coinsignal 1.4.9__cp312-cp312-manylinux2014_x86_64.whl

coinsignal/__init__.py

@@ -0,0 +1,12 @@
+from . import _ext
+from . import data
+from . import feature
+from . import label
+from . import model
+from . import parallel
+from . import sampler
+from . import tools
+
+__version__ = '1.4.9'
+
+__all__ = ['data', 'feature', 'label', 'model', 'parallel', 'sampler', 'tools']

coinsignal/data.py

@@ -0,0 +1,32 @@
+# --------------------------------------------------------------------------------
+# Copyright (c) 2025 Zehao Yang
+#
+# Author: Zehao Yang
+#
+# This module implements data reading and transformation functions:
+# • Reading raw market data from CSV files
+# • Handling static and dynamic symbol selection
+# • Transforming symbol-based data maps into feature maps
+# --------------------------------------------------------------------------------
+
+
+import _ext.data as _data
+
+
+DEFAULT_DATA_PARAMS_DICT = {
+    'static_symbols': [],
+    'dynamic_symbols': {
+        'rule': '',
+        'count': 0
+    },
+    'sources': []
+}
+
+
+def read_data_to_features_map(data_dir, date, look_back_days, look_ahead_days, data_params_dict):
+    features_map, read_data_errors = _data.read_data_to_features_map(data_dir, date, look_back_days, look_ahead_days, data_params_dict)
+    return features_map, read_data_errors
+
+def transform_features_map_to_full_features_df(features_map):
+    full_features_df = _data.transform_features_map_to_full_features_df(features_map)
+    return full_features_df

coinsignal/feature.py

@@ -0,0 +1,39 @@
+# --------------------------------------------------------------------------------
+# Copyright (c) 2025 Zehao Yang
+#
+# Author: Zehao Yang
+#
+# This module implements feature engineering functions:
+# • Adding various features based on trade and reference exchange-market pairs
+# --------------------------------------------------------------------------------
+
+
+import _ext.feature as _feature
+
+
+DEFAULT_FEATURE_PARAMS_DICT = {
+    'features':
+    {
+        'time': [],
+        'macd': [],
+        'rsi': [],
+        'price_move_range': [],
+        'price_trend': [],
+        'price_volume_corr': [],
+        'ohlcv': [],
+        'volatility': [],
+        'high_low_time': [],
+        'volume': [],
+        'basis': [],
+        'fixedstart': [],
+        'funding_rate': [],
+        'funding_time': [],
+        'index': []
+    },
+    'is_validating': False
+}
+
+
+def add_features(features_map, feature_params_dict):
+    features_map = _feature.add_features(features_map, feature_params_dict)
+    return features_map

coinsignal/label.py

@@ -0,0 +1,38 @@
+# --------------------------------------------------------------------------------
+# Copyright (c) 2025 Zehao Yang
+#
+# Author: Zehao Yang
+#
+# This module implements return (label) calculation functions:
+# • Calculating return values for multiple horizons and decay multipliers
+# • Handling time, duration, and range return types
+# • Supporting funding rate adjustment and discrete return calculation
+# --------------------------------------------------------------------------------
+
+
+import _ext.label as _label
+
+
+DEFAULT_RETURN_PARAMS_DICT = {
+    'pred_horizons': [600],
+    'decay_multipliers': [0.5, 0.8, 1, 1.5, 2, 3, 5, 10],
+    'return_type': 'time',
+    'range_direction': 'forward',
+    'range_horizon_limits': [0, 1000],
+    'range_merge_step': 1,
+    'n_multi': 1,
+    'multi_limits': [1, 1],
+    'half_life': 0,
+    'duration_vlimits': None,
+    'duration_qlimits': [0, 1],
+    'duration_handling': False,
+    'is_funding_adjusting': False,
+    'is_horizon_adjusting': False,
+    'is_discrete': False,
+    'y_multiplier': 10000
+}
+
+
+def calculate_return(features_map, return_params_dict):
+    features_map = _label.calculate_return(features_map, return_params_dict)
+    return features_map

coinsignal/model.py

@@ -0,0 +1,76 @@
+# --------------------------------------------------------------------------------
+# Copyright (c) 2025 Zehao Yang
+#
+# Author: Zehao Yang
+#
+# This module implements the main model training and evaluation pipeline:
+# • Handling data splitting, scaling, auxiliary model processing, cross-validation,
+# • Training and evaluating models with various metrics
+# --------------------------------------------------------------------------------
+
+
+import _ext.model as _model
+
+
+DEFAULT_MODEL_PARAMS_DICT = {
+    'group_count': 5,
+    'train_ratio': 0.8,
+    'valid_sub_ratio': 0.25,
+    'test_isolate_time': 0,
+    'is_random_start': False,
+    'weight_limit': 1,
+    'is_sign_balanced': False,
+    'is_cv': True,
+    'cv_split_count': 4,
+    'cv_eval_train_metric': True,
+    'cv_selection_metric': 'mse',
+    'num_boost_round': 1000,
+    'log_evaluation_period': 50,
+    'evaluate_rounds': [400, 700, 1000],
+    'X_scaler_sub_params': {
+        'name': '',
+        'vlimits': None,
+        'qlimits': [0, 1],
+        'inner_scale': 1,
+        'outer_scale': 1
+    },
+    'y_scaler_sub_params': {
+        'name': '',
+        'vlimits': None,
+        'qlimits': [0, 1],
+        'inner_scale': 1,
+        'outer_scale': 1
+    },
+    'auxiliary_model_sub_params': {
+        'name': '',
+        'model_dir': '',
+        'vlimits': None,
+        'qlimits': [0, 1],
+        'sampling_modes': ['train', 'evaluate']
+    },
+    'loss_sub_params': {
+        'name': '',
+        'dof': 10,
+        'scale': 1,
+        'fee': 0.0001
+    },
+    'training_sub_params': {
+        'boosting_type': 'gbdt',
+        'num_leaves': 25,
+        'max_depth': 5,
+        'learning_rate': 0.005,
+        'reg_alpha': 1,
+        'reg_lambda': 1,
+        'min_gain_to_split': 0.1,
+        'extra_trees': True,
+        'objective': 'regression',
+        'metric': '',
+        'num_class': 1,
+        'verbose': -1
+    }
+}
+
+
+def run_model(full_features_df, model_params_dict):
+    model_evaluations = _model.run_model(full_features_df, model_params_dict)
+    return model_evaluations

coinsignal/parallel.py

@@ -0,0 +1,21 @@
+# --------------------------------------------------------------------------------
+# Copyright (c) 2025 Zehao Yang
+#
+# Author: Zehao Yang
+#
+# This module implements parallel processing for feature preparation:
+# • Parallelizing data reading, feature and label calculations, sampling
+# • Progress tracking for parallel processing across multiple dates
+# --------------------------------------------------------------------------------
+
+
+import _ext.parallel as _parallel
+
+
+def prepare_sub_full_features_df(data_dir, date, look_back_days, look_ahead_days, constants_dict, data_params_dict, feature_params_dict, return_params_dict, sampler_params_dict, N, n_completed, progress_update_event):
+    df, mft_df, mft_feature_plan, errors = _parallel.prepare_sub_full_features_df(data_dir, date, look_back_days, look_ahead_days, constants_dict, data_params_dict, feature_params_dict, return_params_dict, sampler_params_dict, N, n_completed, progress_update_event)
+    return df, mft_df, mft_feature_plan, errors
+
+def prepare_full_features_df(data_dir, start_date, end_date, look_back_days, look_ahead_days, constants_dict, data_params_dict, feature_params_dict, return_params_dict, sampler_params_dict):
+    full_features_df, mft_full_features_df, mft_feature_plan, read_data_errors = _parallel.prepare_full_features_df(prepare_sub_full_features_df, data_dir, start_date, end_date, look_back_days, look_ahead_days, constants_dict, data_params_dict, feature_params_dict, return_params_dict, sampler_params_dict)
+    return full_features_df, mft_full_features_df, mft_feature_plan, read_data_errors

coinsignal/sampler.py

@@ -0,0 +1,26 @@
+# --------------------------------------------------------------------------------
+# Copyright (c) 2025 Zehao Yang
+#
+# Author: Zehao Yang
+#
+# This module implements data sampling functions:
+# • Sampling data on time using time intervals
+# • Sampling data on move using smoothed price change
+# • Sampling data on return using weighted random sampling
+# --------------------------------------------------------------------------------
+
+
+import _ext.sampler as _sampler
+
+
+DEFAULT_SAMPLER_PARAMS_DICT = {
+    'sampling_time': 0,
+    'is_time_random': False,
+    'sampling_bp': 0,
+    'rolling_step': 0
+}
+
+
+def sample_data(features_map, sampler_params_dict):
+    features_map = _sampler.sample_data(features_map, sampler_params_dict)
+    return features_map

coinsignal/tools.py

@@ -0,0 +1,118 @@
+# --------------------------------------------------------------------------------
+# Copyright (c) 2025 Zehao Yang
+#
+# Author: Zehao Yang
+#
+# This module provides utility functions and helper classes:
+# • I/O utilities for logging and progress tracking
+# • Configuration loading and parameter updating
+# • Rolling statistics calculations and various data processing helpers
+# • Plotting functions for model performance visualization
+# --------------------------------------------------------------------------------
+
+
+import os
+import sys
+import yaml
+import hashlib
+import numpy as np
+from datetime import datetime
+import _ext.tools as _tools
+
+
+class RegressionIO:
+    def __init__(self, log_file):
+        self.sys_stdout = sys.stdout
+        self.ios = [sys.stdout, open(log_file, 'a', encoding='utf-8')]
+
+    def write(self, text):
+        current_time = get_current_time()
+        if text not in ['\n', '\r']:
+            text = f'[{current_time}] {text}'
+        for io in self.ios:
+            io.write(text)
+
+    def flush(self):
+        for io in self.ios:
+            io.flush()
+
+    def close(self):
+        for io in self.ios[1:]:
+            io.close()
+
+    def __enter__(self):
+        sys.stdout = self
+        return self
+
+    def __exit__(self, exc_type, exc_val, exc_tb):
+        sys.stdout = self.sys_stdout
+        self.close()
+        return False
+
+
+def get_current_dir(current_file):
+    current_dir = os.path.dirname(os.path.abspath(current_file))
+    return current_dir
+
+def get_current_time():
+    current_time = datetime.now().strftime('%Y-%m-%d %H:%M:%S')
+    return current_time
+
+def write_log(log_file, log):
+    with open(log_file, 'w', encoding='utf-8') as f:
+        f.write(log)
+
+def load_config(current_dir):
+    with open(f'{current_dir}/config.yaml', 'r', encoding='utf-8') as f:
+        config = yaml.safe_load(f)
+    config['start_date'] = str(config['start_date'])
+    config['end_date'] = str(config['end_date'])
+    return config
+
+def initialize_random_seed(random_seed):
+    if random_seed is None:
+        random_seed = int(hashlib.md5(f'{datetime.now().timestamp()}_{os.urandom(8).hex()}'.encode()).hexdigest()[:8], 16)
+    np.random.seed(random_seed)
+    return random_seed
+
+def update_params_dict(params_dict, default_params_dict):
+    new_params_dict = {}
+    for key in default_params_dict:
+        if isinstance(default_params_dict[key], dict):
+            new_params_dict[key] = update_params_dict(params_dict.get(key, {}), default_params_dict[key])
+        else:
+            new_params_dict[key] = params_dict.get(key, default_params_dict[key])
+    return new_params_dict
+
+def make_results_dirs(output_dir, model_name, horizons):
+    model_dir = f'{output_dir}/{model_name}'
+    os.makedirs(model_dir, exist_ok=True)
+    os.makedirs(f'{model_dir}/results', exist_ok=True)
+    for horizon in horizons:
+        ret_col = f'ret-{horizon}'
+        os.makedirs(f'{model_dir}/results/{ret_col}', exist_ok=True)
+        os.makedirs(f'{model_dir}/results/{ret_col}/models', exist_ok=True)
+        os.makedirs(f'{model_dir}/results/{ret_col}/plots', exist_ok=True)
+
+def copy_scripts(current_dir, output_dir, model_name, random_seed, scripts_folder='scripts'):
+    model_dir = f'{output_dir}/{model_name}'
+    os.makedirs(f'{model_dir}/{scripts_folder}', exist_ok=True)
+    file_and_folders = [f for f in os.listdir(current_dir) if not f.startswith('.')]
+    for f in file_and_folders:
+        if os.path.isfile(f'{current_dir}/{f}'):
+            with open(f'{current_dir}/{f}', 'r', encoding='utf-8') as file:
+                text = file.read()
+                if f == 'config.yaml':
+                    row = [row for row in text.split('\n') if row.startswith('random_seed:')][0]
+                    text = text.replace(row, f'random_seed: {random_seed}')
+            with open(f'{model_dir}/{scripts_folder}/{f}', 'w', encoding='utf-8') as file:
+                file.write(text)
+        else:
+            copy_scripts(f'{current_dir}/{f}', output_dir, model_name, random_seed, f'{scripts_folder}/{f}')
+
+def validate_and_save_features(output_dir, model_name, full_features_df, mft_full_features_df, mft_feature_plan):
+    comparison_df = _tools.validate_and_save_features(output_dir, model_name, full_features_df, mft_full_features_df, mft_feature_plan)
+    return comparison_df
+
+def summarize_and_save_results(output_dir, model_name, model_evaluations):
+    _tools.summarize_and_save_results(output_dir, model_name, model_evaluations)

coinsignal-1.4.9.dist-info/METADATA

@@ -0,0 +1,23 @@
+Metadata-Version: 2.4
+Name: coinsignal
+Version: 1.4.9
+Summary: coin data analysis and return predicting framework
+Author: Zehao Yang
+Requires-Python: >=3.8
+Description-Content-Type: text/plain
+Requires-Dist: numpy>=1.20.0
+Requires-Dist: pandas>=1.3.0
+Requires-Dist: lightgbm>=3.3.0
+Requires-Dist: scikit-learn>=1.0.0
+Requires-Dist: matplotlib>=3.3.0
+Requires-Dist: pyyaml>=5.4.0
+Requires-Dist: joblib>=1.0.0
+Requires-Dist: cython>=0.29.0
+Dynamic: author
+Dynamic: description
+Dynamic: description-content-type
+Dynamic: requires-dist
+Dynamic: requires-python
+Dynamic: summary
+
+coin data analysis and return predicting framework

coinsignal-1.4.9.dist-info/RECORD

@@ -0,0 +1,13 @@
+coinsignal-1.4.9.dist-info/METADATA,sha256=e9RXNmIpc-qOCrPxJDL4CiQgfpwNAPn3HGxm8d0h9xQ,626
+coinsignal-1.4.9.dist-info/WHEEL,sha256=abiPiewrGIMBJasL3sP4UnsZN4xsTtRXnS1JXgFYjtg,112
+coinsignal-1.4.9.dist-info/top_level.txt,sha256=8xBBVxKSVz2DM3nreUe-A7OGl7vCbK-3njdPu-JtUjw,11
+coinsignal/__init__.py,sha256=LzN6rcSEJm9WonEOjhtXGJg7ev5ZR07IpI1PkUibr9U,269
+coinsignal/_ext.so,sha256=fxnET7txIUeORb73boZww6dT00jgM8JcQixuPgoBf4Q,12959376
+coinsignal/data.py,sha256=VUC5aC2yVFdzFhHHF01hdjGmUOUU9KxNx13rggj3cMM,1077
+coinsignal/feature.py,sha256=kgBVCBXZWBLauptusI5p2Sof0DXJXo7jjq5YZivnaEc,1012
+coinsignal/label.py,sha256=NRZgnje1baDasecNfiPxUcT-wj0NA-CvSgDLLmpsvBo,1207
+coinsignal/model.py,sha256=F-DJhtr_QWwM1wc_ws_XkZqvWRJQKfX7NZ8m1BIO_SQ,2051
+coinsignal/parallel.py,sha256=KKLRUc_LGqSfcfuhHi1vUGyJZHPhsvHbn6imQs1_6JQ,1592
+coinsignal/sampler.py,sha256=3fRBUk2kFBZ_3ef6WdFYx0Hk2Sz1Cq8nB9PV4MFek7I,758
+coinsignal/tools.py,sha256=l4nGZtcXUE5DiUx_yfcniu5klDASI3FgoLwHDJ83oxI,4485
+coinsignal-1.4.9.dist-info/RECORD,,

coinsignal-1.4.9.dist-info/WHEEL

@@ -0,0 +1,5 @@
+Wheel-Version: 1.0
+Generator: setuptools (80.9.0)
+Root-Is-Purelib: false
+Tag: cp312-cp312-manylinux2014_x86_64
+

coinsignal-1.4.9.dist-info/top_level.txt

@@ -0,0 +1 @@
+coinsignal