1 files changed, 133 insertions, 0 deletions
diff --git a/python/cds_curve.py b/python/cds_curve.py
new file mode 100644
index 00000000..4f96fbfb
--- /dev/null
+++ b/python/cds_curve.py
@@ -0,0 +1,133 @@
+from pyisda.curve import YieldCurve, BadDay, SpreadCurve, BasketIndex
+from pyisda.legs import FeeLeg, ContingentLeg
+from pyisda.logging import enable_logging
+import datetime
+import math
+import pandas as pd
+from yieldcurve import YC, ql_to_jp
+from quantlib.settings import Settings
+from quantlib.time.api import Date
+import numpy as np
+from db import dbconn
+from concurrent.futures import ProcessPoolExecutor, as_completed
+from itertools import zip_longest, chain
+
+def get_singlenames_quotes(indexname, date):
+    conn = dbconn('serenitasdb')
+    with conn.cursor() as c:
+        c.execute("SELECT * FROM curve_quotes(%s, %s)", vars=(indexname, date))
+        return [r for r in c]
+
+def build_curve(r, today_date, yc, start_date, step_in_date, value_date, end_dates):
+    spread_curve = 1e-4 * np.array(r['spread_curve'][1:])
+    upfront_curve = 1e-2 * np.array(r['upfront_curve'][1:])
+    recovery_curve = np.array(r['recovery_curve'][1:])
+    sc = SpreadCurve(today_date, yc, start_date, step_in_date, value_date,
+                     end_dates, spread_curve, upfront_curve, recovery_curve, True)
+    return (r['cds_ticker'], sc)
+
+def grouper(iterable, n, fillvalue=None):
+    "Collect data into fixed-length chunks or blocks"
+    # grouper('ABCDEFG', 3, 'x') --> ABC DEF Gxx
+    args = [iter(iterable)] * n
+    return zip_longest(fillvalue=fillvalue, *args)
+
+def build_curves_dist(quotes, args, workers=4):
+    ## about twice as fast as the non distributed version
+    ## non thread safe for some reason so need ProcessPool
+    with ProcessPoolExecutor(workers) as e:
+        fs = [e.submit(build_curves, *(q, args)) for q in grouper(quotes, 30)]
+    return list(chain.from_iterable([f.result() for f in as_completed(fs)]))
+
+def build_curves(quotes, args):
+    return [build_curve(q, *args) for q in quotes if q is not None]
+
+
+Settings().evaluation_date = Date(6, 2, 2017)
+yc = YC()
+jp_yc = ql_to_jp(yc)
+today_date = datetime.date(2017, 2, 6)
+step_in_date = datetime.date(2017, 2, 7)
+value_date = datetime.date(2017, 2, 9)
+start_date = datetime.date(2016, 12, 20)
+end_dates = [datetime.date(2017, 12, 20),
+             datetime.date(2018, 12, 20),
+             datetime.date(2019, 12, 20),
+             datetime.date(2020, 12, 20),
+             datetime.date(2021, 12, 20),
+             datetime.date(2023, 12, 20),
+             datetime.date(2026, 12, 20)]
+
+quotes = get_singlenames_quotes("ig27", today_date)
+args = (today_date, jp_yc, start_date, step_in_date, value_date, end_dates)
+curves = build_curves_dist(quotes, args)
+
+def all_curves_pv(curves, today_date, jp_yc, start_date, step_in_date, value_date, end_dates):
+    r = {}
+    for d in end_dates:
+        tenor = {}
+        coupon_leg = FeeLeg(start_date, d, True, 1., 1.)
+        default_leg = ContingentLeg(start_date, d, True)
+        accrued = coupon_leg.accrued(step_in_date)
+        tickers = []
+        data = []
+        for ticker, sc in curves:
+            coupon_leg_pv = coupon_leg.pv(today_date, step_in_date, value_date, jp_yc, sc, False)
+            default_leg_pv = default_leg.pv(today_date, step_in_date, value_date,
+                                            jp_yc, sc, 0.4)
+            tickers.append(ticker)
+            data.append((coupon_leg_pv-accrued, default_leg_pv))
+        r[pd.Timestamp(d)] = pd.DataFrame.from_records(data,
+                                                       index=tickers,
+                                                       columns=['duration', 'protection_pv'])
+    indexpv = r.mean().unstack()
+    return indexpv.protection_pv-indexpv.duration*0.01
+
+def stack_curves(curves):
+    dates = [d for d, _ in curves[0].inspect()['data']]
+    hazard_rates = np.empty((len(curves), len(dates)))
+    for i, sc in enumerate(curves):
+        hazard_rates[i] = np.array([h for _, h in sc.inspect()['data']])
+    return hazard_rates, dates
+
+def all_curves_pv2(curves, today_date, jp_yc, start_date, step_in_date, value_date, end_dates):
+    tickers = [t for t, _ in curves]
+    hazard_rates, end_dates = stack_curves([c for _, c in curves])
+    r = {}
+    for maturity in end_dates:
+        data = []
+        ig27 = BasketIndex(today_date, start_date, maturity, end_dates, hazard_rates)
+        data.append(ig27.pv(step_in_date, value_date, jp_yc, 0.4))
+        r[pd.Timestamp(d)] = pd.DataFrame.from_records(data,
+                                                       index=tickers,
+                                                       columns=['duration', 'protection_pv'])
+    return r
+
+def forward_hazard_rates(sc):
+    r = []
+    t = []
+    t1 = 0
+    h1 = 0
+    base_date = sc.base_date
+    for d, h in sc.inspect()['data']:
+        h2 = math.log1p(h)
+        t2 = (d - base_date).days / 365
+        r.append( (h2 * t2 - h1 * t1) / (t2 - t1) )
+        t.append(t2)
+        h1, t1 = h2, t2
+    return t, r
+
+ticker, sc = curves[0]
+sc2 = sc.tweak_curve(0.01, inplace=False)
+# class CreditIndex:
+#     def __init__(name, trade_date):
+
+#     @property
+#     def quotes():
+#         pass
+#     forward_hazard_rates()
+#     for d, q in
+#     @quotes.setter
+#     def quotes(val):
+#         self._quotes = val
+#         for d, q in self._quotes: