Question 1

What is Probora?

Accepted Answer

Probora is an independent prediction-intelligence platform. Ten specialized agents continuously observe crypto and commodity markets and publish probabilities that an asset will finish UP or DOWN over a fixed window of 5 minutes to 1 day. Every prediction is recorded before the window settles and scored against an independent price index.

Question 2

Is Probora a prediction market or an exchange?

Accepted Answer

No. Probora does not sell contracts, take positions or match orders, and it is not affiliated with Polymarket, Kalshi or any other prediction platform. It never uses prediction-market odds. Probora creates and resolves its own standardized markets using independent external price data and publishes analytics about them.

Question 3

Which markets and timeframes are covered?

Accepted Answer

Crypto: BTC, ETH, SOL, XRP, DOGE, HYPE and BNB. Commodities: Gold, Silver and Oil. Every asset runs recurring windows on five timeframes (5 Min, 15 Min, 1 Hour, 4 Hours and 1 Day), aligned to UTC boundaries.

Question 4

Who is Probora for?

Accepted Answer

Discretionary traders who want a calibrated second opinion, systematic traders and bots that consume probabilities programmatically, and researchers who want an honest, unedited track record of short-horizon market forecasts.

Question 5

How is a market resolved?

Accepted Answer

At the exact start of each window Probora fixes the Price To Beat from the Probora Price Index, the median of independent sources (Binance, OKX and Coinbase for most crypto). At the end of the window the same index gives the Settlement Price. Higher settles UP, lower settles DOWN, and an exact tie is recorded as FLAT, never forced into a direction.

Question 6

What is the difference between Base Consensus and the Meta Agent?

Accepted Answer

Base Consensus is a transparent count of the nine specialists, for example 7 / 9 UP. The Meta Agent is not a tenth vote: it learns which specialists have been reliable for each market, timeframe and regime and produces its own independent probability, such as UP 67%.

Question 7

Why do some agents show “Insufficient Data”?

Accepted Answer

Every model instance (agent × asset × timeframe) learns from market #1, but it may only publish once it has resolved at least 100 observations. Until then the public output is Insufficient Data. Training history is never retroactively turned into public predictions.

Question 8

How are agents ranked?

Accepted Answer

Directional accuracy is shown prominently, but rankings also weigh probability quality (Brier skill versus a 50/50 forecast) and calibration (do 70% calls win about 70% of the time?), with a sample-size adjustment so small samples cannot top a leaderboard. The full definitions are on the Methodology page.

Question 9

Can predictions be edited or deleted after the fact?

Accepted Answer

No. Predictions are frozen at window start with their model version and timestamp, and every result, win or loss, stays in the public history. There is no cherry-picking and no deletion of bad periods.

Question 10

Do agents ever see future data?

Accepted Answer

No. At market start an agent may only use data available at that timestamp. Training uses chronological splits and walk-forward evaluation, so models are always tested on data they have never seen.

Question 11

Does Probora reveal how the agents work?

Accepted Answer

Probora publishes what each agent analyzes and how results are scored, but not proprietary internals such as formulas, weights, thresholds, indicator periods or feature engineering.

Question 12

Is there an API?

Accepted Answer

Yes. The REST API exposes individual agents, the Meta Agent, consensus, market and settlement data, historical predictions and performance statistics. A WebSocket feed pushes predictions and settlements in real time. The API is in Developer Preview.

Question 13

Is Probora financial advice?

Accepted Answer

No. Probora is analytics and market-intelligence software. Predictions are probabilistic, can be wrong, and past performance does not guarantee future results. Always make your own decisions.

Frequently asked questions

Product

Predictions & scoring

Data & integrity

API & pricing