1 files changed, 106 insertions, 0 deletions
diff --git a/python/index_data.py b/python/index_data.py
new file mode 100644
index 00000000..bf772100
--- /dev/null
+++ b/python/index_data.py
@@ -0,0 +1,106 @@
+from db import dbengine, dbconn
+from dates import bond_cal
+
+import datetime
+import pandas as pd
+serenitasdb = dbengine('serenitasdb')
+
+def insert_quotes():
+    # backpopulate some version i+1 quotes one day before they start trading so that
+    # we get continuous time series in the rb
+    eturns
+    dates = pd.DatetimeIndex(['2014-05-21', '2015-02-19', '2015-03-05','2015-06-23'])
+    df = pd.read_sql_query("SELECT DISTINCT ON (date) * FROM index_quotes " \
+                           "WHERE index='HY' AND tenor='5yr' " \
+                           "ORDER BY date, series DESC, version DESC",
+                           serenitasdb, parse_dates=['date'], index_col=['date'])
+    df = df.loc[dates]
+    for tup in df.itertuples():
+        result = serenitasdb.execute("SELECT indexfactor, cumulativeloss  FROM index_version " \
+                                     "WHERE index = 'HY' AND series=%s AND version in (%s, %s)" \
+                                     "ORDER BY version",
+                                     (tup.series, tup.version, tup.version+1))
+        factor1, cumloss1 = result.fetchone()
+        factor2, cumloss2 = result.fetchone()
+        recovery = 1-(cumloss2-cumloss1)
+        version2_price = (factor1 * tup.closeprice - 100*recovery)/factor2
+        print(version2_price)
+        serenitasdb.execute("INSERT INTO index_quotes(date, index, series, version, tenor, closeprice)" \
+                            "VALUES(%s, %s, %s, %s, %s, %s)",
+                            (tup.Index, 'HY', tup.series, tup.version+1, tup.tenor, version2_price))
+
+def get_index_quotes(index=None, series=None, tenor=None, date=None, years=3):
+    args = locals().copy()
+    if args['years'] is not None:
+        args['date'] = (pd.Timestamp.now() - pd.DateOffset(years=years)).date()
+    del args['years']
+
+    def make_str(key, val):
+        if isinstance(val, list):
+            op = "IN"
+            return "{} IN %({})s".format(key, key)
+        elif isinstance(val, datetime.date):
+            op = ">="
+        else:
+            op = "="
+        return "{} {} %({})s".format(key, op, key)
+
+    where_clause = " AND ".join(make_str(k, v)
+                                for k, v in args.items() if v is not None)
+    sql_str = "SELECT * FROM index_quotes"
+    if where_clause:
+        sql_str = " WHERE ".join([sql_str, where_clause])
+
+    def make_params(args):
+        return {k: tuple(v) if isinstance(v, list) else v
+                for k, v in args.items() if v is not None}
+
+    df = pd.read_sql_query(sql_str, serenitasdb, parse_dates=['date'],
+                           index_col=['date', 'index', 'series', 'version', 'tenor'],
+                           params = make_params(args))
+    df.sort_index(inplace=True)
+    ## get rid of US holidays
+    dates = df.index.levels[0]
+    if index in ['IG', 'HY']:
+        holidays = bond_cal().holidays(start=dates[0], end=dates[-1])
+        df = df.loc(axis=0)[dates.difference(holidays),:,:]
+    return df
+
+def index_returns(df=None, index=None, series=None, tenor=None, date=None, years=3):
+    """computes daily spreads and price returns
+
+    Parameters
+    ----------
+    df : pandas.DataFrame
+    index : str or List[str], optional
+        index type, one of 'IG', 'HY', 'EU', 'XO'
+    series : int or List[int], optional
+    tenor : str or List[str], optional
+        tenor in years e.g: '3yr', '5yr'
+    date : datetime.date, optional
+        starting date
+    years : int, optional
+        limits many years do we go back starting from today.
+
+    """
+    if df is None:
+        df = get_index_quotes(index, series, tenor, date, years)
+    df = (df.
+          groupby(level=['index', 'series', 'version', 'tenor'])
+          [['closespread','closeprice']].
+           pct_change())
+    df.columns = ['spread_return', 'price_return']
+    df = df.groupby(level=['date', 'index', 'series', 'tenor']).nth(-1)
+    coupon_data = pd.read_sql_query("SELECT index, series, tenor, coupon FROM " \
+                                    "index_maturity WHERE coupon is NOT NULL", serenitasdb,
+                                    index_col=['index', 'series', 'tenor'])
+    def add_accrued(df):
+        coupon = coupon_data.loc[df.index[0][1:],'coupon'] * 1e-4
+        accrued = (df.index.levels[0].to_series().diff().
+                   astype('timedelta64[D]')/360 * coupon)
+        return df + accrued
+
+    df['price_return'] = (df.
+                          groupby(level=['index', 'series', 'tenor'])['price_return'].
+                          transform(add_accrued))
+    return df