Editorial Standards & Methodology
Convex is committed to producing accurate, transparent, and timely economic analysis. This page explains how our content is produced, what quality controls are in place, how we handle corrections, and our approach to AI transparency.
How Our Analysis Is Produced
All analysis published on Convex is produced by the Convex Research Desk, an AI-powered editorial system. Our models, built on Claude by Anthropic, synthesize data from 300+ live economic indicators to produce original analysis. The AI is our analytical engine, not a shortcut: it cross-references multiple data sources and analytical frameworks to generate research that reflects the current macro environment with specificity and rigour.
We publish two types of analysis: longform editorials (800–1,300 words) that provide deep, thesis-driven analysis of macro themes, and breaking analysis (300–600 words) that deliver rapid-response analytical takes on significant market developments, typically within minutes of an event.
Data Sources
Our analysis draws on primary economic data from authoritative sources. Every claim and statistic in a Convex article is grounded in this data pipeline:
300+ time series covering Treasury yields, employment, inflation (CPI, PCE, PPI), industrial production, housing, consumer sentiment, financial conditions, and liquidity metrics.
Auction results, issuance data, and fiscal balance data.
Nonfarm payrolls, unemployment rate, CPI components, and wage data.
Crude oil inventories, production data, natural gas storage, and energy market fundamentals.
Net speculative positioning across commodities, currencies, rates, and equity index futures.
Bitcoin and Ethereum network metrics, exchange flows, and derivatives positioning via CoinGecko.
Curated financial news from major wire services and publications, processed for event detection and significance scoring.
Analytical Framework
Our analysis is grounded in established macro-economic frameworks, not pattern-matching or sentiment aggregation. Key pillars of our analytical approach:
- Macro Regime Classification
We classify the current economic environment into one of four regimes, Goldilocks, Reflation, Stagflation, or Deflation, based on the interaction of growth, inflation, and monetary policy data. This regime framework shapes every piece of analysis we publish.
- Cross-Asset Correlation Analysis
Markets do not move in isolation. Our models track correlations across rates, equities, commodities, credit, and currencies to identify divergences, contagion risks, and positioning imbalances.
- Central Bank Policy Tracking
Federal Reserve communications, dot plots, and policy rate expectations are integrated into our macro model. We track the gap between market pricing and our assessment of likely policy outcomes.
- CFTC Positioning Data
Net speculative positioning across major futures markets provides a contrarian signal layer. Extreme positioning often precedes reversals, and we incorporate this data into our analysis systematically.
- Significance Scoring
Not every data release warrants analysis. Our proprietary scoring system evaluates signal magnitude, cross-asset convergence, topic freshness, and event severity to determine whether a development merits publication, ensuring we publish when there is something meaningful to say, not on a fixed schedule.
Quality Controls
Every article published on Convex passes through a multi-stage validation process designed to ensure accuracy, originality, and editorial quality:
- Articles are only generated when our significance scoring system identifies a meaningful development worth analyzing, preventing noise and ensuring every piece has a clear analytical purpose.
- Every article undergoes automated fact-checking against our live data sources. Specific numbers, rates, and percentages cited in articles are validated against the underlying economic data to catch discrepancies.
- Our editorial review system evaluates each article for coherence, logical flow, appropriate tone for professional financial analysis, and consistency with the current macro regime. Articles that are over-hedged, internally contradictory, or tonally inappropriate are rejected.
- Duplicate and overlapping coverage is automatically detected. We enforce topic diversity to ensure we do not publish repetitive analysis on the same theme, even when the underlying data changes incrementally.
- Articles that fail any quality check are rejected and never published. Our system logs rejection reasons for ongoing calibration and improvement.
AI Transparency
We are fully transparent about our use of AI. All articles on Convex are generated by AI (Claude by Anthropic) from live economic data. Every article includes a clear disclosure at the bottom of the page identifying it as AI-generated content.
The AI is our analytical tool, in the same way that a Bloomberg terminal is a tool for a trader or a quantitative model is a tool for a portfolio manager. The Convex Research Desk publishes the analysis; the AI powers it. Our models are configured and supervised to produce analysis consistent with established macro-economic frameworks, not to generate speculative content or sensationalized commentary.
We do not use AI to rewrite or summarize existing news articles. Our content is generated directly from primary economic data, producing original analysis rather than aggregated or paraphrased reporting.
Financial Disclaimer
Convex provides economic analysis for informational purposes only. Our content does not constitute financial, investment, tax, or legal advice. We do not recommend buying, selling, or holding any security or financial instrument.
Past performance of any asset or strategy discussed in our analysis does not guarantee future results. All investment and trading decisions carry risk, and readers should consult qualified financial professionals before making decisions based on any analysis.
While our quality controls validate specific data points against source databases, AI-generated analysis may contain errors or reflect incomplete information. Our content is not a substitute for professional financial advice and should not be relied upon as such.
Correction Policy
Accuracy is fundamental to our credibility. If you identify a factual error in our analysis, an incorrect data point, a misattributed statistic, or a logical inconsistency, please contact us at mail@convextrade.com.
We will review the report and, where warranted, update the article with a correction notice. Corrected articles display an updated modification timestamp. We do not silently alter published analysis, corrections are transparent and traceable.
Editorial Accountability
The named editor responsible for content quality is Ben Bleier, the founder of Convex. Ben designs the quality gates listed above, audits the AI desk's output, and signs off on what gets published. Editorial concerns, factual disputes, and correction requests should reference the editor by name.
The Convex Research Desk is the AI editorial system that drafts articles. It is not a team of human journalists. Every article carries the dual byline "By Convex Research Desk · Edited by Ben Bleier" so the AI/human roles are unambiguous.
Funding & Ownership
Convex is independently founded and self-funded by Ben Bleier. There are no outside investors, no advertisers, no affiliate revenue, and no paid placements influencing editorial coverage. The site is free during beta. There is no paywall and no premium tier with conflict-of-interest implications for indexed content.
If this changes — for example if Convex takes outside investment, introduces sponsored content, or begins charging for analysis — we will disclose the change here and date the disclosure.
Conflict-of-Interest Policy
Articles published on Convex frequently discuss specific tickers, asset classes, and policy outcomes. The editor's personal positions do not influence which assets get covered, what the analytical conclusion is, or how scenarios are framed. Coverage is driven by significance scoring and macro relevance, not by holdings.
Where the editor holds a material long or short position in an asset that is the primary subject of an article, that position will be disclosed inline at the top of the article. Material is defined as more than 1% of the editor's investable net worth in a single ticker. No such disclosures are required for articles where the discussed asset is one of many cross-referenced data points (e.g., gold mentioned alongside twenty other commodities in a regime classification).
Questions about our methodology or editorial process? Contact mail@convextrade.com or visit our about page.