SECFSN Plus Models — Reproducible Financial Modeling Research

SECFSN Plus Models is the fourth layer of my SEC financial-statement research architecture.

It takes prepared financial features and converts them into reproducible modeling evidence for downstream research.

The framework focuses on:

  • Forward-label validation: Aligning financial observations with next-period outcomes while preserving company-period alignment.
  • Time-aware modeling: Separating earlier training data from later held-out evaluation periods.
  • Governed outputs: Preserving model predictions, evaluation information, and company-period identities.

The demonstration uses a deterministic synthetic financial panel and runs the repository’s actual modeling workflow.

Synthetic Layer 04 workflow: 42 input rows, five modeled features, 36 forward returns, 24 training rows through 2020Q4 and 12 held-out rows; Layer 05 index preserved.
Canonical synthetic demo: 42 company-quarter rows yield 36 forward labels. The model uses 24 training rows through 2020Q4 and 12 later held-out rows. This demonstrates chronological workflow, not a point-in-time investment backtest.
Six synthetic company records for 2021Q1 with company-period identifiers, model predictions, expected returns and ranked risk output for Layer 05.
The notebook’s six-row Layer 05 handoff preview preserves company-period identities. Model risk probability is a relative percentile rank, not a calibrated probability; these synthetic outputs do not establish investment performance.

The objective isn’t simply to train a model. It’s to establish a transparent process for evaluating financial models without overstating results from limited samples.

The resulting evidence can be passed to SECFSN 05 for further research synthesis.

GitHub: https://github.com/rossautomatedsolutions/secfsn_plus_models

Research and demonstration purposes only. Not investment advice.

Sharing

Related Articles

  • All Post
  • Articles
  • Blog Post
  • General Business Automation
  • Portfolio
  • Stock Market & Finance