Global Power Shift Index
Geopolitical Risk & Strategic Dynamics Monitor
Measures geopolitical integration vs. isolation. High RII = Isolated/Pariah (e.g., SY, YE). Low RII = Deeply integrated via bilateral defense pacts, multilateral blocs, and economic hubs (e.g., IL, KR, DE, CY).
The Global Power Shift Index (GPSI) is a composite indicator of short-term geopolitical relevance and relative strategic position. It combines five structural dimensions using the current weights loaded dynamically from the DIM_WEIGHTS sheet and applies them to normalized 0–100 dimension scores.
Important caveat: GPSI is not a direct measurement of national power. Baseline scores are structured estimates grounded in public datasets and expert-coded assumptions. Daily movement comes from news-event adjustments, sentiment-gated dimension impacts, persistence/decay rules, and calibrated momentum. The result should be read as a disciplined signal of geopolitical momentum, not as a precise real-world capability measure.
× Economyscore
+ × Militaryscore
+ × Energyscore
+ × Diplomacyscore
+ × Technologyscore
// Each sub-score ∈ [0, 100]
// 1-day movement: dimension-weighted raw change, then calibrated.
// 7/14/30-day movement: raw daily GPSI changes are summed first, then scaled and period-capped.
// Positive scale: ; negative scale: .
// 1-day calibrated movements below pts are zeroed.
// Multi-day cumulative momentum bypasses the daily noise floor so small repeated moves can accumulate.
// Daily calibrated movement is capped at ± pts.
// Multi-day cap = maxDailyDelta × √days.
// Scores are pulled toward the 30-day baseline at /day.
Raw event pressure is processed through a calibrated momentum pipeline loaded from the backend:
1) Sentiment-gated boost: relevant dimension signals move only when scored sentiment is strong enough to pass the confidence gate.
2) Directional scaling: positive movement is scaled by ; negative movement is scaled by .
3) Noise filter: 1-day calibrated movements below pts are zeroed.
4) Hard cap: daily calibrated movement is clamped to ± pts.
5) Mean reversion: scores pull toward the 30-day baseline at /day.
6) Dimensional inertia: scored news days are smoothed using % yesterday state and % today signal.
7) Quiet-day persistence: low/no-news days decay using persistence toward yesterday and pull toward baseline.
Black-swan detection is used as a short-term shock-responsiveness rule, not as a direct measure of real-world power growth. When an event cluster is classified as exceptional, the backend allows the affected dimensions to respond more strongly and reduces normal inertia for that day. This registers strategic discontinuity without treating the event as a proportional increase in material national capability.
The black-swan layer should therefore be interpreted as strategic shock salience: it highlights abrupt discontinuities such as major military escalation, severe coercive shocks, sudden diplomatic breakthroughs, or multi-domain crises. The GPSI score still remains bounded by the active calibration, daily cap, inertia, and sentiment-confidence gate.
Below are the actual 1-day news signals and their mathematical calibration pipeline (scaling, noise filtering, and hard capping) applied to the top active countries.
// Effective Weights Derivation:
// Momentum Weight = Raw (—) × Amplifier (—)
// Signals Weight = Raw (—) × Amplifier (—)
// Momentum is clamped to [−—, +—]
// Signals = Σ(News Events) ÷ —, capped at —
The Base score is already expressed on the 0–100 GPSI scale, while Momentum and Signals are smaller fractional inputs. The current momentum amplifier is , and the current signal amplifier is . Both are loaded dynamically from the backend so that coalition scoring can be recalibrated from the spreadsheet without editing the widget HTML.
Strategic event gatekeeping: Raw headlines are filtered, deduplicated, country-matched, and classified before they influence scores. When available, Gemini is used to classify whether a headline is a material geopolitical event and whether the country is acting, targeted, benefiting, or only being mentioned. If the model call fails or is rate-limited, the backend falls back to a deterministic keyword/lexicon layer so the pipeline can continue with reduced semantic precision.
Shadow layer: Strategically material events produce three diagnostic values in GPI_History: Reality Score, Signaling Score, and Narrative Divergence. Reality Score reflects event substance; Signaling Score reflects media-volume and narrative intensity; Divergence = Signaling Score − Reality Score. These fields help distinguish structural shifts from media over-amplification or underreporting.
Diagnostic transparency: The backend also writes Shadow Status, Shadow Reason, and AI Source diagnostics. The widget reads these values and compares narrative divergence against thresholds loaded dynamically from the backend, rather than using fixed front-end thresholds.
Measures geopolitical integration versus isolation on a 0–100 scale (higher = more isolated). The baseline Connectivity Score is calculated as:
(—% × Alliance Density) +
(—% × Diplomatic Reach) +
(—% × Econ+Tech Capacity) +
(—% × Hard Power)
// RII = 100 - Connectivity
// High-value networks (e.g., NATO, G7) grant −— pts bonus.
// Standard alliances grant −— pts bonus.
// Diplomatic vacuums (Diplomacy < 25) incur +— pts penalty.
// Pariah status (0 alliances, low econ/dipl) incurs +— pts penalty.
The system continuously monitors for systemic tipping points using two primary triggers:
• Volatility Breach: Flags a SURGE or PLUNGE when the calibrated daily delta exceeds — pts.
• Boundary Check: Flags an ATH (All-Time High) alert when a country's structural score exceeds — while maintaining positive momentum.
Detects significant shifts in the global hierarchy through two distinct mechanisms:
• Overtaking Moments: Identifies rank inversions between the current date and the period start date. To prevent clutter, the algorithm filters for unique countries (a country can only appear once as an overtaker or overtaken) and prioritizes events by the magnitude of the rank change.
• Rank Momentum: In the broader rank movement chart, momentum is concurrently calculated as (calibrated Δ) / period days to distinguish sharp, high-impact moves from gradual drift.
The dashboard employs a two-tier narrative architecture to deliver geopolitical analysis, prioritizing generative AI with a deterministic fallback:
1) Primary: Generative AI
A daily backend process runs a Large Language Model (Gemini) that ingests calibrated momentum, media sentiment, and raw news headlines (from diverse news sources) to produce a balanced dense, structured strategic analysis report. It explicitly identifies for the past week (7-Day change) the top gainers and losers, correlating their score shifts with real-world events, strategic positioning, and key watchpoints.
2) Fallback: Deterministic Game-Theoretic Engine
If the AI narrative is unavailable, the widget instantly synthesizes a deterministic synopsis. It evaluates the systemic state based on G2 divergence, regional downside volatility, and asymmetric actor behavior, categorizing the global environment into predefined states such as:
Bipolar Fragmentation, Multipolar Diffusion, Regional Crisis Spillover, Containment Consolidation, Asymmetric Whiplash, or Baseline Adaptation. Note: The signal significance is evaluated against dynamic calibration thresholds. The tactical shifts must exceed the noise floor (—), while strategic shifts approach the maximum daily cap (—).
.png)
