Quandiction
Method
Provable,
not
A laboratory for time-series forecasting on live markets. Every prediction is committed before the outcome, published with the outcome — including the ones we get wrong — and verifiable without us.
Est. 2026
Pre-gate · edge not proven
Open-proof forecasting
Over a long horizon you don't need an edge — only faith. Over ours, faith is not required. Only measurement.
What open-proof means
Open here is not the code. It is the proof.
Anyone can publish a chart of what they would have predicted. Almost nobody publishes what they did predict, at the moment they predicted it, next to what actually happened. The whole design exists to make the second thing structurally unavoidable.
Before
The protocol is declared
What gets measured, on which data, over how much forward time, after which costs, and what counts as a failure — all written down before any result exists. Changing it later means starting a new record, in public.
During
The prediction is committed
Each forecast is written to an append-only ledger the moment it is made, with a hash. There is no edit path and no delete path. A bad call cannot quietly stop existing.
After
The record is sealed
The ledger is anchored to the Bitcoin blockchain with OpenTimestamps. You can verify the entire history — outcomes, misses, abandoned hypotheses — without our cooperation and without trusting us.
What this page does not have
Not signals. Not trades. Only proof.
Nothing to show yet — and that is the point. The system is in build. Until it passes its own exam in public, there is no result to display; we put that in writing before it became inconvenient.
Signals to tradeA subscriptionPerformance numbersA backtest on the storefrontAdvice of any kind
The gate
One exam stands between here and the word “works”.
A live-market test whose rules were fixed in advance: the forward window, the metrics, the cost model, and the definition of failure. No result from before those rules were written can be used to pass it.
Backtests are the currency this industry has been paying in for twenty years, and it has lost its value. The gate is a refusal to pay in it. If the system fails, the failure gets published in the same format a success would have — a project that cannot show a zero never built the machine to measure one.
The open question
One number decides it. Nobody has measured it.
Two reservoirs, identical in every respect but one: one is classical, one is quantum. Same inputs, same readout, same costs, same judge. The difference between them — measured forward, on live markets, under rules fixed before the first prediction — is the entire question.
Δ = QRC − ESN · net of costs
[ not yet measured ]
Primary endpoint · pre-registered · publishes at the gate, either sign
Nobody knows the sign of this number. We intend to find out.
Sealed until the gate
These pages exist. They are simply not open yet.
They are not ideas for later — each one is the public face of something already built or already decided. They open on the day the exam is decided, whichever way it goes.
Archive
Every prediction, every outcome, every hypothesis we dropped — with its timestamp anchor.
Opens at the gateOracles
The leaderboard of configurations. Names and track records public, including the retired ones.
Opens at the gateIndices
Fragility, Liquidity, Regime. Open formulas. Free. Context, not signals.
In preparationMethod
The whitepaper, the metric spec, and the pre-registration in full — so the exam can be judged, not believed.
In preparation
Waitlist
Two emails. No more than that.
One when the exam begins and its rules go public. One when it is decided — pass or fail, same email either way. That is the whole arrangement.