Changelog

See what's new in Perspectec - we're constantly improving our platform.

All Types

Features
Improvements
Bug Fixes
Security
Performance

v1.1.46 AI Features Major

Enhanced SARIMAX-X Forecasting with Exogenous Regressors & Diagnostics

Jun 12, 2026
Business Value: Significantly improves forecast accuracy and model transparency by incorporating external economic drivers and trend data, while streamlining the forecasting workflow and providing better diagnostic visibility into model performance.

Introduced SARIMAX-X Phase 2 with support for exogenous regressors including FRED macroeconomic data and Google Trends, plus automatic integration by default. Added comprehensive diagnostics panel, seasonal projection capabilities, positivity guards, and Census retail factors. Users can now run forecasts for all segments with a single button, with per-candidate exogenous variable selection and credit-per-segment charging.

Enhanced SOTP Segment Forecasting with SARIMAX-X

Jun 12, 2026
Business Value: Delivers more accurate and granular financial forecasts for equity research analysts, enabling better-informed investment decisions through advanced statistical modeling.

The Sum-of-the-Parts (SOTP) valuation model now incorporates Step 9 segment forecasts using advanced SARIMAX-X methodology. This enhancement improves forecast accuracy by consolidating segment-level predictions into the rollup calculation, providing more sophisticated time-series analysis for equity research models.

Financial Data Extraction Improvements - 11 of 11 Benchmark Tickers Now Fully Validated

Jun 12, 2026
Business Value: Users can now generate equity research reports with higher confidence in financial data accuracy, reducing manual verification work and enabling faster, more reliable financial analysis across all major benchmark companies.

Fixed critical validation issues in financial statement processing, including a crash when analyzing UBER reports and improved Net Income comparison logic against Income Statement line items. These fixes, combined with comprehensive updates to the extraction engine, have achieved full validation across all 11 benchmark companies (TSLA, MSFT, GOOG, NVDA, WMT, and others), significantly improving data accuracy and reliability in equity research reports.

Fixed Netflix Weighted-Average Share Count Extraction Accuracy

Jun 12, 2026
Business Value: Improves accuracy of share count metrics in financial models, ensuring more reliable earnings-per-share calculations and equity valuations for Netflix research reports.

Corrected a critical bug in the source verification modal for Netflix's weighted-average share count data where false Calculated/Consolidated-Fallback flags were being incorrectly applied. The fix implements scale-aware context handling to ensure accurate financial metric extraction and validation during equity research analysis.

Improved Financial Forecasting with Maturity-Aware Growth & AI Driver Discovery

Jun 12, 2026
Business Value: Delivers more accurate and contextually-aware financial forecasts by leveraging company maturity metrics and AI-driven identification of material business drivers, reducing manual analysis effort for equity researchers.

Enhanced the forecasting engine with maturity-aware growth caps that adjust predictions based on company lifecycle stage, and introduced LLM-powered driver discovery to automatically identify key financial drivers from 10-K filings. Fixed critical issues with MD&A extraction targeting and SARIMAX exogenous variable ranking to eliminate spurious trend factors and improve forecast accuracy.

Improved Financial Model Setup & Forecast Accuracy

Jun 12, 2026
Business Value: Provides users with more accurate and transparent financial models by ensuring sub-segment allocations match actual filings and reducing confusion between estimated vs. filed data sources.

Enhanced the financial modeling workflow with more robust handling of sub-segment cost allocations, improved forecast period projection for edge cases, and better distinction between filed figures and estimates. Sub-segment costs now reconcile exactly to filed totals using proportional allocation, while verification now only highlights figures sourced from actual filings rather than displaying estimates as unverified data.

Improved Financial Data Scrolling with Always-Visible Horizontal Scrollbar

Jun 12, 2026
Business Value: Reduces friction when analyzing wide financial tables, allowing users to more efficiently explore financial data during equity research workflows.

Enhanced the financial data tables in the model setup workflow with an always-visible, sticky horizontal scrollbar that remains accessible at the bottom of the viewport. This improvement eliminates the need to scroll down to access the scrollbar when viewing wide financial datasets, providing a more seamless experience when navigating through extensive financial information.

v1.1.45 Bug Fixes Major

Critical billing and data integrity fixes for model setup

Jun 10, 2026
Business Value: Eliminates financial leakage from failed billing enforcement, recovers lost analysis results from timeout errors, and ensures accurate financial driver calculations for research models.

Fixed three critical issues affecting financial modeling: (1) billing gate now prevents Anthropic API charges without successful user payment, ensuring no revenue leakage; (2) resolved bug where 504 errors were discarding completed, paid analysis results; (3) fixed driver matrix calculation to properly exclude overlapping memo lines, eliminating incorrect quarterly Fix clicks. These fixes improve data reliability and financial accuracy in equity research models.

Enhanced Model Setup Wizard with Improved Data Reconciliation & Validation

Jun 10, 2026
Business Value: Accelerates equity research workflows by automating financial model setup, reducing manual data entry errors, and ensuring P&L integrity through intelligent reconciliation—enabling analysts to focus on insights rather than data compilation.

This release significantly improves the Model Setup Wizard's financial modeling capabilities with intelligent quarterly-to-annual cost derivation, enhanced P&L completeness checks, and clearer reconciliation states. The wizard now auto-populates missing filing components as sub-rows, eliminates manual data hunting, and provides upfront validation with clear error messaging. Additionally, Claude Haiku has been removed in favor of more capable models (Sonnet 4.6+ and Opus 4.8) for segment classification and unit-driver estimation.

Driver Matrix UI Improvements & Stability Fixes

Jun 10, 2026
Business Value: Provides a cleaner, more stable interface during financial model configuration, reducing user confusion and improving the efficiency of model setup workflows.

Enhanced the financial model setup interface with automatic cleanup of empty sub-rows and more compact rendering for empty subscriptions. Fixed a visual instability issue where cells were flickering in and out of the driver matrix during sweep operations, improving the overall user experience during model configuration.