Proof, not percentages.
Trade less.
Hold longer.
Crossview ranks the liquid universe at the horizons where the noise clears: which names carry the drift, which regime you are in, and whether to wait for the pullback. Read only.
One screen. Every market. Today's answer.
Every morning the run ranks every liquid name and tells you what is worth doing over the next 1, 3 or 7 days. Each market gets its own model, trained on its own history.
Markets
Every name, ranked
How every name moved over 1, 3 and 7 days, which moves are smooth and which are wild, the model's forecast, and where each name ranks against the field. Sort any column, tap any name.
Pro
The model's picks
The model's top tenth by forecast, which names entered today and which left, and the ones it would stay out of.
Pro
Pairs that snap back
Two names that usually move together, measured so we know the link is real, and the gap between them when one has drifted too far. Rebuilt fresh each day.
Events
What the crowd expects
Prediction markets priced into one daily snapshot: what the crowd thinks will happen, and how much money is behind it.
Long read
The month-ahead read
Which names are worth holding for a month, what kind of market you are in right now, and the one piece of guidance per name: wait for the dip, or hold off.
Chat
Just ask
Ask what to buy, sell or hold. The answer comes from the same numbers on your screen, never from thin air.
Every option the book carries is listed in one overview on the Markets page, in three groups: longs, shorts and others, the last holding the exits, the rotations and any leg still unproven. Each row carries the name's own chart beside its decision inputs, so you can compare the options at once before choosing which to act on.
Today's book
Every morning, for US, ASX, SGX and LSE stocks and for currencies: the same five lists. What is new, what is held, what is closing. You see what changed in ten seconds, without re-reading a screen.
Today's book
Five short lists a day
New buys, new sells, what to keep holding, what to close, and the swaps, never more than five names each. Every name carries its suggested size and the measurement behind it.
Swaps
Close one, open another
When one name leaves the book and another takes its place on the same day, you see them as a pair, with the opened name's measured 30 day record beside it.
Sizing
Sizing from an account budget
Equities are 1x, 2x or 3x by 30 day vol. fx is sized from an account level gross notional budget, a multiple of equity, and the names inside the budget are equally weighted. Four profiles carry the book's own measured drawdown from the starting capital: cautious 1.0 at 4.42%, standard 3.0 at 12.80%, aggressive 5.0 at 20.58%, and the broker ceiling 12.984 at 46.34%. The profiles are selected by the house bankroll rubrics rather than by Sharpe, and those rubrics would pick about 27x, above what a broker permits. The book loses half the starting capital at 14.4x and first liquidates at 45.95x on 22 June 2023, both past the ceiling. The broker's own initial margin on the major currencies runs 2.5% to 3.8%, and applied to this book's currency mix it gives the 12.984x ceiling. At 50:1 a single 200 bps session move consumes the whole of the posted margin. The size an account starts on is a rule rather than a fixed constant: the profile with the highest williams score among those whose drawdown from the starting capital stays at or under 25%, which today is aggressive at 5.0. A defensive tape opens no new longs and carries a held long at 1x; the shorts are the defensive trade and stay open at the regime weight.
Hit rates are the measured classes' own legs at 7 and 30 days, with the null p values beside them. fx fades the 2.0 z-score band: a long when the close is 2.0 standard deviations below its trailing 60 session mean, a short above, a 30 session hold. The numbers beside the fx book are the shipped selection's own, the five largest absolute z60 a leg: 63.9% hits and +32.1 bps net of fees over 1,111 trades (docs/research/2026-09-26-fx-strategy-ideas-measured.md). The older P1 measurement of the same band on the seven majors reads 55.7% hits and +27.9 bps (docs/research/2026-09-22-fx-range-edges-measured.md). Unproven means the leg does not beat its own null and drops one step, never below 1x, and on fx it means the fade measurement is missing from disk. The book shows hit rates and a winsorised mean, never raw means: on eq_lse the top 1 percent of name-days carry 99 percent of the P&L. Every class reads its own venue regime file, fx included, and fx has no news feed yet, so its news read stays ungrounded until one exists.
The playbook
The layer above the daily book: what each market actually is, which measured edges apply, and which strategy families are still proving themselves in paper.
Range markets
fx gets fades
fx is a range market: 98.1 percent of 30 day windows close within +-10 percent of where they started, and the drift sits near zero. The measured edge is the fade at the 2.0 z-score band on the majors, a 30 session hold: the P1 majors measurement put it at 55.7% hits and +27.9 bps net of fees over 1,214 trades, null p 0.00 (docs/research/2026-09-22-fx-range-edges-measured.md).
Range vs tails
Why equities get bands
Each market gets the edge that fits its shape. fx ranges, so it gets fades. Equities are a tails market: the middle of the distribution is noise and the top and bottom deciles carry the signal, so equities get tails bands instead.
Paper status
In-sample until the forward twin
Every strategy family is in-sample until its forward twin, so the playbook shows paper status, paper day N of 30, never a backtest number as a promise. Sub-day scalps are suppressed regardless of pnl: they contradict the measured doctrine.
Same posture as the rest of the terminal: a family in paper is being tested, not recommended. The playbook states the market, the measurement and the paper day, and the decision stays yours.
Why the longer view
Short windows are mostly noise. The measured edge lives at weeks, and the regime decides the entry.
Noise
A signal measured over a few days is mostly noise. Our house paper on this is in the writing section: in-sample metrics predicted held-out outcomes at 0.077 and below, within noise of zero.
Weeks
The ranking model knows more about the next month than about tomorrow, and sentiment carries its edge at a 30 day hold on liquid names. That is the horizon the terminal is built for.
Entry
Quiet tape, wait for the pullback. A strong trend, sentiment is crowded. Defensive tape, hold off. The entry rule is conditional on the regime, not on enthusiasm.
Why people subscribe
The benefit is focus. One ranked view of every liquid market before your day starts, and no conflict of interest: we sell the screen, never the trades.
Hours back, every day
The daily build ranks every name and every pair before you sit down. Your morning starts at the shortlist, not the screener.
Know what is priced in
Event markets show what the crowd already expects. Your edge is where the book disagrees with your view, not where you are merely bullish.
Right horizon, right entry
Every name carries its measured horizon and its regime-conditional entry guidance. The question becomes what to hold and when, not whether the screen is green.
One surface, all asset classes
Crypto, US, ASX, SGX and LSE equities, FX and the merged eq-all view, each with its own model trained on its own history.
Honest, auditable claims
Every model ships its holdout metrics in its manifest. When a model is weak, the product says so instead of painting over it.
The terminal filters, you decide
The models rank, exclude and time. What you bring is the outside view: what changed in the world, what you can afford to lose, and whether this quarter looks like the ones we measured.
The maths, in plain words
Enough to judge the machinery, not the calibration. Every number below is computed from data available before it is acted on, which is the whole point of a point-in-time pipeline.
Causal pipeline
No lookahead, ever
Every feature is built on a strict trailing window. Nothing sees the future of the bar it scores. A pipeline that leaks looks great in backtests and loses in production, so we ban the pattern outright.
Forecasts
Where the drift points
The daily model reads each name's direction and how wild it swings, then ranks it against today's whole field. Strong means strong relative to everything else, today.
Pairs
Linked, not just similar
Two prices that merely look alike can drift apart forever. Two prices that are truly linked snap back. We test for the real link before we ever show you the gap.
The map
Names that behave alike
Every measure of every name folded into one picture. Names near each other have behaved alike across returns, shape, volume and forecast. You see the clusters a spreadsheet cannot show you.
Shape
How the move moved
Trend fit, worst drawdown and jumpiness over 60 sessions describe the character of a name, not just its return. A clean grind and a violent spike are different trades.
Events
Implied, not asserted
Event probabilities come from liquid prediction markets, not from us. We curate the snapshot daily and show the price, the volume and the disagreement.
Access
Crossview is in testing. The people using it now are testers, not customers, so access is free and there is nothing to buy. The plans below are what we expect to charge once testing ends, published so you can see where this goes.
Free while we test
Sign in with your email and the whole terminal opens: every asset class, the GBM book, pairs, event markets and the chat rail. No card, no countdown, no seat to buy.
Expected pricing after testing. Annual billing is ten months for twelve. Nothing here is on sale yet.
Core
$249
per month
- The leaderboard and series charts
- News and sentiment
- Scatter and UMAP
- Event markets
- Timeframe insights
Professional
$499
per month
- Everything in Core
- The GBM book
- Pairs
- On-chain panels and chat
Institutional
$999
per month
- Everything in Professional
- Three seats
- Priority support
The read-only API is priced on request, to the deployment: seats, request volume and distribution all count. [email protected]
Not a charting platform. Crossview is a read-only decision surface and deliberately stays out of TradingView territory.
Honesty and limits
Read only. No orders, no execution, no advice, no spread, no flow, no kickback. Forecasts are one input among many. Nothing here is a recommendation. Models vary by asset class and their holdout metrics ship in each model manifest. The models rank and filter. The decision, the size and the stop are yours.
Questions, answered
How fresh is the data?
Crypto builds hourly. Equities and FX build each night at 22:30 UTC for the latest session, and use the latest session on or before the date you pick.
Can the terminal trade for me?
No. The terminal is read only. You make the decisions and place orders yourself.
Is this financial advice?
No. Every number here is one input among many.
What do I add that the models cannot?
The outside view. Models only know the history they were trained on. You know what changed in the world this week, what you can afford to lose, and whether this quarter looks like the ones the measurements came from. The terminal ranks and filters. You decide, size and stop.
What is UMAP?
A two dimensional map of all metrics together. Names near each other behaved alike across the return ladder, shape, volume and forecast. It shows clusters, and is not a signal by itself.
What is a rotation?
One action that closes a name leaving a leg and opens the name that takes its place in the same leg that day, shown as a pair. The panel carries the opened leg's winsorised 30 day mean beside it.