01
Forecast
Published model rows are separated by asset, timeframe, issue date, target date, and horizon so the site does not mix live forecasts with after-fact scoring.
Methodology
Methodology
How ProsQuant turns forecasts into a decision environment.
ProsQuant is designed as a layered research and monitoring system: model forecasts are published first, then selection, risk review, execution state, and post-trade feedback are kept visible instead of being hidden behind a single black-box signal.
02
Selection
The public layer ranks the strongest available setup by current horizon context rather than forcing every model into one universal signal.
03
Risk review
A forecast can be directionally interesting and still be filtered if the current setup does not pass the decision and risk gates.
04
Execution state
The site shows runtime status, signal coverage, and bot state separately, so the forecast layer and execution layer remain readable.
05
Feedback loop
Closed target bars and closed trades feed the review layer, helping distinguish forecast quality from execution conditions.
Horizon discipline
Each horizon has its own context.
H=1 and H=3 should not be visually treated as the same forecast. The interface separates next-target forecasts from selected-horizon forecasts to avoid mixing tomorrow's decision with a multi-day target.
Model pools
Pools follow the published rows.
When only two model rows exist for a horizon, the public pool should show those two rows; it should not inherit a larger pool from another horizon.
Runtime separation
Forecasts and trading state are different layers.
The forecast board can publish research information even when the bot is stopped, filtered, or waiting for a qualified setup.