PhD-led deep-learning research ยท benchmark-proven

The market's next 48 hours,
candle by candle.

PRISM is a proprietary multi-timeframe neural forecasting engine that predicts full OHLC candles โ€” with direction probability and calibrated confidence โ€” across 11 global markets simultaneously. Benchmarked head-to-head against the strongest models in the field. It won.

See it live โ†’ Request API access
11markets
4timeframes
254candles / forecast
3.8M+candles ingested
48hhorizon
Benchmark results

We put it against every class of model that matters.

The new wave of AI foundation forecasters โ€” the class of Kronos, Chronos and TimesFM, of which we benchmark the strongest openly runnable, Kronos, head-to-head. The classical quant standards every trading desk still runs โ€” RiskMetrics EWMA, GARCH-family AR volatility, random-walk and rolling-median baselines. And our own previous 54-million-parameter deep system. Identical out-of-sample windows, identical markets, no excuses: PRISM forecasts the entire 48-hour path with the lowest error across the intraday spectrum and wins the overwhelming majority of head-to-head matchups.

PRISM vs classical quant models โ€” full-path error
Full 48-hour path error vs the classical quant standards. Lower is better.
Head-to-head market wins vs Kronos and the previous deep system
Head-to-head wins vs the AI foundation model and our previous deep system.
11 / 11markets won vs Kronos on 4-hour candles
11 / 11markets won vs our previous 54M system on 1-hour candles
โˆ’50%error vs the previous production system on 4-hour candles
Model report

Evaluated like a frontier model. Reported like one.

Big-league AI models publish their numbers โ€” so do we. Every figure below is measured on strictly held-out, chronologically future data the model never trained on.

60%peak error reduction vs the no-skill baseline (daily range)
11 / 11markets with a positive edge on 1-hour candles
254exactly-consistent candles per forecast, 4 timeframes
ร—3independently trained experts vote on every candle
TimeframeRange MAEError reduction*Close-path MAEClose-path RMSE
15 minutes0.079%โˆ’54%1.04%1.98%
1 hour0.147%โˆ’58%1.05%2.00%
4 hours0.307%โˆ’58%1.10%2.07%
Daily0.765%โˆ’60%1.33%2.43%
* per-candle range error vs the no-skill baseline, averaged across all 11 markets and the full 48-hour horizon, out-of-sample.
Validation protocol โ€” what professional models require, and what we do:
โœ“ Chronological walk-forward evaluation โœ“ Zero-leakage input audit โ€” every feature provably known at decision time โœ“ Out-of-sample-only reporting โœ“ Adversarial multi-model benchmarking โœ“ Deep ensembling โ€” 3 independent experts โœ“ Live re-anchoring every 15 minutes
Under the hood โ€” without giving away the keys

A research-grade architecture, engineered end-to-end.

PRISM is not a fine-tuned public model. It is a ground-up, proprietary architecture developed by PhD-level deep-learning researchers over months of controlled experiments โ€” leakage-audited data pipelines, walk-forward validation, adversarial benchmarking. The exact mechanisms stay in the vault; the building blocks below tell you why it works.

Market ingestion
11 markets ร— 4 timeframes
price ยท volume ยท flow ยท sessions
Nร— attention blocks
Multi-Head Self-Attention
Normalization ยท Gated FFN
Multi-Head Self-Attention
Normalization ยท Gated FFN
multi-resolution encoders,
weights shared across markets
Fusion core
global cross-market
attention state
48 temporal queries
every future hour decodes
its own latent state
Emission heads
OHLC path engine
Direction probability
Confidence ฯƒ
Consistency lock
ร—3 expert ensemble
votes on every candle
Deep learning research
Prediction power, quantified

It doesn't just predict. It knows how sure it is.

Every one of the 254 forecast candles carries its own calibrated uncertainty โ€” tight when the model is confident, wide when the market is treacherous. And the edge is not one lucky market: it spans metals, FX majors, indices, energy and crypto.

Calibrated confidence by horizon
Self-estimated uncertainty grows honestly with the horizon โ€” per timeframe.
Per-market error reduction
Error reduction vs the statistical baseline, market by market (4-hour candles).
New ยท AI news briefing

The model reads the chart. Now something reads the news beside it.

A forecast tells you where the numbers point. It does not know the Fed spoke an hour ago. So every symbol now carries a short briefing that puts the two together โ€” market headlines read against PRISM's own forecast, refreshed through the day.

โœฆ What the news means for XAUUSD
โ—† neutral medium risk confidence 45%
Fed hawkishness lifts dollar; gold forecast flat despite rate risk

Warsh's accommodation language pushed the 2-year yield and the dollar higher, typically headwinds for gold โ€” yet the model predicts almost no movement over the next 24 hours.

Support zone under pressure 4513.13 โ€“ 4530.00
Resistance if the dollar fades 4560.00 โ€“ 4573.52
bearish 2-year yield and dollar spike
๐Ÿ“ฐ
Headlines that actually move your market

Central bank releases, official data and market wires โ€” matched to the instrument rather than dumped on it. A Fed decision reaches all eleven; an oil inventory number reaches WTI.

โš 
Levels, marked on your chart

Where the news and the forecast agree, and where the risk of being wrong is highest โ€” drawn as soft bands on the live chart, not buried in a paragraph.

๐Ÿ””
It tells you when price gets there

Reach a flagged level and a warning appears on the page, and in your inbox if you want it. It clears itself when price moves on.

โš–
Conditions, never instructions

It reports what the briefing said about a level โ€” never buy, sell or close. It does not know your position, and it will not pretend to.

Included on selected plans, and machine-readable through the API as /v1/briefing. See the plans โ†’

What you get

One engine. Every answer a trader asks.

โ—”Full OHLC candles

Not a single number โ€” complete open/high/low/close paths: 192 ร— M15, 48 ร— H1, 12 ร— H4, 2 ร— D1, all describing the same 48 hours, all exactly consistent.

โ–ฒโ–ผDirection probability

P(up) for every candle, from the forming bar to the 2-day close โ€” calibrated, never contradicting the drawn path.

ยฑฯƒCalibrated confidence

Per-candle uncertainty bands you can size positions with. Wide ฯƒ = stand back; tight ฯƒ = the model means it.

โ›“11 markets, one brain

Gold, silver, EURUSD, GBPUSD, USDJPY, USDCHF, NZDUSD, S&P 500, Dow, WTI, Bitcoin โ€” forecast jointly, not in isolation.

โŸณLive every 15 minutes

Self-healing data pipeline re-anchors the forecast at every new quarter-hour candle, around the clock.

๐Ÿ›กFails gracefully

Holiday? Weekend? Data outage? The engine keeps forecasting with whatever the market gives it.

Coverage

What it forecasts โ€” exactly.

11
Supported markets
XAUUSDGold XAGUSDSilver EURUSDEuro GBPUSDPound USDJPYYen USDCHFFranc NZDUSDKiwi US500S&P 500 US30Dow Jones WTICrude Oil BTCUSDBitcoin

Metals ยท FX majors ยท US indices ยท Energy ยท Crypto โ€” forecast jointly by one engine.

4
Supported timeframes
M1515 minutes H11 hour H44 hours D1Daily

All four views describe the same 48 hours and agree with each other exactly โ€” the 15-minute candles aggregate perfectly into their hour, their 4-hour block, their day.

254
Future candles per forecast
M15192 candles
H148 candles
H412 candles
D12 candles

Every candle with full OHLC + direction probability + ยฑฯƒ confidence, re-anchored every 15 minutes.

From the blog

Latest articles

All articles

See the next 48 hours before they happen.

Open the live forecast โ†’ Get it in MetaTrader 5 Request the API
๐Ÿ“ฑ PRISM in your pocket โ€” free app, no store account needed