CycleNavigator: Visualizing Economic and Political Cycles Through AI

Leveraging Long Cycles for Strategic Intelligence

The CycleNavigator project represents a sophisticated fusion of economic history, data visualization, and generative AI. It delivers insights into four major macroeconomic and geopolitical cycles — Business (9 years), Kondratiev (50 years), Finance (80 years), and Hegemony (250 years) — through interactive Plotly visualizations and GPT-enhanced analysis.

This article explores the nature of these cycles, the code architecture that brings them to life, and how the integration of historical patterns with modern AI creates a powerful intelligence tool for researchers, analysts, and decision-makers.

The Four Long Cycles: Origins and Characteristics

Business Cycle (≈9 years) — The "Heartbeat" of Investment and Inventory

The smallest yet most empirically validated cycle captures the equipment investment and inventory oscillations first documented by French statistician Clément Juglar in 1862. Recent spectral analysis of OECD and IMF time series continues to confirm this 7-11 year pattern in modern economies.

Kondratiev Wave (≈50 years) — The Long Technological Wave

Named after Russian economist Nikolai Kondratiev, this 45-60 year cycle is connected to technological innovation clusters. The code defines this relationship explicitly:

CYCLES = { "K-Wave": 50, "Business": 9, "Finance": 80, "Hegemony": 250 }

Each technological revolution — steam engine, railways, electricity, ICT, AI — emerges in approximately half-century intervals, and the code visually superimposes these waves to reveal their overlapping patterns.

Long Credit Cycle (≈80 years) — Debt Cycles and Major Financial Crises

The Bank for International Settlements' long-term database suggests that the total credit-to-GDP ratio follows a pattern of excess→collapse→deleveraging at approximately 60-100 year intervals. These credit cycles precipitate systemic crises such as the 1929 Great Depression and the 2008 Global Financial Crisis.

Hegemony Cycle (≈250 years) — The Rhythm of Global Order Transitions

Scholars like Kindleberger, Modelski, Wallerstein, and Ray Dalio observe that maritime hegemonic transitions (Netherlands→Britain→United States) occur at intervals of approximately 200-300 years. While this cycle has the weakest quantitative rigor among the four, it serves as a widely used framework for explaining the "quasi-periodic hierarchical restructuring" of international order.

The Logical Architecture of the Code

Wave Visualization Engine

def half_sine(xs, period, amp):
    phase = np.mod(xs - CENTER, period)
    y = amp * np.sin(np.pi * phase / period)
    y[y < 0] = 0
    return y

This function generates half-sine waves for each cycle to visualize periodicity. The 30-step gradient represents the intensity of each cycle, offering an intuitive representation of time's slow progression.

Historical Similarity Mapping via SBERT Embeddings

_embed_model = SentenceTransformer("sentence-transformers/all-MiniLM-L6-v2")
_all_sentences = [(yr, ev["event_en"]) for yr, evs in EVENTS.items() for ev in evs]
_embeddings = _embed_model.encode([s for _, s in _all_sentences], convert_to_tensor=True)

Through sentence embeddings, the system automatically connects historically similar events such as "Panic of 1873 ↔ Great Depression of 1929 ↔ Global Financial Crisis of 2008." Users can immediately verify these similarities via hover tooltips on the chart.

Real-time News Integration via Brave API

def fetch_cycle_news():
    markers = []
    for cyc, kw in NEWS_KEYWORDS.items():
        res = brave_search(kw, 1, freshness_days=2)
        if res:
            markers.append({
                "cycle": cyc,
                "title": res[0]["title"],
                "year": datetime.utcnow().year,
                "url": res[0]["url"]
            })
    return markers

This function retrieves news headlines from the last 48 hours for each cycle keyword and displays them as star icons (★) on the chart. This visually superimposes "fast events on slow waves," connecting long-term cycles with current news.

GPT-Powered Analysis Engine

BASE_PROMPT = (
    "You are **CycleNavigator AI**, an expert in economic history, geopolitics, and long-wave analysis "
    "(9-year Business, 50-year K-Wave, 80-year Finance, and 250-year Hegemony cycles). "
    "Answer **in Korean**, maintaining both scholarly accuracy and practical clarity. "
    "✦ Structure every reply as follows: ① Summarize the core question → ② Specify how it relates to the four cycles "
    "→ ③ Explain the historical & data evidence → ④ Offer implications / outlook. "
    "Use numbered points, bullet lists, and concise paragraphs to keep the logic clear."
)

The system prompt enforces a four-stage structure: "①Question summary→②Cycle relevance→③Historical/data evidence→④Implications/forecast," generating responses with clearly delineated facts, context, and outlook. This structured prompting enhances the consistency and reliability of AI responses.

Practical Utility from an AI Perspective

Minimizing Cognitive Load

The combination of Plotly's zoom and hover capabilities with GPT summaries significantly streamlines the "data scan→interpretation→report draft" process. Experimental research indicates that such interactive charts substantially improve decision-making accuracy. Users can quickly comprehend and analyze complex economic patterns.

def build_chart(start: int, end: int, lang: str = "KO"):
    # ... chart construction logic ...
    fig.add_trace(go.Scatter(
        x=[yr], y=[yv], mode="markers",
        marker=dict(color="white", size=6),
        customdata=[[cyc, txt, sim]],
        hovertemplate=(
            "Year %{x} • %{customdata[0]}<br>"
            "%{customdata[1]}<br>"
            "Similar: %{customdata[2]}<extra></extra>"
        ),
        showlegend=False))

Real-time Context Provision

The headlines (★) retrieved through the Brave API instantly connect ongoing events with long-cycle contexts. For instance, a hegemony researcher can immediately evaluate US-China technology regulation news from the perspective of "K-Wave peak + hegemonic transition phase."

erifiable Narratives

By open-sourcing cycle_events.json along with links to the Maddison Project GDP, BIS credit, and war databases, users can modify and extend the data themselves to validate hypotheses. This is a key element in enhancing transparency and credibility.

EVENTS_PATH = pathlib.Path(__file__).with_name("cycle_events.json")
with open(EVENTS_PATH, encoding="utf-8") as f:
    RAW_EVENTS = json.load(f)
EVENTS = {int(item["year"]): item["events"] for item in RAW_EVENTS}

Extensibility as an Educational, Research, and Strategic Plugin

The Gradio UI is deployed as a single file, making it immediately executable on Colab or local GPUs. This facilitates easy extension into economic history lectures, asset management risk dashboards, or policy institute scenario tools.

def create_app():
    with gr.Blocks(
        theme=gr.themes.Soft(),
        css="""
           #discord-badge{position:fixed; bottom:10px; left:50%;
                          transform:translateX(-50%);}
        """
    ) as demo:
        # ... UI components ...

Conclusion — Transforming 'Deep Currents' into Contemporary Decision Tools

CycleNavigator AI repackages useful hypothetical waves through data, visualization, and generative AI rather than presenting them as "proven laws." This enables users to comprehend complex macro, financial, and geopolitical phenomena at a glance, while verifying evidence and engaging in discussion in real-time.

Particularly noteworthy is how AI extends human knowledge and intuition. Users gain a tool that bridges the gap between historical pattern recognition and current event analysis. Like Mark Twain's insight that history doesn't repeat but "rhymes," CycleNavigator captures the "rhymes" of historical patterns, enabling deeper understanding of present and future.

When slow long cycles (9-50-80-250 years) and fast headlines (Brave API) meet on a single canvas, users experience a novel intelligence paradigm where they can reconstruct the past, interpret the present, and actively design future scenarios. This transcends mere data visualization, maximizing the synergy between AI and human intelligence to create a new decision-making paradigm.

Repo(Open-Source) : https://huggingface.co/spaces/openfree/Cycle-Navigator

• license: apache 2.0
• community: https://discord.gg/openfreeai