Danelfin alternative · cited papers, not "AI"

Framler — Danelfin alternative that publishes the math instead of branding it "AI"

Danelfin's "AI Score" is a 1-10 rating with proprietary methodology. Framler is the opposite — every factor weight cites a paper, every regime threshold is documented, every conformal calibration constant is on the methodology page.

Try NVDA →Open methodology

Framler vs Danelfin

One brands the algorithm "AI". The other publishes every equation.

FeatureFramlerDanelfin
Free, no credit cardYesLimited free + paid Pro/Premium
Methodology disclosurePublic + invariantsProprietary "AI"
Score type0–100 continuous1–10 ordinal AI Score
Statistical foundation13 cited papers, one per factorAI branding, no math
Regime detectionBOCPD HMMNot disclosed
Prediction intervalsMondrian conformalNo
Universe size1,000+ curated~1,500 US-listed
Sector breakdownPer-sector pages + leaderboardsNative
Counter-AI positioningPure deterministic statsAI-marketing front

What you get on Framler

Cited math vs branded "AI"

Danelfin presents an "AI Score" 1-10 with proprietary methodology — you trust the brand, not the math. Framler does the inverse: every factor cites a peer-reviewed paper (Novy-Marx 2013 quality, Jegadeesh-Titman 1993 momentum, Bernard-Thomas 1989 PEAD, Sloan 1996 accruals, Loughran-McDonald 2011 NLP, Schmidt-Stadtmüller 2006 tail dependence, and 7 more), every weight is published, every invariant is audited live at /coherence.

Continuous 0-100 vs ordinal 1-10

A 1-10 ordinal score loses gradients — the difference between an 8 and a 9 is the same as between a 4 and a 5 within Danelfin's rating, even though the underlying signal strength almost certainly isn't. Framler's 0-100 continuous composite preserves the gradient, making cross-sectional ranking more honest.

Counter-AI positioning

Post-2026 AI fatigue, "we publish every equation and run live invariants" sells stronger than "we have proprietary AI you can't inspect". Both Danelfin and Framler scrape similar data, but the trust positioning is opposite. Framler's 26 daily crons contain ZERO LLM calls — pure deterministic statistics.

Common questions

Does Framler use AI?

Deliberately no in scoring. The 26 daily crons that compute scores contain zero Anthropic API calls — pure Bayesian statistics, conformal prediction, copula tail-dependence math. We do use Anthropic for user-triggered AI explanations of model output, but those wrap already-computed deterministic scores; they don't generate them.

Why is the universe smaller?

Framler's factor pipeline reads SEC EDGAR, ClinicalTrials.gov, options chains, short-interest data, and computes Bayesian regime posteriors per ticker. 1,000+ names is the current quota-respecting limit; coverage grows as data sources scale. Danelfin's 1500 universe relies on lighter per-ticker depth.

How do I verify the methodology?

/methodology has every factor formula and academic citation. /coherence runs 9 structural invariants live (per-day pass/fail). /track-record now shows rolling Information Coefficient — 1d + 7d horizons calibrated 2026-05-17, 30d matures ~2026-06-26.

Explore Framler

Popular tickers
Browse by sector
How it works

See the math, not the branding

Look up any of 1,000+ tickers, see the 13-factor breakdown, regime context, prediction interval, and pattern flags. Free during early access.

Try a ticker →See track record
Free Danelfin Alternative | Framler