Qubx 0.1.84__cp311-cp311-manylinux_2_35_x86_64.whl → 0.1.86__cp311-cp311-manylinux_2_35_x86_64.whl

qubx/_nb_magic.py

@@ -1,7 +1,6 @@
 """"
 Here stuff we want to have in every Jupyter notebook after calling %qube magic
 """
-import importlib_metadata
 
 import qubx
 from qubx.utils import runtime_env
@@ -15,11 +14,19 @@ def np_fmt_short():
 
 def np_fmt_reset():
     # reset default np printing options
-    np.set_printoptions(edgeitems=3, infstr='inf', linewidth=75, nanstr='nan', precision=8,
-                        suppress=False, threshold=1000, formatter=None)
+    np.set_printoptions(
+        edgeitems=3,
+        infstr="inf",
+        linewidth=75,
+        nanstr="nan",
+        precision=8,
+        suppress=False,
+        threshold=1000,
+        formatter=None,
+    )
 
 
-if runtime_env() in ['notebook', 'shell']:
+if runtime_env() in ["notebook", "shell"]:
 
     # - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
     # -- all imports below will appear in notebook after calling %%alphalab magic ---
@@ -39,19 +46,18 @@ if runtime_env() in ['notebook', 'shell']:
     # - - - - Learn stuff - - - -
     # - - - - Charting stuff - - - -
     from matplotlib import pyplot as plt
-    from qubx.utils.charting.mpl_helpers import fig, subplot, sbp
+    from qubx.utils.charting.mpl_helpers import fig, subplot, sbp, plot_trends, ohlc_plot
 
     # - - - - Utils - - - -
     from qubx.pandaz.utils import scols, srows, ohlc_resample, continuous_periods, generate_equal_date_ranges
 
     # - setup short numpy output format
     np_fmt_short()
-    
+
     # - add project home to system path
     add_project_to_system_path()
 
     # show logo first time
-    if not hasattr(qubx.QubxMagics, '__already_initialized__'):
+    if not hasattr(qubx.QubxMagics, "__already_initialized__"):
         setattr(qubx.QubxMagics, "__already_initialized__", True)
         logo()
-

qubx/core/series.pxd

@@ -15,7 +15,7 @@ cdef class TimeSeries:
     cdef public long long timeframe
     cdef public Indexed times
     cdef public Indexed values
-    cdef float max_series_length
+    cdef public float max_series_length
     cdef unsigned short _is_new_item
     cdef public str name
     cdef dict indicators        # it's used for indicators caching
@@ -28,8 +28,12 @@ cdef class TimeSeries:
 
 
 cdef class Indicator(TimeSeries):
-    cdef TimeSeries series
-    cdef TimeSeries parent
+    cdef public TimeSeries series
+    cdef public TimeSeries parent
+
+
+cdef class IndicatorOHLC(Indicator):
+    pass
 
 
 cdef class RollingSum:

qubx/core/series.pyx

@@ -142,6 +142,14 @@ cdef class TimeSeries:
     def __len__(self) -> int:
         return len(self.times)
 
+    def loc(self, str t):
+        _t = np.datetime64(t, 'ns').item()
+        ix = int(np.searchsorted(self.times.values, _t, side='right'))
+        ix = min(ix, len(self.values))
+        if ix == 0:
+            raise ValueError(f"Time {t} not found in {self.name} !")
+        return np.datetime64(self.times.values[ix - 1], 'ns'), self.values.values[ix - 1]
+
     def _on_attach_indicator(self, indicator: Indicator, indicator_input: TimeSeries):
         self.calculation_order.append((
             id(indicator_input), indicator, id(indicator)
@@ -169,7 +177,7 @@ cdef class TimeSeries:
             # - disable first notification because first item may be incomplete
             self._is_new_item = False
 
-        elif time - self.times[0] >= self.timeframe:
+        elif (_dt := time - self.times[0]) >= self.timeframe:
             # - add new item
             self._add_new_item(item_start_time, value)
 
@@ -186,6 +194,8 @@ cdef class TimeSeries:
 
             return self._is_new_item
         else:
+            if _dt < 0:
+                raise ValueError(f"Attempt to update past data at {time_to_str(time)} !")
             self._update_last_item(item_start_time, value)
 
         # - update indicators by new data
@@ -300,6 +310,9 @@ def _wrap_indicator(series: TimeSeries, clz, *args, **kwargs):
 
 
 cdef class Indicator(TimeSeries):
+    """
+    Basic class for indicator that can be attached to TimeSeries
+    """
 
     def __init__(self, str name, TimeSeries series):
         if not name:
@@ -309,7 +322,7 @@ cdef class Indicator(TimeSeries):
         self.name = name
 
         # - we need to make a empty copy and fill it 
-        self.series = TimeSeries(series.name, series.timeframe, series.max_series_length)
+        self.series = self._instantiate_base_series(series.name, series.timeframe, series.max_series_length)
         self.parent = series 
         
         # - notify the parent series that indicator has been attached
@@ -318,6 +331,9 @@ cdef class Indicator(TimeSeries):
         # - recalculate indicator on data as if it would being streamed
         self._initial_data_recalculate(series)
 
+    def _instantiate_base_series(self, str name, long long timeframe, float max_series_length):
+        return TimeSeries(name, timeframe, max_series_length)
+
     def _on_attach_indicator(self, indicator: Indicator, indicator_input: TimeSeries):
         self.parent._on_attach_indicator(indicator, indicator_input)
 
@@ -345,6 +361,17 @@ cdef class Indicator(TimeSeries):
         return _wrap_indicator(series, clz, *args, **kwargs)
 
 
+cdef class IndicatorOHLC(Indicator):
+    """
+    Extension of indicator class to be used for OHLCV series
+    """
+    def _instantiate_base_series(self, str name, long long timeframe, float max_series_length):
+        return OHLCV(name, timeframe, max_series_length)
+
+    def calculate(self, long long time, Bar value, short new_item_started) -> object:
+        raise ValueError("Indicator must implement calculate() method")
+
+
 cdef class Lag(Indicator):
     cdef int period
 
@@ -729,7 +756,7 @@ cdef class OHLCV(TimeSeries):
             # Here we disable first notification because first item may be incomplete
             self._is_new_item = False
 
-        elif time - self.times[0] >= self.timeframe:
+        elif (_dt := time - self.times[0]) >= self.timeframe:
             b = Bar(bar_start_time, price, price, price, price, volume, bvolume)
 
             # - add new item
@@ -740,6 +767,9 @@ cdef class OHLCV(TimeSeries):
 
             return self._is_new_item
         else:
+            if _dt < 0:
+                raise ValueError(f"Attempt to update past data at {time_to_str(time)} !")
+
             self._update_last_item(bar_start_time, self[0].update(price, volume, bvolume))
 
         # - update indicators by new data

qubx/core/utils.pyx

@@ -50,5 +50,5 @@ cpdef recognize_timeframe(timeframe):
         tf = np.int64(timeframe.item().total_seconds() * NS) 
 
     else:
-        raise ValueError('Unknown timeframe type !')
+        raise ValueError(f'Unknown timeframe type: {timeframe} !')
     return tf

qubx/math/__init__.py

@@ -1 +1 @@
-from .math import percentile_rank
+from .stats import compare_to_norm, percentile_rank, kde

qubx/math/stats.py

@@ -30,13 +30,30 @@ def compare_to_norm(xs, xranges=None):
     fit = stats.norm.pdf(sorted(xs), _m, _s)
 
     sbp(12, 1)
-    plt.plot(sorted(xs), fit, 'r--', lw=2, label='N(%.2f, %.2f)' % (_m, _s))
-    plt.legend(loc='upper right')
+    plt.plot(sorted(xs), fit, "r--", lw=2, label="N(%.2f, %.2f)" % (_m, _s))
+    plt.legend(loc="upper right")
 
-    sns.kdeplot(xs, color='g', label='Data', shade=True)
+    sns.kdeplot(xs, color="g", label="Data", fill=True)
     if xranges is not None and len(xranges) > 1:
         plt.xlim(xranges)
-    plt.legend(loc='upper right')
+    plt.legend(loc="upper right")
 
     sbp(12, 2)
     stats.probplot(xs, dist="norm", sparams=(_m, _s), plot=plt)
+
+
+def kde(array, cut_down=True, bw_method="scott"):
+    """
+    Kernel dense estimation
+    """
+    from scipy.stats import gaussian_kde
+
+    if cut_down:
+        bins, counts = np.unique(array, return_counts=True)
+        f_mean = counts.mean()
+        f_above_mean = bins[counts > f_mean]
+        if len(f_above_mean) > 0:
+            bounds = [f_above_mean.min(), f_above_mean.max()]
+            array = array[np.bitwise_and(bounds[0] < array, array < bounds[1])]
+
+    return gaussian_kde(array, bw_method=bw_method)