Every morning before the market opens, I like to check in with Schwab’s Market Open Update — their daily summary of macro conditions, economic data, and big-picture sentiment.

But let’s be honest: most days, I don’t want to read another macro recap.

So I built a script.

Why I Built It

As a systematic trader and developer, I look for ways to simplify edge detection and automate the decision-making process. I’m particularly interested in how market tone affects my position sizing — whether I should go in heavy, stay light, or skip the day entirely.

Rather than relying on my own mood while reading the headlines, I decided to let a large language model do the reading and decide:
Is the sentiment today bullish, bearish, or mixed?

How It Works

The script pulls the latest update from Schwab’s site and feeds it to a language model — either Claude 3.5 Sonnet from Anthropic or GPT-4 from OpenAI.

From there, it distills the article into one word:
Bullish, Bearish, or Mixed.

That value is logged in a .csv-style file (market_sentiment.log) alongside the date and the model used. The log gets referenced by my trading bot, which adjusts entry size and aggression based on that day’s sentiment.

Here’s an actual log output from today:

date,sentiment,model
2025-04-11,Mixed,anthropic,claude-3-5-sonnet-20241022

Simple. Reliable. Non-emotional.

Why Use AI for This?

Language models are particularly good at tasks like:

  • Summarizing macro narratives
  • Detecting optimism or caution in tone
  • Cutting through jargon and filler to highlight risk sentiment

It turns out that what I was doing manually — skimming a few paragraphs and making a gut call — is exactly the kind of repetitive judgment task that AI can handle well.

Soon, I won’t make sizing decisions based on “how I feel about the open.” I’ll let the model read it, log it, and move on.

Want to Try It?

If you’re curious to see it in action or want to integrate it into your own workflow, you can clone the full repo from GitHub. It’s lightweight, local, and easy to plug into any script-based system.

👉 GitHub Repo: Market Sentiment Logger

In a future post, I’ll show how I use this sentiment signal as a core input in my trading bot’s engine, and explain how it dynamically adjusts entry size and aggressiveness based on the broader market tone.

Until then, feel free to fork, tweak, or use it as-is — and keep your decision-making clean and data-backed.