> ## Documentation Index > Fetch the complete documentation index at: https://systematica.mintlify.site/llms.txt > Use this file to discover all available pages before exploring further. # Base > systematica.generic.base ## `Feature` ```python theme={null} Feature( id: str = None, loader_or_data: Callable = , timeframe: str = '1d', start: str = None, end: str = None, s1: str = 'BTCUSDT', s2: str = 'ETHUSDT', suffix: str = 'h5', use_close: bool = True, freq: str = None, signal_model: str = None, long_entries: float = nan, long_exits: float = nan, short_entries: float = nan, short_exits: float = nan, clean: bool = True, select_symbols: str | int | List[str | int] = None, portfolio_config: Dict[str, Any] = _Nothing.NOTHING, model_params: Dict[str, Any] = _Nothing.NOTHING, metrics: str | List[str] = None, metrics_kwargs: Dict[str, Any] = _Nothing.NOTHING, use_rolling: bool = False, storage: str | optuna.storages._base.BaseStorage = None, pruner: optuna.pruners._base.BasePruner = , sampler: optuna.samplers._base.BaseSampler = , study_name: str = None, direction: str | List[str] = 'maximize', load_if_exists: bool = False, n_trials: int = 100, n_completed_trials: int = None, timeout: float = None, n_jobs: int = 1, catch: Iterable[Type[Exception]] | Type[Exception] = _Nothing.NOTHING, callbacks: Iterable[Callable[[optuna.study.study.Study, optuna.trial._frozen.FrozenTrial], None]] | None = None, gc_after_trial: bool = False, show_progress_bar: bool = False, verbose: bool = True, tracker: Type = None, ) ``` Configuration for a feature calculation. This class encapsulates the necessary configuration for computing a feature. It includes the feature's name, a callable to compute the feature, a list of required data keys, and optional parameters for the computation. Method generated by attrs for class Feature. ### Descendants * `systematica.generic.base.Model` ### Instance variables * `symbols: List[str]`: Symbols. * `loader_name: str`: Loader name. * `frequency: str`: Frequency. * `optuna_metadata: Dict`: Metadata for Optuna study tracking. Returns dictionary containing key configuration parameters excluding runtime objects (tracker, sampler, pruner, etc.). * `callbacks: Iterable[Callable[[optuna.study.study.Study, optuna.trial._frozen.FrozenTrial], None]] | None`: List of callback functions that are invoked at the end of each trial. Each function must accept two parameters with the following types in this order: Study and FrozenTrial. * `catch: Iterable[Type[Exception]] | Type[Exception]`: A study continues to run even when a trial raises one of the exceptions specified in this argument. Default is an empty tuple, i.e. the study will stop for any exception except for TrialPruned. * `clean: bool`: Clean signals. Defaults to `True`. * `close: vectorbtpro.data.base.Data`: * `data: vectorbtpro.data.base.Data`: * `data_symbols: vectorbtpro.data.base.Data`: * `direction: str | List[str]`: Optimization objective, either `minimize` or `maximize`, default is `maximize`. * `end: str`: The end date for the data in string format (e.g., `YYYY-MM UTC`). If None, use `datetie.utcnow()` to fetch the latest data possible. Default to `None`. * `end_date: str`: * `freq: str`: Frequency of the data. Used when `cross_validate` is True. Defaults to `None`. * `gc_after_trial: bool`: Flag to determine whether to automatically run garbage collection after each trial. Set to `True` to run the garbage collection, `False` otherwise. It runs a full collection by internally calling `gc.collect`. If you see an increase in memory consumption over several trials, try setting this flag to `True`. Defaults to `False`. * `high: vectorbtpro.data.base.Data`: * `id: str`: Custom searchable id. * `load_if_exists: bool`: Flag to control the behavior to handle a conflict of study names. In the case where a study named study\_name already exists in the storage, a DuplicatedStudyError is raised if load\_if\_exists is set to False. Otherwise, the creation of the study is skipped, and the existing one is returned. * `loader_or_data: Callable`: Default loader or data. If loader, it must accept timeframe as retriever. More information in `data/loader`. Defaults to `load_clean_data`. * `long_entries: float`: Long entry signals. * `long_exits: float`: Long exit signals. * `low: vectorbtpro.data.base.Data`: * `metrics: str | List[str]`: Performance metrics to compute. Defaults to `None`. * `metrics_kwargs: Dict[str, Any]`: Parameters for metric config. If `None`, uses default parameters based on the model type. This is used to specify parameters such as `window` or `minp` for rolling metrics. If using cross-validation, this should be set to `None` as it is not applicable. If using rolling metrics, this should be set to a dictionary with parameters for the metric config. For example, `{"window": 20}`. Defaults to `None`. * `model_params: Dict[str, Any]`: A dictionary of optional keyword arguments needed to compute the model. * `n_completed_trials: int`: The number of successfully completed trials before stopping the optimization. If `None`, it will run for `n_trials` trials. TO be effective, the number of trials should be higher than the number of completed trials. default is `None`. * `n_jobs: int`: Number of parallel jobs for optimization (`-1` for using all available `CPUs` cores), default is `-1`. * `n_trials: int`: The number of trials for the optimization. This is the maximum number of trials that will be run. If `n_completed_trials` is set, the optimization will stop when the number of completed trials reaches this value. If `n_completed_trials` is not set, the optimization will run for `n_trials` trials. If `n_trials` is set to `None`, it will run indefinitely until `n_completed_trials` is reached. If `n_trials` is set to `0`, it will not run any trials. If `n_trials` is set to a negative value, it will run indefinitely until `n_completed_trials` is reached. Defaults to `100`. * `open: vectorbtpro.data.base.Data`: * `portfolio_config: Dict[str, Any]`: Default VectorBT portfolio parameters used in the systematica package. These parameters are used to define the portfolio simulation and its behavior. The default values are set to None or specific values as per the requirements of the systematica package. * `price: vectorbtpro.data.base.Data`: * `pruner: optuna.pruners._base.BasePruner`: Optuna Pruner. If `None`, defaults to Asynchronous Successive Halving Algorithm. Defaults to `None`. * `s1: str`: First symbol. * `s2: str`: Second symbol. * `sampler: optuna.samplers._base.BaseSampler`: Optuna Sampler. If `None`, defaults to TPE (Tree-structured Parzen Estimator) algorithm. Defaults to `None`. * `select_symbols: str | int | List[str | int]`: Symbol name (e.g., `BTCUSDT` or `ETHUSDT`) or column index (e.g., `0` or `1`) to keep. If None, no filtering. * `short_entries: float`: Short entry signals. * `short_exits: float`: Short exit signals. * `show_progress_bar: bool`: Flag to show progress bars or not. To show progress bar, set this True. Note that it is disabled when n\_trials is None, timeout is not None, and n\_jobs. * `signal_model: str`: Signal model used for generating signals. * `start: str`: The start date for the data in string format (e.g., `YYYY-MM UTC`). Default to `None`. * `start_date: Dict[str, Any]`: * `storage: str | optuna.storages._base.BaseStorage`: Database URL. If this argument is set to None, `InMemoryStorage` is used, and the Study will not be persistent. * `study_name: str`: The name of the (optuna and tracker) study. * `suffix: str`: Extension (suffix) of the file to load. Defaults to `None`. * `timeframe: str`: Timeframe of the file to be loaded. Defaults to `None`. * `timeout: float`: Stop study after the given number of second(s). None represents no limit in terms of elapsed time. The study continues to create trials until the number of trials reaches n\_trials, timeout period elapses, stop() is called or, a termination signal such as SIGTERM or Ctrl+C is received. * `tracker: Type`: Tracker instance for logging and monitoring the optimization process. If `None`, no tracking is performed. If you want to use a tracker, you can pass an instance of a tracker class that inherits from `BaseTracker`. For example, you can use `NeptuneTracker`. This tracker will log the optimization process and metrics to the tracker service. Default to None. * `use_close: bool`: Use close price if set to True. Otherwise, use open price. * `use_rolling: bool`: Flag to use rolling method via `BaseWalkForwardRolling`. Note that both rolling and cv models can be passed. Fallbacks to `BaseWalkForwardCV` if model is not using `from_rolling` splitter without error message. Defaults to `False`. * `verbose: bool`: Whether to print Optuna logs during the optimization process. If True, prints logs to the console. Default is `True`. ### Methods #### `check_direction` ```python theme={null} check_direction( self, attribute: str, value: str | list[str], ) ``` Validate optimization directions. **Parameters**: | Name | Type | Default | Description | | ----------- | ----------------- | ------- | ------------------------------ | | `attribute` | `attrs.Attribute` | `--` | The attribute being validated. | | `value` | `BaseStatArb` | `--` | The model instance. | **Raises**: | Type | Description | | ------------ | --------------------------------------------- | | `ValueError` | If direction is not "maximize" or "minimize". | | `TypeError` | If direction is not a string or a list | #### `check_metrics` ```python theme={null} check_metrics( self, attribute: str, value: str | list[str], ) ``` Validate metrics. **Parameters**: | Name | Type | Default | Description | | ----------- | ----------------- | ------- | ------------------------------ | | `attribute` | `attrs.Attribute` | `--` | The attribute being validated. | | `value` | `BaseStatArb` | `--` | The model instance. | **Raises**: | Type | Description | | ------------ | ----------------------------------------------------- | | `TypeError` | If metrics is not a string or list. | | `ValueError` | If metric has not been configured and is not tunable. | #### `asdict` ```python theme={null} asdict( self, ) ‑> Dict[str, Any] ``` Convert instance to a shallow dict. **Returns**: | Type | Description | | ------------ | ------------- | | `Ttp.Kwargs` | Feature dict. | ## `FeatureExpression` ```python theme={null} FeatureExpression( expr: str, include_pckg: Dict[str, Any] = None, kwargs: Dict[str, Any] = None, ) ``` StatArb expression model. Method generated by attrs for class FeatureExpression. ### Instance variables * `expr: str`: Expression string. Must contain valid Python code and can be single-line or multi-line. * `include_pckg: Dict[str, Any]`: External packages used in expr. * `kwargs: Dict[str, Any]`: Additional key-word arguments used in `vbt.IF.from_expr` ## `Model` ```python theme={null} Model( id: str = None, loader_or_data: Callable = , timeframe: str = '1d', start: str = None, end: str = None, s1: str = 'BTCUSDT', s2: str = 'ETHUSDT', suffix: str = 'h5', use_close: bool = True, freq: str = None, signal_model: str = None, long_entries: float = nan, long_exits: float = nan, short_entries: float = nan, short_exits: float = nan, clean: bool = True, select_symbols: str | int | List[str | int] = None, portfolio_config: Dict[str, Any] = _Nothing.NOTHING, model_params: Dict[str, Any] = _Nothing.NOTHING, metrics: str | List[str] = None, metrics_kwargs: Dict[str, Any] = _Nothing.NOTHING, use_rolling: bool = False, storage: str | optuna.storages._base.BaseStorage = None, pruner: optuna.pruners._base.BasePruner = , sampler: optuna.samplers._base.BaseSampler = , study_name: str = None, direction: str | List[str] = 'maximize', load_if_exists: bool = False, n_trials: int = 100, n_completed_trials: int = None, timeout: float = None, n_jobs: int = 1, catch: Iterable[Type[Exception]] | Type[Exception] = _Nothing.NOTHING, callbacks: Iterable[Callable[[optuna.study.study.Study, optuna.trial._frozen.FrozenTrial], None]] | None = None, gc_after_trial: bool = False, show_progress_bar: bool = False, verbose: bool = True, tracker: Type = None, model: Any = None, preprocess_func: Callable = None, ) ``` Configuration for a feature calculation. This class encapsulates the necessary configuration for computing a feature. It includes the feature's name, a callable to compute the feature, a list of required data keys, and optional parameters for the computation. Method generated by attrs for class Model. ### Ancestors * `systematica.generic.base.Feature` ### Descendants * `systematica.walk_forward.base.BaseWalkForward` ### Instance variables * `model_name: str`: Model name. * `model: Any`: Compute model, which defines the method to compute the feature. * `model_cls: Type`: * `model_output: numpy.ndarray | pandas.core.frame.DataFrame | systematica.signals.base.Signals`: * `preprocess_data: pandas.core.frame.DataFrame`: * `preprocess_func: Callable`: Preprocessing function applied to price data. If None, the model uses price data by default. Defaults to None. ### Methods #### `updated_metrics_kwargs` ```python theme={null} updated_metrics_kwargs( self, use_rolling: bool = None, ) ``` Set metrics kwargs. **Parameters**: | Name | Type | Default | Description | | ----------- | ----------------- | ------- | ------------------------------ | | `attribute` | `attrs.Attribute` | `--` | The attribute being validated. | | `value` | `BaseStatArb` | `--` | The model instance. | **Raises**: | Type | Description | | ------------ | ----------------------------------------------------- | | `TypeError` | If metrics is not a string or list. | | `ValueError` | If metric has not been configured and is not tunable. | #### `get_model_output` ```python theme={null} get_model_output( self, to_numpy: bool = True, **kwargs, ) ‑> numpy.ndarray | pandas.core.series.Series | pandas.core.frame.DataFrame ``` Computes the model output. **Parameters**: | Name | Type | Default | Description | | ---------- | ------ | ------- | ------------------------------------------------------------------------------------------- | | `to_numpy` | `bool` | `False` | If `True`, converts output to a `np.array`. Default is `False`, returning a `pd.DataFrame`. | **Returns**: | Type | Description | | ---------------------------- | ------------------ | | `np.ndarray or pd.DataFrame` | Model output data. | #### `get_signals` ```python theme={null} get_signals( self, validate_model: bool = True, ) ‑> systematica.signals.base.Signals ``` Generated trading signals. **Parameters**: | Name | Type | Default | Description | | ---------------- | ------ | ------- | -------------------------------------------------------- | | `validate_model` | `bool` | `True` | Check if model parameters are valid. Defaults to `True`. | **Returns**: | Type | Description | | --------- | -------------------------- | | `Signals` | Generated trading signals. |