Add variants that retrieve price information in real time

- Add `mod_investpy_pme` module with `investpy_pme` and `investpy_verbose_pme` using `investpy` module, which uses Investing.com as the data source. - Also, fix a couple of issues with hypothesis testing and making sure the input dates are sorted.
ymyke · Mar 13, 2022 · 885e6e6 · 885e6e6
2 parents 1cf75c8 + 0c8cee8
commit 885e6e6
Show file tree

Hide file tree

Showing 7 changed files with 203 additions and 6 deletions.
diff --git a/README.md b/README.md
@@ -40,6 +40,45 @@ Notes:
   case, the IRR is for the underlying period.
 - `verbose_pme`: Calculate PME for evenly spaced cashflows and return vebose
   information.
+- `investpy_pme` and `investpy_verbose_pme`: Use price information from Investing.com.
+  See below.
+
+## Investpy examples -- using investpy to retrieve PME prices online
+
+Use `investpy_pme` and `investpy_verbose_pme` to use a ticker from Investing.com and
+compare with those prices. Like so:
+
+```python
+from datetime import date
+from pypme import investpy_pme
+
+common_args = {
+    "dates": [date(2012, 1, 1), date(2013, 1, 1)],
+    "cashflows": [-100],
+    "prices": [1, 1],
+}
+print(investpy_pme(pme_ticker="Global X Lithium", pme_type="etf", **common_args))
+print(investpy_pme(pme_ticker="bitcoin", pme_type="crypto", **common_args))
+print(investpy_pme(pme_ticker="SRENH", pme_type="stock", pme_country="switzerland", **common_args))
+```
+
+Produces:
+
+```
+-0.02834024870462727
+1.5031336254547634
+0.3402634808264912
+```
+
+The investpy functions take the following parameters:
+- `pme_type`: One of `stock`, `etf`, `fund`, `crypto`, `bond`, `index`, `certificate`.
+  Defaults to `stock`.
+- `pme_ticker`: The ticker symbol/name.
+- `pme_country`: The ticker's country of residence. Defaults to `united states`.
+
+Check out [the Investpy project](https://github.com/alvarobartt/investpy) for more
+details.
+
 
 ## Garbage in, garbage out
 

diff --git a/pypme/__init__.py b/pypme/__init__.py
@@ -1,2 +1,3 @@
-__version__ = '0.1.3'
+__version__ = '0.2.0'
 from .pme import verbose_pme, pme, verbose_xpme, xpme
+from .mod_investpy_pme import investpy_verbose_pme, investpy_pme
diff --git a/pypme/mod_investpy_pme.py b/pypme/mod_investpy_pme.py
@@ -0,0 +1,72 @@
+"""Calculate PME and get the prices from Investing.com via the `investpy` module.
+
+Important: The Investing API has rate limiting measures in place and will block you if
+you hit the API too often. You will notice by getting 429 errors (or maybe
+also/alternatively 503). Wait roughly 2 seconds between each consecutive call to the API
+via the functions in this module.
+
+Args: 
+- pme_type: One of "stock", "etf", "fund", "crypto", "bond", "index", "certificate".
+  Defaults to "stock".
+- pme_ticker: The ticker symbol/name.
+- pme_country: The ticker's country of residence. Defaults to "united states".
+
+Refer to the `pme` module to understand other arguments and what the functions return.
+"""
+
+from typing import List, Tuple
+from datetime import date
+import pandas as pd
+import investpy
+from .pme import verbose_xpme
+
+
+def get_historical_data(ticker: str, type: str, **kwargs) -> pd.DataFrame:
+    """Small wrapper to make the investpy interface accessible in a more unified fashion."""
+    kwargs[type] = ticker
+    if type == "crypto" and "country" in kwargs:
+        del kwargs["country"]
+    return getattr(investpy, "get_" + type + "_historical_data")(**kwargs)
+
+
+def investpy_verbose_pme(
+    dates: List[date],
+    cashflows: List[float],
+    prices: List[float],
+    pme_ticker: str,
+    pme_type: str = "stock",
+    pme_country: str = "united states",
+) -> Tuple[float, float, pd.DataFrame]:
+    """Calculate PME return vebose information, retrieving PME price information from
+    Investing.com in real time.
+    """
+    dates_as_str = [x.strftime("%d/%m/%Y") for x in sorted(dates)]
+    pmedf = get_historical_data(
+        pme_ticker,
+        pme_type,
+        country=pme_country,
+        from_date=dates_as_str[0],
+        to_date=dates_as_str[-1],
+    )
+    # Pick the nearest price if there is no price for an exact date:
+    pme_prices = [
+        pmedf.iloc[pmedf.index.get_indexer([x], method="nearest")[0]]["Close"]
+        for x in dates_as_str
+    ]
+    return verbose_xpme(dates, cashflows, prices, pme_prices)
+
+
+def investpy_pme(
+    dates: List[date],
+    cashflows: List[float],
+    prices: List[float],
+    pme_ticker: str,
+    pme_type: str = "stock",
+    pme_country: str = "united states",
+) -> Tuple[float, float, pd.DataFrame]:
+    """Calculate PME and return the PME IRR only, retrieving PME price information from
+    Investing.com in real time.
+    """
+    return investpy_verbose_pme(
+        dates, cashflows, prices, pme_ticker, pme_type, pme_country
+    )[0]
diff --git a/pypme/pme.py b/pypme/pme.py
@@ -1,5 +1,7 @@
 """Calculate PME (Public Market Equivalent) for both evenly and unevenly spaced
-cashflows. Calculation according to
+cashflows. 
+
+Calculation according to
 https://en.wikipedia.org/wiki/Public_Market_Equivalent#Modified_PME
 
 Args:
@@ -11,7 +13,6 @@
 Note:
 - Both `prices` and `pme_prices` need an additional item at the end for the last
   interval / point in time, for which the PME is calculated.
-- Obviously, all prices must be in the same (implicit) currency.
 - `cashflows` has one fewer entry than the other lists because the last cashflow is
   implicitly assumed to be the current NAV at that time.
 
@@ -127,6 +128,8 @@ def verbose_xpme(
     Requires the points in time as `dates` as an input parameter in addition to the ones
     required by `pme()`.
     """
+    if dates != sorted(dates):
+        raise ValueError("Dates must be in order")
     if len(dates) != len(prices):
         raise ValueError("Inconsistent input data")
     df = verbose_pme(cashflows, prices, pme_prices)[2]

diff --git a/pyproject.toml b/pyproject.toml
@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "pypme"
-version = "0.1.3"
+version = "0.2.0"
 description = "Python library for PME (Public Market Equivalent) calculation"
 authors = ["ymyke"]
 license = "MIT"
@@ -14,12 +14,14 @@ python = ">=3.8, <3.10"
 xirr = "^0.1.8"
 numpy-financial = "^1.0.0"
 pandas = "^1.4.1"
+investpy = "^1.0.8"
 
 [tool.poetry.dev-dependencies]
 pytest = "^5.2"
 ipykernel = "^6.9.1"
 black = "^22.1.0"
 hypothesis = "^6.39.3"
+pytest-mock = "^3.7.0"
 
 [build-system]
 requires = ["poetry-core>=1.0.0"]

diff --git a/tests/test_investpy_pypme.py b/tests/test_investpy_pypme.py
@@ -0,0 +1,72 @@
+import pytest
+from datetime import date
+import pandas as pd
+from pypme.mod_investpy_pme import investpy_pme, investpy_verbose_pme
+
+
+@pytest.mark.parametrize(
+    "dates, cashflows, prices, pme_timestamps, pme_prices, target_pme_irr, target_asset_irr",
+    [
+        (
+            [date(2012, 1, 1), date(2013, 1, 1)],
+            [-100],
+            [1, 1],
+            ["2012-01-01"],
+            [20],
+            0,
+            # B/c the function will search for the nearest date which is always the same
+            # one b/c there is only one and therefore produce a PME IRR of 0
+            0,
+        ),
+        (
+            [date(2012, 1, 1), date(2013, 1, 1)],
+            [-100],
+            [1, 1],
+            ["2012-01-01", "2012-01-02"],
+            [20, 40],
+            99.62,
+            # In this case, the "nearest" option in `investpy_verbose_pme`'s call to
+            # `get_indexer` finds the entry at 2012-01-02. Even though it's far away
+            # from 2013-01-01, it's still the closest.
+            0,
+        ),
+    ],
+)
+def test_investpy_pme(
+    mocker,
+    dates,
+    cashflows,
+    prices,
+    pme_timestamps,
+    pme_prices,
+    target_pme_irr,
+    target_asset_irr,
+):
+    """Test both the verbose and non-verbose variant at the same to keep things simple.
+
+    Note that this test does _not_ hit the network / investing API since the relevant
+    function gets mocked.
+    """
+    mocker.patch(
+        "pypme.mod_investpy_pme.get_historical_data",
+        return_value=pd.DataFrame(
+            {"Close": {pd.Timestamp(x): y for x, y in zip(pme_timestamps, pme_prices)}}
+        ),
+    )
+    pme_irr, asset_irr, df = investpy_verbose_pme(
+        dates=dates,
+        cashflows=cashflows,
+        prices=prices,
+        pme_ticker="dummy",
+    )
+    assert round(pme_irr * 100.0, 2) == round(target_pme_irr, 2)
+    assert round(asset_irr * 100.0, 2) == round(target_asset_irr, 2)
+    assert isinstance(df, pd.DataFrame)
+
+    pme_irr = investpy_pme(
+        dates=dates,
+        cashflows=cashflows,
+        prices=prices,
+        pme_ticker="dummy",
+    )
+    assert round(pme_irr * 100.0, 2) == round(target_pme_irr, 2)
diff --git a/tests/test_pypme.py b/tests/test_pypme.py
@@ -7,7 +7,7 @@
 
 
 def test_version():
-    assert __version__ == "0.1.3"
+    assert __version__ == "0.2.0"
 
 
 @pytest.mark.parametrize(
@@ -92,12 +92,18 @@ def test_for_valueerrors(list1, list2, list3, exc_pattern):
     assert exc_pattern in str(exc)
 
 
+def test_for_non_sorted_dates():
+    with pytest.raises(ValueError) as exc:
+        xpme([date(2000, 1, 1), date(1900, 1, 1)], [], [], [])
+    assert "Dates must be in order" in str(exc)
+
+
 @st.composite
 def same_len_lists(draw):
     n = draw(st.integers(min_value=2, max_value=100))
     floatlist = st.lists(st.floats(), min_size=n, max_size=n)
     datelist = st.lists(st.dates(), min_size=n, max_size=n)
-    return (draw(datelist), draw(floatlist), draw(floatlist), draw(floatlist))
+    return (sorted(draw(datelist)), draw(floatlist), draw(floatlist), draw(floatlist))
 
 
 @given(same_len_lists())
@@ -109,6 +115,8 @@ def test_xpme_hypothesis_driven(lists):
         )
     except ValueError as exc:
         assert "least one cashflow" in str(exc) or "All prices" in str(exc)
+    except OverflowError as exc:
+        assert "Result too large" in str(exc)
     else:
         assert xnpv(df["PME", "CF"], pme_irr) == 0
         assert xnpv(df["Asset", "CF"], asset_irr) == 0