NaNs in prices

cvx/simulator/builder.py

@@ -293,7 +293,12 @@ def set_weights(self, time: datetime, weights: pd.Series) -> None:
         :param weights: series of weights
         """
         assert isinstance(weights, pd.Series), "weights must be a pandas Series"
-        self[time] = (self._state.nav * weights) / self._state.prices
+        valid = self._state.prices.dropna().index
+        # check that you have weights exactly for those indices
+        if not set(weights.dropna().index) == set(valid):
+            raise ValueError("weights must have same index as prices")
+
+        self[time] = (self._state.nav * weights[valid]) / self._state.prices[valid]
 
     def set_cashposition(self, time: datetime, cashposition: pd.Series) -> None:
         """
@@ -305,7 +310,13 @@ def set_cashposition(self, time: datetime, cashposition: pd.Series) -> None:
         assert isinstance(
             cashposition, pd.Series
         ), "cashposition must be a pandas Series"
-        self[time] = cashposition / self._state.prices
+
+        valid = self._state.prices.dropna().index
+        # check that you have weights exactly for those indices
+        if not set(cashposition.dropna().index) == set(valid):
+            raise ValueError("cashposition must have same index as prices")
+
+        self[time] = cashposition[valid] / self._state.prices[valid]
 
     def set_position(self, time: datetime, position: pd.Series) -> None:
         """
@@ -315,6 +326,12 @@ def set_position(self, time: datetime, position: pd.Series) -> None:
         :param position: series of number of stocks
         """
         assert isinstance(position, pd.Series), "position must be a pandas Series"
+
+        valid = self._state.prices.dropna().index
+        # check that you have weights exactly for those indices
+        if not set(position.dropna().index) == set(valid):
+            raise ValueError("position must have same index as prices")
+
         self[time] = position
 
     def __iter__(self) -> Generator[tuple[pd.DatetimeIndex, _State], None, None]:
@@ -365,6 +382,11 @@ def __setitem__(self, time: datetime, position: pd.Series) -> None:
         assert isinstance(position, pd.Series)
         assert set(position.index).issubset(set(self.assets))
 
+        valid = self._state.prices.dropna().index
+        # check that you have weights exactly for those indices
+        if not set(position.dropna().index) == set(valid):
+            raise ValueError("position must have same index as prices")
+
         if self.market_cap is not None:
             # compute capitalization of desired position
             cap = position * self._state.prices

tests/resources/priceNaN.csv

@@ -0,0 +1,6 @@
+,A,B,C,D,E,F,G
+2013-01-01,1673.78,23311.98,,3735.12,1462.42,2711.25,2518.99
+2013-01-02,1686.9,23311.98,,3735.12,1462.42,2711.25,2518.99
+2013-01-03,1663.95,23398.6,63312.46,3714.99,1459.37,,2509.51
+2013-01-04,1655.65,23331.09,62523.06,3689.34,1466.47,,2516.81
+2013-01-07,1646.95,23329.75,61932.54,3699.14,1461.89,,2523.77

tests/test_builder.py

@@ -274,3 +274,35 @@ def test_input_data(prices):
     for t, state in b:
         print(state.volume)
         pd.testing.assert_series_equal(state.prices, state.input_data["volume"])
+
+
+def test_weights_on_wrong_days(resource_dir):
+    prices = pd.read_csv(
+        resource_dir / "priceNaN.csv", index_col=0, parse_dates=True, header=0
+    )
+
+    b = _builder(prices=prices, initial_cash=50000)
+    t = prices.index
+
+    for t, state in b:
+        with pytest.raises(ValueError):
+            b.set_weights(
+                t[-1], pd.Series(index={"A", "B", "C"}, data=[0.5, 0.25, 0.25])
+            )
+
+        with pytest.raises(ValueError):
+            b.set_cashposition(t[-1], pd.Series(index={"A", "B", "C"}, data=[5, 5, 5]))
+
+        with pytest.raises(ValueError):
+            b.set_position(t[-1], pd.Series(index={"A", "B", "C"}, data=[5, 5, 5]))
+
+    for t, state in b:
+        b.set_weights(
+            t[-1],
+            pd.Series(
+                index=prices.loc[t[-1]].dropna().index,
+                data=np.random.rand(6),
+                dtype=float,
+            ),
+        )
+    # assert False