Strategy Marketplace — Public Research Templates

What Is the Strategy Marketplace?

The Strategy Marketplace is a public collection of published experiments from all SentiSignal users. It transforms individual research into a shared knowledge base where you can browse proven configurations, learn from other researchers, and use their experiments as starting points for your own analysis.

Accessing the Marketplace

You can reach the Strategy Marketplace in two ways:

• From the SentiLab page → click the "Strategy Marketplace" tab
• Directly via the /experiments page

Publication Quality Gate

Not every saved experiment can be published. SentiSignal enforces a quality gate powered by AI evaluation to ensure the Marketplace contains only meaningful, non-duplicate research.

How Evaluation Works

When you submit an experiment for publication, an LLM (Google Gemma-3-27b) evaluates it and assigns a score from 0 to 100 based on six criteria:

1. Duplication check: Are the filters identical to an existing published experiment? Duplicates are heavily penalized.
2. Filter quality: Does the experiment use thoughtful, non-default parameters? Pure default settings (BTC / 7d / minQuality 0.7) receive a penalty because they represent the minimum-effort analysis.
3. Statistical significance: Did the results achieve FDR-corrected significance?
4. Correlation strength: How strong is the reported effect?
5. Data quality: Were quality filters applied? How many data points were used?
6. Uniqueness: Does this experiment test an uncommon asset, unusual lag range, or novel filter combination?

Minimum Score: 60/100

Only experiments scoring 60 or above are published publicly. Below 60, the experiment remains in your private saved collection but does not appear in the Marketplace.

What Gets Published

Published experiments include several AI-generated fields to make them browsable:

• llm_title: An AI-generated descriptive title summarizing the experiment
• llm_description: An AI-generated summary explaining the experiment's approach and findings
• llm_slug: A URL-friendly identifier for direct linking
• llm_quality_score: The 0–100 score from the evaluation
• All original filters: The complete configuration so anyone can reproduce the analysis

Browsing the Marketplace

You can filter published experiments by:

• Asset: Find experiments for BTC, ETH, EUR/USD, Gold, or any supported asset
• Analysis type: Correlation Sweep, Source Predictability, or Narrative Clustering
• Quality score: Sort by score to find the highest-quality research first

Cloning and Modifying

The most practical use of the Marketplace is cloning. When you find an interesting experiment:

1. Click to view its full configuration and results
2. Clone it to create a copy in your saved experiments
3. Modify the parameters — change the asset, extend the time range, adjust quality filters
4. Re-run with your modifications to see how results change

This lets you build on other researchers' work rather than starting from scratch. If someone found an interesting signal for BTC, you can clone their configuration and test it on ETH, SOL, or any other asset.

Contributing to the Marketplace

Publishing high-quality experiments helps the entire community. To maximize your chances of passing the quality gate:

• Use non-default parameters — show that you thought about your filter choices
• Include adequate data (90+ day ranges)
• Test assets or configurations that are not already heavily represented
• Add strategy notes to explain your reasoning

Why This Matters

Research is more powerful when shared. The Strategy Marketplace turns SentiSignal from a solo research tool into a collaborative intelligence platform. By browsing what others have found and contributing your own discoveries, you accelerate the collective understanding of sentiment-price relationships. Next, learn how to turn these research findings into actionable strategies in From Experiment to Trading Strategy.