Stop guessing where prices will go. Start measuring what your book can survive.
Every morning, for the stocks you actually hold: how bad the next two weeks could get, which names would do the damage, which ones jump without warning, and what weights stop any one stock from taking over. No price predictions, and a published report card on how often our ranges held.
89–93% (research books, one year: 2026, 18 two-week windows per book)of real fortnights inside our self-corrected 90% range
Five research books of 8 to 50 stocks, one year
3.9–5.3% (research books, one year: 2026, 36 weekly rebalances per book)worst drawdown of the risk-parity research books
Books that chased forecasts: up to 46%
5 of 5 (research books, one year: 2026, 36 weekly rebalances per book, after 10 bp costs)research books where risk parity beat equal weight after trading costs
Research books, one year
30 minbetween refreshes of how your stocks move together and how much they move
Coming for desks (Tier 2 and 3)
Research books: 2026 out of sample, one year, 18 two-week windows and 36 weekly rebalances per book. Every product below states its limit next to its proof. The demo book above is live and refreshes after every US close.
The idea
We measure the storm, not the weather.
A price target feels useful and fails quietly. A range can be checked, fortnight after fortnight, and corrected when reality runs wider. That is the whole product in one picture.
A price target: one line, one guess
Right about half the time. In our tests, direction calls made from recent prices were about a coin flip. So we don’t sell them.
A range: how far the next two weeks could reasonably go
Checked against reality. In 2026, real fortnights landed inside our 90% range 89–93% (research books, one year: 2026, 18 two-week windows per book) of the time, across five research books.
Think of a hurricane cone. It never says exactly where the storm will land. It says how wide the danger zone is, and forecasters publish how often storms stayed inside it. We do the same for your money.
Products
Eleven products. Each answers one question.
Every product is built around a question a portfolio manager actually asks. The answers arrive on one page per book; the live examples below are the public demo book.
Products 1–4
The bad-morning page
When a sell-off starts, a manager asks three things: how bad can the next two weeks get, who causes it, and which of my stocks are hidden gap risks. These four answer them on one page per book.
Product 1 · Daily Joint Stress Generator
If the next two weeks go badly, how bad, and how does it unfold?
Every day we play out joint scenarios of the next fortnight for your exact book, with your stocks moving together the way they have lately. Then we pull out the worst one in twenty and the worst one in a hundred, and show them as money and as a day-by-day path. Not “risk is elevated”: a number you can act on.
Proof.
Real two-week outcomes landed inside the self-corrected 90% range 89.4–93.1% (research books, one year: 2026, 18 two-week windows per book) of the time on five research books, and losses past the 1-in-20 line stayed in the regulator’s green zone on all five (research books, one year).
Limit.
In a liquidity panic that runs past recent co-movement, losses can still run past the bad-fortnight average.
For.
Portfolio managers, risk officers
Live: the chart at the top of this page is this product, on the demo book.
Product 2 · Expected Shortfall with Named Culprits
Which of my holdings would do the damage?
Inside those worst fortnights, we split the loss by stock. You see each holding's share of the money next to its share of the damage, so you know where to trim: a few lively names can do far more damage than their share of the money.
Limit.
For reporting and trimming by hand. Shares of a tail loss need a large sample of fortnights to settle, so they never feed an automatic optimiser.
For.
Portfolio managers, risk officers
Today each book shows the names carrying the most risk; the split of the bad-fortnight loss itself arrives with the deep sample. See it on the demo book.
Product 3 · Concentrated-Position Risk Sheet
How exposed is my biggest position, and will my stop get hit?
For any position: its 1-in-20 and 1-in-100 ten-day loss, how often it has violent half-hours, and the chance it touches your stop at any point in the next ten days, not just at the end. A stop gets touched far more often than the stock ends below it, because prices wander on the way.
Proof.
Single-stock ten-day ranges held 89.4–93.1% (research books, one year: 2026, 18 two-week windows per book) of real outcomes after self-correction (research books, one year).
Limit.
The first half-hour after the open is still under-ranged, so the one-day band is a guide, not a 90% promise. The middle of the range is never a price target.
Which of my stocks look calm but jump without warning?
Two stocks can have the same volatility and completely different behaviour: one moves a little every day, the other sits still and then gaps. A covariance matrix gives them the same weight. The card reads five years of half-hours for each holding and flags the gap-driven ones.
Rules.
A stock is flagged only when at least two of three jump measures (fat tails, the largest half-hour, how fast the tail falls off) sit far outside a 97-stock reference, and it stands out within your own book. High volatility or a fast rise alone never sets a flag.
Limit.
The card describes shape, never direction or return: it says how a stock's risk arrives, not whether to own it.
For.
Portfolio managers, CIOs
Loading the live demo book...
Products 5–6
Desk add-ons
Once the core page is trusted, these two bring the same measurements to hedging and to the trading day. Both are in development, for Tier 2 and 3.
Product 5 · Tail-Hedge Overlay SizerIn development
How much index protection do I need to cap a bad fortnight?
Pick the most you are willing to lose in a 1-in-20 fortnight. We measure how much index protection would have paid off across your book's own bad fortnights, and size it to bring the loss back to your limit. Hedges sized from average betas are measured on ordinary days; this one is measured in the bad fortnights, which is when you need it.
Limit.
Sizes protection; it does not price options, which needs market option prices. Because the rarest market-wide drops are under-counted, the sized hedge is a minimum.
For.
Derivatives desks, portfolio managers
Product 6 · 30-Minute Intraday Risk DeskIn development
What is my book doing right now, half-hour by half-hour?
How your stocks move together and how much they move refresh every 30 minutes, and the ten-day range re-anchors to the live price. The opening half-hour is treated as what it is: the most violent half-hour of the day. A bell-curve model treats every half-hour the same.
Limit.
Opening-half-hour ranges are execution guides, not calibrated bands.
For.
Execution and risk desks
Products 7–8
Easy first steps
Nothing to connect to your trading systems. A risk officer or analyst can start with these right away, free.
Product 7 · Continuous Five-Check Self-Audit
Can I trust these numbers?
Every build is compared with what happened, on five checks: do the stocks move together the right amount, do they fall together, are moves the right size, did the range hold, and how often was the 1-in-20 loss line crossed. If reality has been running wider than our range, the range widens automatically, and you see by how much. It can also audit the loss estimates from your current vendor.
Proof.
Before correction the range held 81.5–83.0% (research books, one year: 2026, 18 two-week windows per book) of fortnights on the 16- to 50-stock research books; the correction, using only earlier fortnights, brought that to 91.8–93.1% (research books, one year: 2026, 18 two-week windows per book). The 8-stock book needed none (research books, one year).
Limit.
It corrects after the first stretched fortnight, not before. One year is 18 fortnights: about ±15 points of uncertainty on coverage. A vendor's loss series supports the breach count, not all five checks.
For.
Risk officers, compliance
Loading the live demo book...
Product 8 · Pre-Trade Universe Screener
Before I buy it: is this stock smooth or jumpy?
Drop in any list of tickers. In about a second, without running a single simulation, each one has its shape measured against a 97-stock reference: how fat its tails are, how violent its worst half-hour was, how much of its movement arrives in sudden jumps. Same index, very different animals.
Limit.
Shape only: no view on valuation, quality or direction.
For.
Analysts, portfolio managers
Kurtosis of 30-minute returns, five years (log scale)
Log scale; higher means more of the movement arrives in rare, large half-hours.
Weights set by risk alone, every week, from measured co-movement and volatility, never from a forecast.
Product 9 · Analytic Risk-Parity Rebalancer
What weights stop any one stock from taking over my risk?
Every week, weights are set so each stock contributes the same share of the book's swings, using the measured co-movement and volatility of your stocks, never a return forecast. Equal money is not equal risk: the wildest names carry several times the risk of the calmest. Risk parity evens that out.
Proof.
Beat equal weight on 5 of 5 (research books, one year: 2026, 36 weekly rebalances per book, after 10 bp trading costs) research books after costs, trading 12–25% of the book a week. Worst drawdown 3.9–5.3% (research books, one year: 2026, 36 weekly rebalances per book), while books that chased forecasts fell as much as 46% (research books, one year: 2026, 36 weekly rebalances per book) (research books, one year).
Limit.
The edge over equal weight is modest, about +0.1 to +0.4 Sharpe at a weekly cadence, and it disappears if you rebalance daily. One year carries about ±0.6 Sharpe of noise.
For.
Allocators, self-directed investors
Loading the live demo book...
Products 10–11
Before you trade
Two tools that turn the same measurements into a decision you are about to make.
Product 10 · Diversification X-Ray
I own 24 stocks. How many bets is that really?
We measure how much your holdings move together, translate it into the number of independent bets they add up to, and show where the risk actually sits: which groups of names move as one, their share of the money and their share of the risk.
Limit.
Groupings shift as markets change, so the X-ray refreshes with every daily run. Built on the same measured co-movement as the rebalancer.
For.
Self-directed investors, portfolio managers
Loading the live demo book...
Product 11 · What-If Trade Preview
What will this trade do to my risk before I place it?
Move a weight and see, before you trade, how that stock's share of your risk, your book's swing and your number of independent bets change. Raising one stock from 4% to 12% of the money can make it a far bigger share of the risk.
Limit.
Built on the same co-movement and volatility as the rebalancer, so the preview and the weekly weights never disagree. It previews risk, not return.
For.
Portfolio managers, self-directed investors
Loading the live demo book...
Compared
Compared with the tools you already know.
Factor models describe risk with a matrix. Bell-curve simulations draw paths with thin tails. We draw your book’s paths with each stock’s real habits, and grade ourselves in public.
What you get
Factor model (for example Barra)
Bell-curve simulation
CausaQuant books
The output
A covariance matrix and betas, no paths
Paths with thin, symmetric tails
Joint ten-day paths for your exact book, each stock with its own shape
How often it refreshes
Daily or monthly
When someone re-runs it
Full paths every trading day; a 30-minute refresh is coming for desks
Extreme half-hours
Not modelled
Barely modelled: thin tails. Volatility models (the GARCH family) overshoot instead: when one stock's variance blows up, worst half-hours come out 25 to 80% deeper
Modelled from each stock's own five years, with worst cases that stay bounded; the rarest are still under-counted
The opening half-hour
Not modelled
Treated like any other half-hour
Modelled as the day's most violent half-hour, still narrower than reality
Stocks falling together
Not modelled
Understated: bell-curve coupling
Modelled with heavy-tailed coupling, and checked on every build
Ten-day range for a diversified book
Not produced
About as good, once self-corrected
About as good, once self-corrected, and published with its report card
Gap-prone stocks
Invisible: variance only
Invisible
Flagged on the jump-risk ledger and in the screener
Published report card
None
None
Five checks on every build, every fortnight scored
Adding a stock
Factor mapping and refit
Re-estimate the matrix
A rebuild of your book, included in every tier
Limits
What we promise, and what we refuse to fake.
Anyone can sell a crystal ball. We sell measured ranges, with every limitation stated next to the number it limits.
What you can count on
Ten-day ranges that hold.89.4–93.1% (research books, one year: 2026, 18 two-week windows per book) of real fortnights landed inside the self-corrected 90% range, on five research books from 8 to 50 stocks.
Sell-offs where stocks fall together.Our paths let your stocks drop in the same half-hour far more often than a bell-curve model does, and whether they fall together often enough is one of the five checks on every build.
A loss line that passed the regulator's test.Breaches of the 1-in-20 line stayed in the green zone on all five research books.
Calm, low-turnover rebalancing.Risk parity beat equal weight on all five research books after costs, with worst drawdowns of 3.9–5.3% (research books, one year: 2026, 36 weekly rebalances per book).
What it cannot do
Forecast direction.The middle of any range called up or down about as often as a coin flip. We don't sell it, and the middle of the range is never a price target.
Use simulated paths to pick weights.Feeding simulated tails or volatility into an optimiser never beat plain risk parity out of sample. So the simulator measures risk and culprits; the weights come from measured co-movement and volatility.
Match the rarest half-hours.The most extreme half-hours appear less often in our paths than in real markets. In a liquidity panic losses can run past the bad-fortnight figure; the jump badges flag the stocks where this matters most.
Promise a one-day range.The first half-hour after the open is still under-ranged, so one-day bands are guides, not 90% promises. The ten-day range is the calibrated one.
Correct before reality moves.The range widens after the first stretched fortnight, not before it.
Escape a short history.The research is one year: 18 fortnights (about ±15 points on coverage) and 36 weekly rebalances (about ±0.6 Sharpe). A multi-year walk-forward is the next milestone.
Who it is for
Built for two kinds of buyers.
The portfolio manager with an eye for tape
You already have conviction in your names. You need an honest X-ray of how they behave together under stress: which one quietly carries gap risk, how wide the next fortnight really is, and what weight change stops one volatile name from running your P&L.
The serious investor building wealth at home
The discipline of an institutional risk desk for your own book: a morning page that says how bad things could get, a warning when our own range has had to stretch, and weekly weights that keep drawdowns small.
See your own book on one page.
Send us your tickers and weights: the data check is free, and your first morning page follows the build. Beta members build their first book free.