — all in seconds instead of months (or longer).
Try one of the three linked conversations above. For the rebalance example, check the saved December 31, 2024 stock list and the January 1 through April 10, 2025 test before changing the rules.
NexusTrade - No-Code Automated Trading and Research
Perform financial research and deploy algorithmic trading strategies
nexustrade.io
No comments yet.