BACKTESTING DISCLAIMER — All DPVI scores, lead time figures, velocity readings, and market correlations are derived from historical backtesting and illustrative modeling, for informational purposes only. Past signal performance does not guarantee future results. All live signal outputs shown are illustrative, based on GrayStak methodology applied to publicly available information, and do not constitute trading recommendations. This material does not constitute investment advice.
GS
Escalation Signal Detection
Political events have always been unpredictable.
The escalation pattern that precedes them has not been.
5h 15m
Lead time before Capitol breach — Jan 6 backtest
Jan 6
Primary proof case
June 1, 2026
Pilot opens (tentative)
The Framework
Political escalation has structure.
The cascade is legible before market consensus catches up.
Traditional political risk operates on horizons of years. Breaking news and sentiment models operate coincident with events. Neither captures what happens in the days and weeks between, when the cascade is moving fast enough to matter to a desk but has not yet reached the news cycle. That window is where GrayStak operates.
What GrayStak detects
GrayStak does not predict events. It scores the trajectory of political-escalation cascades, in the window between when escalation begins moving and when market consensus forms. That window — measured in days to weeks — is where variant perception lives.
An esoteric signal class measured systematically. It does not fit existing political-risk research budgets or alt-data sourcing budgets. It fits the signal evaluation pods of multistrats and systematic macro shops operating at scale.
The detection mechanism
01 — Structured Forces
Institutional pressure, regulatory language, legislative dynamics. These set the conditions. IPSI tracks this layer continuously.
02 — Strategic Actors
Elites read structural conditions and coordinate publicly before they act. ESSI measures that coordination through statement volume, platform reach, language tone, and timing patterns across institutional accounts.
03 — Trigger Events
Physical manifestation of the conditions above. FOSI captures field signal in live operations. MNSI follows as media catches up.
Time horizon comparison
Traditional political risk
2yr – 10yr
GrayStak
Days – weeks
Breaking news / NLP sentiment
Coincident
The Problem
By the time consensus forms,
the position is already exposed.
Signal and coverage arrive in the wrong order. Field traces appear first. Institutional language shifts. Media fragments. By the time a national consensus has formed, the window has closed and the move is underway.
T+0h
Cascade traces appear — institutional language, elite coordination, field signals
T
GrayStak dual-trigger fires. DPVI state: Critical.
T+3h
Scattered media. No consensus. Window still open.
T+5h
Event materializes. Market begins to react.
T+6.5h
National saturation. Volatility peaks. Most desks react.
Lead time window
Alert to event materialization
5h+
The Foundational Insight
Escalation
compresses rungs
before it arrives.
Political escalation rarely appears without precedent. It moves through layers — institutional pressure, elite coordination, field activity, narrative fragmentation — and leaves traces at each layer before any single outlet has the story. The convergence of those traces is the signal.
Published — The Consensus Gap, GrayStak Substack
Field observer activity shifts
T+0h
Institutional language changes — Congress.gov, Federal Register
T+1h
Media narrative fragmentation begins
T+1.5h
DPVI sustained Critical, positive velocity — dual-trigger fires
T
Scattered media coverage begins
T+3h
Event materializes — market begins to reprice
T+5h
National media consensus — volatility peaks, window closed
T+6.5h
Structured Forces
Institutional pressure that primes the environment. IPSI.
Strategic Actors
Elites coordinate publicly before they act. ESSI detects this first.
Trigger Events
Physical manifestation. FOSI confirms. MNSI follows.
DPVI Model Architecture
Four independent signal layers,
weighted into one composite score,
recalculated every six hours.
The DPVI is not a news aggregator or a sentiment model. It is a multi-layer convergence score: four independent signal inputs, weighted and synthesized every six hours into a single machine-readable output. Each layer is sovereign. The composite fires when they converge. The cascade is structurally legible if measured systematically — each layer moves on a different clock, but the convergence pattern is consistent across event types.
FOSI0.30 + MNSI0.25 + ESSI0.25 + IPSI0.20 = DPVI
The defensible position
FOSI generates a class of data that does not exist in any database and cannot be scraped, purchased, or replicated. The other three sub-indices process public information faster and more systematically than human consensus forms. That combination is the moat.
FOSI is sourced from the GrayStak Media field network in live operations. For historical cases, FOSI is reconstructed from documented public field reporting — conflict journalism, primary footage, and corroborating sources — under a transparent methodology. Each sub-index has its own provenance; the four-layer cascade is validated end-to-end against the documented record.
FOSI
30%
Field Observation Sub-Index
Field observers extract signal from environments where conventional data sources fail or arrive late. Crowd dynamics, physical escalation, and local conditions captured before they enter any feed.
MNSI
25%
Media Narrative Sub-Index
Fragmentation velocity across media layers. Measures divergence before convergence — the pre-consensus signal.
ESSI
25%
Elite Signaling Sub-Index
Statement volume, platform reach, language tone, and coordination patterns across institutional accounts. Detects elite activity before physical events occur.
IPSI
20%
Institutional Pressure Sub-Index
Federal Register, Congress.gov filings, regulatory language. The institutional paper trail of escalation.
Data inputs: Perigon · Congress.gov · Federal Register · X/Tweet API · yfinance
The Signal
A scored signal,
delivered via API,
recalculated every six hours.
Each scoring cycle produces a state, a composite score, a velocity delta, and a timestamp. Output is machine-readable and integrates into existing desk infrastructure without new workflow.
The signal does not interpret events. It scores cascade trajectory and fires when threshold and velocity converge. Interpretation is the desk’s job. Detection is ours.
API delivery — alert routing configurable to desk preferred channels
DPVI OUTPUT — BACKTESTED EVENT
EVENT ID
DPVI-2021-0106-001
Jan 6, 2021
DPVI SCORE
68.95 / 100
Critical
STATE
CRITICAL
↑ sustained from morning
VELOCITY Δ
+6.90pts
/ 6h at trigger
TRIGGER
DUAL [T+V] FIRED
TIMESTAMP
2021-01-06 14:00 UTC
LEAD → BREACH
5h 15m █
(verified)
DPVI State Machine
Five states,
scored against threshold
and velocity.
A score at Elevated for 48 hours is noise. A score moving from 28 to 71 in a single 6-hour cycle is a signal. Both conditions must fire to alert — absolute level plus velocity, with no human in the loop.
DPVI Composite Score
—
Select a state →
Anticipatory
Event calendar window open. Pre-positioning monitor active.
Calendar
Baseline
Normal environment. No escalation signal detected.
0–29
Elevated
Convergence beginning. Alert threshold approaching.
30–59
Critical
Score ≥60. Dual-trigger required to alert.
60–100
Resolving
Peak passed. Score decelerating. Reassess exposure.
↓
Signal Architecture
Velocity
over
probability.
Markets price probability. Probability is static — it does not capture trajectory. A situation where the cascade has been accelerating for 24 hours is not the same as one where it has plateaued for two weeks. The velocity tells you which one you are in.
GrayStak does not ask how likely an event is. It asks how fast the conditions are moving, and in which direction. That is why it leads.
Jan 6 in the data
Jan 6 morning, 14:00 UTC: DPVI sustained Critical with positive velocity (+6.90pts cycle-over-cycle, four-layer composite). Both conditions fired. The dual trigger captured the difference between elevated noise and active escalation, 5h 15m before the Capitol breach.
Why two conditions, not one
[T] THRESHOLD
Level confirms it matters
DPVI crosses the Critical boundary. Structural pressure, elite coordination, and field signals have converged. Not a spike — a sustained elevation.
[V] VELOCITY
Rate confirms it is moving
Sustained positive velocity inside Critical territory. The situation is accelerating right now, not sitting at a stable plateau.
[T+V] DUAL TRIGGER FIRES
Materially elevated and actively moving
A score at 50 for 48 hours is noise. A score that moves from 57 to 71 in 6 hours is a signal. The dual trigger encodes that distinction with no human judgment in the loop.
Proof of Concept — Primary Backtesting Event
January 6, 2021.
The dip before the spike.
GrayStak alert
09:00 ET
DPVI Critical sustained, +6.90 velocity. VIX at 25.20 and drifting.
Capitol breach
14:15 ET
Event materializes. VIX at 22.63 (still suppressed).
VIX intraday peak
15:25 ET
26.77. Volatility surface fully repriced.
Lead time
5h 15m / 6h 25m
To breach event / to VIX peak
07:00 ET
DPVI re-enters Critical band
09:00 ET
▶ DPVI DUAL-TRIGGER — CRITICAL
14:15 ET
Capitol breach — event materializes
15:25 ET
VIX peak — volatility surface repriced
Lead
5h 15m to breach · 6h 25m to volatility peak
VIX 1-min intraday — Jan 6, 2021 (CBOE)
DPVI sustained Critical at 09:00 ET, with VIX at 25.20 and drifting toward its 22.14 intraday low. The volatility surface did not respond until 14:30 ET (Senate evacuation), peaking at 26.77 by 15:25 ET. The DPVI signal led the VIX response by over six hours.
Overlay independence
VIX is shown here because Jan 6 produced a particularly clean volatility signature. The DPVI signal is overlay-independent: the same trajectory can be evaluated against rates, crude, sector indices, or any market variable relevant to the desk’s book.
January 6 is the primary proof case. 13-event validation suite in progress across 2020–2026.
Validation Framework
13 events.
Designed to
break the model.
Cases span acute political violence, electoral shock, legislative brinkmanship, financial-political shock, and geopolitical escalation. Different event types, geographies, and market regimes by design. Each case is scored end-to-end across all four sub-indices: IPSI, ESSI, and MNSI from documented public record; FOSI from documented field reporting under a transparent methodology, with the GrayStak Media field network as the live-operations source.
Selection principle
Events were not chosen because the signal worked. They were chosen to stress-test the cascade architecture across the widest possible range of escalation types. January 6 was the first completed run. Four-layer scoring with documented FOSI sourcing and verified CBOE intraday VIX confirmed the architecture.
1
Completed
12
In Progress
5
Event types
Full backtesting suite — 2020–2026
Event
Type
Status
01
January 6, 2021 — Capitol Breach
2021
Acute Violence
COMPLETE
02
George Floyd / BLM Escalation
2020
Acute Violence
IN PROGRESS
03
2020 U.S. Election Night
2020
Electoral Shock
IN PROGRESS
04
2022 Midterms — Red Wave Failure
2022
Electoral Shock
IN PROGRESS
05
SVB Collapse — Regulatory Cascade
2023
Financial-Political
IN PROGRESS
06
Debt Ceiling Brinkmanship
2023
Legislative
IN PROGRESS
07
Dobbs Decision Leak
2022
Policy Escalation
IN PROGRESS
08
Ukraine Invasion Onset
2022
Geopolitical
IN PROGRESS
09
Taiwan Strait — Pelosi Visit
2022
Geopolitical
IN PROGRESS
10
Israel-Hamas Escalation Onset
2023
Geopolitical
IN PROGRESS
11
Trump Conviction / Legal Escalation
2024
Electoral Shock
IN PROGRESS
12
Iran War Onset — Operation Epic Fury
2026
Geopolitical
IN PROGRESS
13
Trump-Iran Signal Loop
2026
Geopolitical
IN PROGRESS
January 6 — Signal Sequence
The sequence
is structural,
not random.
Each layer moves at a different time. That timing maps to how political escalation actually unfolds.
ESSI — Jan 2–3
Elites coordinate. 4 days before breach.
Composite trigger — Jan 6 14:00 UTC
DPVI sustained Critical, +6.90 velocity. 5h 15m before breach.
MNSI — Jan 7–8
Media peaks. Window already closed.
VIX — Flat then spike
Market lagged the signal until the move was underway.
Two lead times: a 4-day anticipatory window as ESSI hit ceiling on Jan 2, and a 5h+ action window when the dual trigger fired. Both structural.
DPVI composite & sub-indices · Jan 2–8, 2021 · Backtested
The composite was rising before the market priced it.
Signal Metrics
What a quant
gets from
this signal.
A Sharpe ratio derived from a 13-event backtesting set would not be statistically meaningful, and we will not present one. What the pilot produces instead is access to the live signal, full backtesting documentation, and the data needed for the desk to construct its own attribution model against its own book.
The DPVI generates a timing advantage. Alpha extraction depends on position sizing, instrument selection, and conviction — all decisions that belong to the PM. We provide the raw input: lead time window, market movement during that window, hit rate across the backtesting set. The desk prices the edge.
What formal validation will produce
Sharpe contribution · Information ratio · Hit rate (statistically valid N) · Factor correlation matrix · Max drawdown of hedged vs. unhedged exposure · Attribution by event type
Hit Rate — Backtesting Set
1/1
Events with measurable market movement following Critical-state trigger. Primary proof case: January 6, 2021. Architecture performed as theoretically predicted. 13-event validation suite in progress; formal statistical validation follows full suite completion.
Factor Independence
~0
Structural uncorrelation with NLP sentiment scores at alert time. DPVI fires before media consensus — by construction, it cannot correlate with signals derived from that consensus.
Fits across primary buyer categories
→ Systematic Macro
→ Discretionary Macro
→ Fixed-Income RV
→ Commodity Pods
→ Risk Overlay
→ Multi-Strat Risk
Downside Protection Pattern
1/1
DPVI Critical state preceded VIX spike on the primary proof case by 3.5 hours. Pattern to be validated across the full suite.
Recent Case Study
Case 12 / 13 — In Active Validation
Operation
Epic Fury.
The runway was visible.
A 46-day escalation runway was visible in the data ahead of Iran War onset. Brent Crude surged 60%+ as Iran blockaded the Strait of Hormuz. Case 13 (March 23–26, 2026) produced measurable intraday moves on every ESSI event within the same 6-hour cycle.
The most complex event in the validation suite. Four-layer scoring across IPSI, ESSI, MNSI, and FOSI. Each sub-index sourced from documented public record and GrayStak Media field coverage.
Primary market exposure
Brent Crude · WTI · S&P 500 · 10Y Treasury · DXY · Gold
DPVI BACKTEST — IRAN WAR ONSET / SIGNAL LOOP
CASE REF
BKTEST-012 / 013
Feb–Mar 2026
DPVI SCORE
Critical band
Four-layer
STATE
CRITICAL
↑ from ANTICIPATORY Jan 13
VELOCITY Δ
Sustained positive
/ across event window
TRIGGER
DUAL [T+V] FIRED
Feb 28, pre-event
LEAD TIME
46 days
anticipatory runway
BRENT CRUDE
+60%+
Hormuz blockade
Case 13 — Intraday Signal Loop
March 23–26: each Trump Truth Social post produced a measurable Brent Crude move within the same 6-hour scoring cycle. De-escalation language scored correctly. State machine held CRITICAL throughout, with macro state anchored against intraday noise.
Provenance: confirmed — fully documented public record through March 2026
Competitive Position
GrayStak is not
a political risk
research product.
In a news-driven economy where political volatility moves markets at unusual frequency, the products available to institutions fall into two categories. Traditional political risk research operates on multi-year horizons and provides analyst judgment. NLP sentiment and breaking-news products operate coincident with events. The window between — days to weeks, where the cascade is moving fast enough to matter to a desk but has not yet reached the news cycle — is where GrayStak operates and is structurally unaddressed by either category.
GrayStak is an Escalation Signal platform: quantitative, state-aware, velocity-triggered, delivered via API.
Traditional Political Risk Research
Annual reports, 2–10 year horizons, structural assessments. Useful for long-term capital allocation. Not useful for a desk managing exposure this week.
NLP Sentiment / Breaking News
Coincident or lagging by construction. Measures what is already in the narrative. By definition, the window is closing when these products fire.
Institutional Geopolitical Risk Indices
BGRI, Caldara–Iacoviello GPR. Measure market attention to risks once consensus has formed enough to generate analyst coverage. Lagging by construction.
GrayStak
Days to weeks. Trajectory detection before consensus forms. Upstream of where institutional indices measure. Scored systematically.
Team
The team
building the
signal layer.
Political dynamics, quantitative signal architecture, and hedge fund market structure. The founding team has operated across all three.
Beachhead, then adjacent markets
Phase 1 (now): Hedge fund trading desks — systematic and discretionary macro, fixed-income relative-value, commodity pods, and risk-overlay functions at long-only managers.
Phase 2: Enterprise risk management, board-level political exposure, cyber risk.
The product stays the same. The pitch evolves.
CS
Christopher Sweat
Co-Founder & CEO
Political dynamics, escalation pattern analysis, commercial strategy. Built GrayStak from the ground up — signal theory, architecture, and go-to-market. Founder of GrayStak Media, the public coverage layer producing the field observation signal that feeds FOSI.
RL
Ryan Lazarus
Co-Founder
Media and technology startup background. Brings product instinct and distribution thinking to GrayStak — how signals get packaged, positioned, and adopted at scale.
ES
Eddie Satterly
Technical Advisor — CTO/COO
Infrastructure architecture, ML pipeline design, and acquisition experience. Shaped GrayStak technical stack and UX philosophy from first principles.
KC
Kevin Chen
CIO Advisor & First Validator
Institutional CIO. First sophisticated buyer to evaluate GrayStak methodology against real desk workflow. Ongoing advisory and validation.
Signal Pricing
Priced as a signal,
not as alt data.
Alt data is priced against the data sourcing budget, typically in the $25–30K range. A scored signal that produces measurable timing advantage is priced against alpha P&L and sits in a different commercial conversation. GrayStak is priced for the second category.
Stage 1 — Gated Pilot Entry
$5K
paid pilot reservation
Live DPVI access through the next Critical-state event or 90 days, whichever comes first. Anticipatory state observation, sub-index analysis, and weekly debriefs throughout. The $5K is credited toward annual access on conversion.
→
Stage 2 — Annual Signal Access
$50K
per trading desk, per year
Full DPVI access. Per-desk pricing, not enterprise license. Pilot $5K applied to first-year cost.
Why signal pricing is different
Alt data sourcing budget
Data sourcing team owns this
$25–30K
Signal budget
PM and CIO own this
$50K+
Upside and downside
The signal serves both upside positioning and downside protection. Political shocks produce correlated unwinding across pods that strategy labels said were uncorrelated. The cost of being blindsided once — the deleveraging that compresses across factor exposures — exceeds the annual cost of the signal by a wide margin.
The Ask
One desk.
One pilot.
Event-gated conversion.
The pilot is structured as a paid evaluation. $5K reserves access to the live signal and is credited toward the $50K annual fee on conversion. The pilot runs through the next Critical-state event or 90 days, whichever comes first. Anticipatory state observation, sub-index analysis, and weekly debriefs are delivered throughout the pilot window regardless of Critical-state activity. A Critical-state event during the pilot accelerates the conversion conversation.
$5K
Pilot, credited toward annual
$50K
Annual, per desk
June 1
Pilot opens (tentative)
Pilot Structure
→
One trading desk, full DPVI access during pilot
→
Live score and state machine alerts, every 6-hour cycle
→
$5K credited toward annual fee on conversion
→
January 6 backtesting documented; additional cases in active validation
→
Weekly signal debrief — what fired, why, what the market did
→
API delivery, alert routing configurable to desk preferred channels
Pilot window opens June 1, 2026. Founder conversations available now to align scope and success criteria.
Escalation Signal
01 / 17