Voltacent Quantitative Technologies
Direction, magnitude and volatility models running live in production, available through one API.
Hybrid XGBoost classification
xgboost_20260916_103341
Product stage
RESEARCH / STAGING
Research
RUN-LEVEL PROVENANCE
Live trading
NOT CLAIMED
Telemetry
CANARY SNAPSHOT · NOT LIVE
Market data
NOT CONNECTED
| Date | Asset | Direction | Volume | Net bps | Status |
|---|
The source snapshot records 9 recent trades, but row-level details are not present in the available project copy. No rows are fabricated.
Company
A quantitative research and inference platform for systematic trading workflows.
Mission
Make model provenance, validation design and deployment status inspectable before any output reaches a risk or execution layer.
Current stage
product Stage
RESEARCH / STAGING
live Trading
NOT CLAIMED
lean Validation
NOT YET TESTED
telemetry Provider
CANARY SNAPSHOT · NOT LIVE
market Data
NOT CONNECTED
Platform
Voltacent is a single pipeline, not a bundle of features. Each stage is inspectable — select one to see its inputs, outputs, model type and failure behaviour.
Research engine
For each validation year Y+1, train on all available history from 2002 through Y, then evaluate only on Y+1.
Fold rule
For each validation year Y+1, train on all available history from 2002 through Y, then evaluate only on Y+1.
Temporal control
Point-in-time features and strict temporal causality prevent future bars from entering training or inference inputs.
LEAN distinction
This research walk-forward design is separate from QuantConnect LEAN backtesting. LEAN status remains NOT YET TESTED.
Expanding window
Train on 2002 through Y, then validate only on Y+1.
Strict causality
Point-in-time features use only information available at the bar timestamp.
Full retrain
The complete model is retrained independently for every annual fold.
Not LEAN
Research OOS validation is separate from QuantConnect LEAN backtesting.
Representation research: a causal dilated TCN autoencoder was explored for 32-dim market-regime embeddings during gen-2 research. The production triad does not use TCN latents — direction, magnitude and volatility are served from gradient-boosted models on tabular features.
Threshold research
These are historical threshold-study observations, not an operating promise or guarantee.
Threshold τ
0.50
Net Sharpe
-2.86
Repository run artifact identified by run ID
Threshold τ
0.58
Net Sharpe
+0.69
Repository run artifact identified by run ID
Threshold τ
0.60
Net Sharpe
+1.85
Repository run artifact identified by run ID
Threshold τ
0.62
Net Sharpe
+2.96
Repository run artifact identified by run ID
Threshold τ
0.65
Net Sharpe
+3.99
Repository run artifact identified by run ID
Signal Explorer
Illustrative price, confidence, regime and decision fields on one surface. No market-data or inference provider is connected.
Canary telemetry snapshot
A historical canary snapshot recorded in Voltacent telemetry.json. Values describe that recorded observation only and do not stream or imply current status.
Ledger
The source snapshot records 9 recent trades, but row-level details are not present in the available project copy. No rows are fabricated.
Inference contracts
Direction probability, expected return and volatility forecasts remain separate inputs to a downstream decision, risk and execution layer.
1# Interface example — no public production endpoint is claimed2curl -X POST https://api.example.voltacent.io/v1/inference/direction \3 -H "Authorization: Bearer $VOLTACENT_API_KEY" \4 -H "Content-Type: application/json" \5 -d '{"symbol":"EURUSD","horizon":"1H","as_of":"<ISO-8601>"}'{ "example": true, "model_run_id": "xgboost_20260916_103341", "horizon": "1H", "direction_probability": null, "production_status": "STAGING"}Output: Probability and model provenance. Values in this payload are examples, not observations.
Product lineup
Layer 01
OPEN FOR ALLA personal quant research agent that runs on your own terminal, against your own data. You own everything — your data never leaves your machine. Bundled Claude and Gemini intelligence, routed per task. Research tooling, not advice: it analyzes backtests, explains walk-forward results, and suggests features to test.
Layer 02
ADD-ONVoltacent market data plugged straight into your agent. The retainer that keeps the agent fed.
Layer 03
ACCESS ONLYThe production model triad — direction, magnitude, volatility — served as an API. Models are never distributed; API keys or nothing. Request access for the private beta.
Pricing
Private beta is free while we validate. Prices below are the planned tiers at launch — nothing is billed today.
Free
Snapshot API data, docs, playground. No live data.
$29/mo
Local research agent, bundled Claude + Gemini with fair-use quota, your terminal, your data.
$99/mo
Everything in Agent, plus the Voltacent data feed, larger AI quota.
$199/mo
Everything above, plus the Signals API with the production triad, highest quotas, priority support.
Token quotas apply per tier — throttled, never hard-cut. Annual billing at two months off once generally available.
Infrastructure
Repository-backed research, inference interfaces, observability contracts and downstream execution are kept as explicit boundaries.
Data Lake
Point-in-time multi-timeframe inputs. No public market-data connection is claimed.
Exists today
EvidenceNot yet deployed
PlannedResearch results
Historical results remain connected to the exact model run that produced them.
1.444
run xgboost_20260903_200844
52.1%
run xgboost_20260903_200844
1.424
run lightgbm_20260908_170458
0.0475
run xgboost_reg_20260912_160548
2.155
run xgboost_reg_20260912_160548
5.853
run xgboost_20260916_103341
−2.80%
run xgboost_20260916_103341
−6.09% (2020)
run xgboost_20260916_103341
Review the methodology, ask technical diligence questions, or discuss future platform access.
Legal
Voltacent Quantitative Technologies provides quantitative technology and software infrastructure. It is not financial advisory, brokerage, fund management, or investment advisory services.
Research
Model development and out-of-sample evaluation on historical data. Results are measurements of a research protocol, not trading outcomes.
Backtesting
Simulated execution over historical data under modelled cost assumptions. Simulation cannot reproduce real fills.
Staging
A deployment status for integration work. It does not establish live trading, operational availability or performance.
Canary telemetry snapshot
A recorded staging observation from telemetry.json, dated 16 September 2026. It is historical and does not establish current or live status.