Built for institutional commodity desks

Real-world activity,
audited into tradable
commodity signals.

QNT42 captures industrial intensity, geopolitical friction, biosphere health and climate stress from physical sources — then validates every layer against forward returns before it reaches your desk. No black boxes, no in-sample heroics.

Validated composites
6
Top audit IC
+0.45
Re-audit cadence
90d
ZC=F · CORN  /  EXAMPLE PIPELINE
NTL VIIRS 500m
Stage 01
Raw NTL radiance
signal/noise2.4 dB
Stage 02
Climatology-corrected
residual MAD0.38
Stage 03
Z-score & audit overlay
z · t-1−1.32σ

Coverage · 5 commodities · 23 industrial hubs · 10 conflict zones

Industrial Climate Geopolitical

Independent layers
of validated alpha.

QNT42 triangulates commodity activity across independent data layers — each evaluated separately before aggregation, each with its own audit profile.

LAYER 01
Industrial Activity
Nighttime light intensity at high resolution across global industrial hubs — refineries, ports, manufacturing zones. Robust median z-score with seasonal climatology detects structural shifts in operational intensity ahead of official data.
NTL · VIIRS 23 hubs Top IC +0.45
LAYER 02
Geopolitical Risk
IC-weighted composite of nighttime light activity across active conflict zones — Ukraine, Middle East, Taiwan Strait. Documented regime breaks since 2022 anchor the baseline. Maps directly to supply disruption risk for affected commodities.
10 conflict zones regime-break aware Top IC +0.30
LAYER 03
Climate & Biosphere
ENSO ONI cycle, US drought conditions, MODIS NDVI/EVI vegetation health — translated into directional bias for soft commodities. Each factor independently audited at 30/60/90 day horizons across corn, soybeans and wheat.
ENSO · USDM · MODIS 75+ factors Top IC +0.30

Every commodity, every layer
at a glance.

No aggregate scores. No black boxes. Every commodity delivers a layer table showing the current z-score, audit tier and information coefficient for each indicator. PMs read convergence directly.

ZC=F · CBOT
Corn
Three validated layers · convergent bearish
multi-layer view
updated daily
audit master_v1.0
Layer
Z-score
Tier
Audit IC
Status
IndustrialNTL · ag-processing hubs
−1.32σ
VAL
+0.448 60d horizon
Bearish
ClimateENSO · US drought
−0.03σ
VAL
+0.190 90d horizon
Neutral
BiosphereMODIS · Mato Grosso · Pampa
+0.06σ
VAL
+0.301 90d horizon
Neutral

How a PM reads this

Industrial bearish, validated
NTL activity at corn-processing hubs running −1.32σ below 5y climatology. 60d IC +0.45 with DSR 0.99. Survives macro residualization. Tier-1.
Climate near neutral, validated
ENSO ONI and US drought composite at −0.03σ. 90d IC +0.19. Layer is statistically validated but currently shows no directional pressure.
Biosphere near neutral, validated
MODIS vegetation indices across Mato Grosso and Argentine Pampa at +0.06σ. 90d IC +0.30. Currently constructive but not yet directional.
One validated layer signals bearish; two remain neutral. The platform never produces a composite score. The convergence call is the PM's. Layer table delivered via dashboard or REST.

Why a quant PM
asks different questions.

Most alternative-data vendors optimize for narrative. We optimize for survival under a research-team's interrogation. The difference is what shows up in a due-diligence checklist.

Capability
Typical alt-data vendor
QNT42
Walk-forward out-of-sample validationnon-overlapping window stability
In-sample backtest only
4 non-overlapping windows
Deflated Sharpe correctionmulti-trial Sharpe deflation
Raw Sharpe / IC reported
Bailey & López de Prado · DSR published
Macro residualizationedge vs. risk-premium decomposition
Not performed
VIX + USD broad · 21d OLS
Sign consistency & inversionregime-dependent reversal screen
Global IC reported only
Per-window check · transparent inversion
Tier classification publishedvalidated · indicative · research
Hidden — all signals presented equal
Public 3-tier · per indicator
Layer transparencyper-driver decomposition
Single conviction score
No composite · PM reads convergence
Re-audit cadencerefresh of validation surface
Ad-hoc / on request
Quarterly · event-triggered resets

Built to survive
institutional due diligence.

Our indicators are not just predictive in-sample. Every layer is validated through a multi-step framework designed to expose overfitting, regime dependence, sign instability and macro redundancy before any indicator reaches production.

01  Walk-forward validation
Out-of-sample stability
Every indicator is evaluated across four non-overlapping out-of-sample windows. Stability across windows — not in-sample fit — is the primary acceptance criterion. Indicators that work in one window but not others are flagged as research-phase, not validated.
4 non-overlapping windows
02  Deflated Sharpe Ratio
Multi-trial correction
PnL Sharpe ratios are deflated to account for the number of configurations tested, the skewness and kurtosis of returns, and sample size. The DSR probability quantifies the likelihood that observed performance reflects genuine edge rather than multi-trial noise.
Bailey & López de Prado (2014)
03  Macro-controlled IC
Edge vs. risk premium
Both indicator and forward returns are residualized against observable macro factors — VIX changes, USD broad index — before correlation is measured. An indicator whose raw IC collapses after residualization is redundant with macro factors and should not be sold as standalone alpha.
OLS residualization · 21d windows
04  Tier classification
Honest factor grading
Every indicator carries an explicit tier. Validated indicators meet IC significance, walk-forward stability, sign consistency and DSR thresholds simultaneously. Indicative indicators show evidence but require longer history. Research-phase indicators are excluded from production aggregation.
3-tier · quarterly cadence
05  Robust statistics
Outlier-resistant baselines
Z-scores are computed using robust median + MAD (Median Absolute Deviation) against five-year seasonal climatology. Documented regime breaks reset baselines after major structural events — invasions, embargoes, infrastructure attacks.
Median + MAD · 5y climatology
06  Sign consistency
Directional integrity
When the audit detects IC<0, the production signal is inverted before serving — preserving the convention signal>0 = bullish across all models. Sign inversion is flagged transparently in every audit report. No hidden direction-flipping.
Per-window check · transparent inversion

Bring your research team
signals they can defend.

QNT42 is available to a limited number of institutional partners. Request access to receive the full audit report and validation methodology under NDA.

Limited availability · Institutional only · NDA on request

Talk to research, not sales.

Every access conversation is routed directly to our research team. We open with the audit report, the methodology paper and a sample layer table for the commodity that matters to you.

@
hi@qnt42.com
Direct to research desk
§
Methodology paper
Released under NDA
✓ Thank you — we will be in touch shortly.