FishFyndr Documentation

1. Data Source: RentCast API

FishFyndr sources core property and rental data from RentCast API, which aggregates property listings, rental estimates, and sales records from multiple MLS feeds and public sources.

  • Update Frequency: Data is updated weekly to capture the latest sales, listings, and rental market changes.
  • Coverage: Coverage is strongest in metro areas, especially Atlanta, and may vary in rural zip codes.

2. Comparable and Rent-to-Value Calculations

FishFyndr calculates property comparables (comps) and Rent-to-Value (RTV) using a custom algorithm that finds similar properties based on proximity, property type, and size. The key steps of our methodology are:

  • Only properties of relevant types (Single Family, Condo, Townhouse) are included in the analysis.
  • Nearby rental properties are identified using spatial matching (within approximately a 7 mile radius).
  • Comparables are filtered to match the subject property’s type, then ranked by similarity in square footage.
  • The 5 closest comps (by square footage) are selected for each property.
  • The average rent price from these comps is assigned as the property’s estimated rent.
  • Rent-to-Value (RTV) is calculated as: estimated monthly rent × 12 ÷ property price.

This approach aims to mirror how a local agent or appraiser would select relevant comparables for property analysis, but with greater speed and objectivity.

3. Property Valuation Model

FishFyndr’s property price estimates are generated by a custom-trained machine learning model using historical listing and sales data. Our pipeline and model design mirror modern Automated Valuation Models (AVMs) with a focus on interpretability and data quality.

  • Data Preparation: Core data is pulled from RentCast and internal databases. We remove irrelevant features (e.g., agent/office info, MLS numbers), clean dates, and engineer new features such as property age, price per square foot, and seasonal factors.
  • Categorical Encoding: Categorical variables (like city, county, property type, zip code, listing status) are one-hot encoded to help the model learn differences between locations and property types.
  • Spatial Clustering: Latitude/longitude pairs are clustered (KMeans, k=10) to group properties into location-based clusters, adding neighborhood context.
  • Feature Engineering: Additional features include price per sqft, lot size per bedroom, new construction flag (age ≤ 2 years), and cyclic (sin/cos) month features for market seasonality.
  • Model Training: We use an XGBoost regressor trained on log-transformed sale prices to improve stability and handle outliers. The model is trained/tested separately for homes and for land parcels.
  • Evaluation: Model accuracy is measured with Mean Absolute Error (MAE) on a held-out test set. Final models are exported with feature column order to ensure prediction compatibility.
  • Limitations: Model predictions rely on available data and may not reflect recent renovations, unique property features, or current market volatility. Use estimates as a starting point—not as a replacement for a licensed appraisal.

The FishFyndr valuation model is regularly retrained with fresh data and engineered for transparency, so users can trust the rationale behind every automated price.

4. AI Assistant (Under the Hood)

FishFyndr’s AI Assistant leverages OpenAI’s GPT-4 architecture with a custom dataset of Atlanta-area property data. The AI is fine-tuned to answer real estate, rental, and investment queries using up-to-date FishFyndr datasets, public property records, and general market insights.

  • Summarizes and interprets dashboard metrics
  • Explains real estate terminology
  • Guides users through using the app
  • Does not provide legal or financial advice

5. Additional Information

  • Coverage Area: FishFyndr is focused on the Atlanta metro area, but national expansion is underway.
  • Data Privacy: No personal or user-submitted data is sold or shared. Authentication is powered by Firebase.
  • Methodology Transparency: All calculation algorithms are open for review—contact us for more details or suggestions.
  • Support: For questions or bug reports, use the Contact Us page.

6. Legal / Disclaimer

  • Informational Purposes Only: FishFyndr is provided for general informational and research purposes. The data, analytics, and AI responses do not constitute investment, legal, tax, or financial advice.
  • No Guarantees: While we strive for accuracy, all information is provided “as is” and may be incomplete or contain errors. FishFyndr does not guarantee the accuracy, completeness, or suitability of any data for your particular needs.
  • Consult Professionals: Always consult with a licensed real estate agent, attorney, or financial professional before making investment or purchasing decisions.
  • User Responsibility: Use of FishFyndr is at your own risk. By using this platform, you acknowledge and accept these terms.
  • See Terms: For more information, please review our Terms of Service.

7. Roadmap & Feedback

We’re always improving FishFyndr! Planned features include: custom report exports, investment scenario modeling, and additional city support. Your feedback directly shapes our roadmap.

FishFyndr © 2025