The market has weather.
We map the climate.
History doesn’t repeat itself, but it often rhymes. Historacle reveals the patterns shaping what’s next.
Know where the cycle stands right now, see what history says usually happens next, and stress-test your view against 45 years of real market episodes — then see exactly what it would mean for your stocks and portfolio. Calibrated ranges and transparent evidence, never point predictions.
Private beta · Macro data from FRED & EIA, refreshed daily · Invite-only access
Built on authoritative data sources
Turn a macro view into a historical evidence check.
Name a recession, oil shock, Fed pivot, or soft landing and Historacle pulls from the projection-eligible record, then shows the range similar moments actually delivered. The output is a conditional scenario range you can audit through the facts behind it.
A live taste of the matching engine. Inside, the full instrument adds regime filtering, conformal calibration, and the dual analog + statistical cones.
98-episode record · 65 projection-eligible episodes · every match transparent
Every forecast, graded.
Most forecasters never keep a report card. Historacle does — every projection and every judgement call is archived and scored against what actually happened, whichever way it lands. That discipline is what makes the ranges worth trusting.
Every model is scored on decades of real market episodes it never trained on, and has to beat simple baselines before it reaches the product. Changes that fail are not shipped.
The cone is a calibrated risk range, not decoration — sized so outcomes land inside it about as often as the label claims, and checked against what actually happened rather than asserted.
Projections are compared with the outcomes that followed and the result is kept, whichever way it goes. The record is maintained continuously, not assembled after the fact.
Historacle shows which outputs are calibrated, which beat baselines, and which are most valuable as scenario ranges, so confidence goes to evidence instead of the loudest forecast.
- 01
Replay history
Each model is run against decades of real market episodes it never trained on, so results reflect performance on the unseen, not a curve fitted to the past.
- 02
Calibrate the range
Bands are tuned so the cone reads as an honest risk range — outcomes should land inside it about as often as the label says, and that is measured rather than assumed.
- 03
Separate edge from noise
Every market is checked against simple baselines, so the product can show where the model genuinely adds signal and where the range itself is the useful output.
- 04
Show confidence clearly
Results carry confidence labels, supporting evidence, and watchpoints, so you can see how much weight an output deserves.
Calibration, guardrails, and uncertainty shown together — never a point estimate on its own
One scenario, all the way through
Set a macro view once and carry it from the big picture down to your own holdings. The macro engine is free — live stocks, your portfolio, and exports are Premium.
- 01
Set a scenario
FreeMove six core macro sliders — Fed policy, oil, natural gas, CPI, the dollar, unemployment — plus gold and copper match-only drivers, then toggle condition flags like recession or stagflation. Or start from a research-backed preset such as 2008 GFC or 1995 Goldilocks.
Try it live below - 02
See the macro projection
FreeHistoracle matches the full macro picture at once and returns projection cones across ten indicators, a 24-month semantic regime-weight strip, and every matched analog laid bare beneath the chart.
Try it live below - 03
Drill into a stock
PremiumCarry the scenario into a supported ticker — a 12-month outcome built from the market path, how the stock responds, its own risk, and any catalysts you pick — with the range of outcomes and how often it ends below today's price spelled out.
See it - 04
Stress-test your portfolio
PremiumRun your holdings under the same scenario: a macro-exposure read, a portfolio P/L cone, per-holding contribution, and an analog stress test against the worst matched episode.
See it - 05
Export the report
PremiumTake it with you — CSV and chart-image export for any cone or analysis. Every scenario also encodes to a shareable link, so colleagues open the exact view you saw — that part's free.
See the forces moving any stock under your scenario.
Historacle turns one macro view into a full 12-month stock distribution using the market path, the stock's adopted response, company risk, and selected catalysts. Valuation remains separate decision context.
Under this scenario, NVDA is projected to decline ~8% over 12 months, with an elevated 58% of illustrative outcomes below today's price. The live tool additionally shows its treatment receipt and keeps unadopted cohort effects out of the result.
Live prices · DCF & owner-earnings valuation · sector β · catalyst library — on any ticker
See where your portfolio is exposed before the scenario hits.
Run the same macro view across every holding. Historacle decomposes your exposure, draws a portfolio-level outcome cone, ranks the holdings driving the result, and compares the whole book with the harshest matched episode.
Each bar is a holding’s weight × its 12-month median under the scenario — its real pull on the portfolio.
High market sensitivity — β 1.11, ~1.1× the S&P 500. Unsupported holdings fail the portfolio cone closed instead of receiving a generic sector substitute.
Each holding’s market sensitivity and drift applied to that episode’s actual macro path — “a book like yours, then.”
Under this scenario the book is projected to lose ~15% over 12 months, with an elevated 69% of illustrative outcomes below today's price — the same support-gated stress test the live tool runs on your real holdings.
Export the report
PremiumCSV and chart-image export for any cone, stock, or stress test. Shareable scenario links are free for everyone.
Live prices · real βs · analog stress test · CSV & chart export
Real historical precedent and structured statistics
Historacle combines two complementary approaches — real historical analogs and regime-aware statistical relationships — to generate forward paths with transparent ranges and matched episodes.
Bayesian Markov regime-switching
A Hamilton filter produces normalized semantic weights over six cycle phases. The map is a context lens, not a calibrated probability forecast.
Kernel-regression decomposition
Conditional-expectation estimates give each macro driver's marginal response, with an explicit support diagnostic so thin evidence is flagged rather than hidden. The channels share historical evidence and are not independent confirmation.
Bootstrap percentile cones
Resampling across matched episodes produces fat-tailed 10–90 bands that reflect the way market outcomes actually disperse, rather than assuming a tidy distribution.
Correlation-aware, joint-coherent
A regime-conditioned Bayesian VAR keeps every channel coherent — when oil, the Fed, and the dollar move together, the projection respects how they actually co-move, so cones don't over-widen and a match has to hold across the whole picture.
Join the Historacle beta
We’re opening access in small, hand-picked batches so we can support every tester properly. Request access and we’ll email your invite when it’s ready.
- Full scenario engine — all macro sliders, condition flags, and the Bull/Bear/custom cones
- Bayesian regime detection and transparent historical-analog matching across ten markets
- Founding-tester pricing locked in when paid plans launch