File size: 4,635 Bytes
d4fd27a
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
81e234f
d4fd27a
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112

---
license: mit
task_categories:
- time-series-forecasting
- tabular-classification
- other
pretty_name: Financial Structural Breaks & Regime Detection Benchmark
tags:
- finance
- econophysics
- algorithmic-information-theory
- structural-breaks
- time-series
- anomaly-detection
size_categories:
- 1M<n<10M
---

# Financial Structural Breaks & Regime Detection Benchmark

**Maintainer:** [Algoplexity](https://github.com/algoplexity)  
**Primary Repositories:**
1.  **The Coherence Meter:** [GitHub Repo](https://github.com/algoplexity/Coherence-Meter) (Horizon 0)
2.  **The Computational Phase Transition:** [GitHub Repo](https://github.com/algoplexity/computational-phase-transitions) (Horizon 1)

## 1. Overview
This repository serves as the **immutable data artifact** for the Algoplexity research program into **Algorithmic Information Dynamics (AID)** in financial markets. 

It contains a large-scale collection of non-stationary, continuous financial time series, specifically curated to benchmark methods for **Structural Break Detection** and **Market Regime Diagnosis**. This data underpins the validation of two distinct methodologies:
*   **The Coherence Meter:** A statistical, falsification-driven framework comparing "Stethoscope" (univariate) vs. "Microscope" (multivariate) approaches.
*   **The AIT Physicist:** A transformer-based diagnostic tool that maps market dynamics to **Wolfram Complexity Classes** (e.g., Rule 54 vs. Rule 60) to detect "Computational Phase Transitions."

## 2. Dataset Utility
This dataset allows researchers to reproduce key findings from the associated papers, including:
*   The **"Cost of Complexity"** curve (MDL analysis).
*   The **-27.07% Early Warning** signal in algorithmic entropy.
*   The distinct topological signatures of **Systemic** vs. **Exogenous** crashes.

## 3. Dataset Structure
The data is stored in highly compressed **Parquet** format, optimized for scientific computing and cloud-based ingestion.

### Files
*   **`X_train.parquet`**: The primary feature set containing thousands of continuous financial time series.
*   **`y_train.parquet`**: The ground-truth labels indicating the precise timestamp of structural breaks.
*   **`X_test.parquet` / 'y_test.parquet'**: Out-of-sample series (derived from the Falcon forecasting challenge) used for generalization testing.

### Schema
**Features (`X_train.parquet`)**:
*   `id` (string): Unique identifier for the time series.
*   `period` (int): Sequential time step.
*   `value` (float): The continuous signal (price/return).

**Labels (`y_train.parquet`)**:
*   `id` (string): Unique identifier.
*   `structural_breakpoint` (int): The time step where the regime shift formally occurs.
*   `label` (int): Class identifier (0 = No Break, 1 = Break).

## 4. Provenance
*   **Source:** Derived from the **CrunchDAO** research competitions (Structural Break & Falcon).
*   **Preprocessing:** Data has been anonymized, standardized, and formatted for both statistical analysis (rolling variance) and algorithmic encoding (quantile binning).

## 5. Universal Loading (Python)
This dataset is designed to be ingested directly from the cloud, removing dependencies on local storage or Google Drive.

```python
from huggingface_hub import hf_hub_download
import pandas as pd

def load_benchmark_data(filename):
    """
    Fetches data from the Algoplexity Benchmark Repository.
    Uses local caching for offline capability.
    """
    repo_id = "algoplexity/computational-phase-transitions-data"
    print(f"--- Fetching {filename} from Scientific Repository ---")
    
    local_path = hf_hub_download(
        repo_id=repo_id, 
        filename=filename, 
        repo_type="dataset"
    )
    return pd.read_parquet(local_path)

# Usage
df_features = load_benchmark_data("X_train.parquet")
df_labels = load_benchmark_data("y_train.parquet")
```

## 6. Citation
If you use this data in your research, please cite the associated Algoplexity repositories:

```bibtex
@misc{ait_physicist_2025,
  author = {Mak, Yeu Wen},
  title = {The Computational Phase Transition: Quantifying the Algorithmic Information Dynamics of Financial Crises},
  year = {2025},
  publisher = {GitHub},
  journal = {GitHub repository},
  howpublished = {\url{https://github.com/algoplexity/computational-phase-transitions}}
}

@misc{coherence_meter_2025,
  author = {Mak, Yeu Wen},
  title = {The Coherence Meter: A Hybrid AIT-MDL Framework for Early-Warning Structural Break Detection},
  year = {2025},
  publisher = {GitHub},
  journal = {GitHub repository},
  howpublished = {\url{https://github.com/algoplexity/Coherence-Meter}}
}
```