Token Usage shows how much the AI is being used — think of it as a fuel gauge for the system.

What Are Tokens?

When the AI reads articles, analyzes sentiment, and makes predictions, it processes text in chunks called "tokens." A token is roughly 3/4 of a word. So "The stock market rose today" is about 5 tokens.

The dashboard tracks tokens to help you understand:

• How much work is the AI doing?
• Which tasks use the most resources?
• How much is each part of the pipeline costing?

Dashboard Cards

Card	What It Tells You
Total Tokens	All tokens consumed in the selected period
Total Calls	How many times the AI was called
Avg Tokens / Call	Average size of each AI request
Operations	How many different types of AI tasks were performed

Chart: Tokens per Period

Shows token usage over time. You can spot:

• Daily patterns — predictions use more tokens than article analysis
• Growing trends — more articles = more tokens
• Peaks — days with unusually high usage

Operation Types

Operation	Why It Uses Tokens
Sentiment	Reading articles and scoring sentiment
Surprise	Scanning for unexpected events
Psychology	Analyzing cognitive biases
Prediction	Making price forecasts
Critique	Reviewing own predictions
Intermediate Prediction	Rough updates during the day

Filtering

• Daily / Weekly / Monthly: See usage at different time scales
• Operation filter: Focus on a specific task type

What to Watch

• Is usage growing? More articles = more tokens = more analysis
• Which operation costs the most? Predictions usually use the most
• Are there usage spikes? A sudden jump might mean a lot of new articles or a configuration change