Bootstrapping Separate OIS Curves for Different Valuation Dates
Summary
The document discusses a QuantLib bootstrapping error caused by combining Japanese yen overnight indexed swap market data from two trade dates in one curve. Instruments from different dates can share maturity pillars, triggering a duplicate-pillar error, but the deeper issue is that a term structure represents market conditions as of one valuation date.
The guidance is to build an independent OIS curve for each date using that date’s instrument quotes and evaluation date. A curve calibrated to one day’s market data is appropriate for discounting cash flows as of that day; adding the next day’s observations does not create a meaningful single-date curve. The example uses data from consecutive February 2017 dates. It gives the conceptual fix but does not provide a corrected implementation or discuss curve construction choices beyond separating the dates.
Key ideas
- A bootstrapped curve represents market conditions as of a specific valuation date.
- Quotes from different trade dates should not be combined in a single OIS curve.
- Build a separate curve using each date’s instruments and set the evaluation date accordingly.
- Duplicate maturity pillars may expose the problem, but the underlying issue is mixing valuation dates.
Tags
Full text
# Bootstrapping OIS Curve with data from different days data
# Bootstrapping OIS Curve with data from different days data
I have the following problem bootstrapping the JPY OIS Curve. The bootstrapping itself works when havin one set of data, e.g. for the date 2017-02-09. I have all my instruments and as said bootstrapping and receiving the OIS curve. If i extend the dataset to a second date 2017-02-10 I receive all the data into my bootstrapper which makes no sense with the error message:
> more than one instrument with pillar February 10th, 2017
which results to the same maturities due to data from different dates as explained above. I have somehow to loop over the fdates but I have no idea how this could work in that context.Help would be appreciated.
```
import QuantLib as ql
import pandas as pd
import matplotlib.pyplot as plt
def Convert(Period):
unit =[]
if Period[-1:] == 'D':
unit = ql.Days
elif Period[-1:] == 'M':
unit = ql.Months
elif Period[-1:] == 'W':
unit = ql.Weeks
elif Period[-1:] == 'Y':
unit = ql.Years
period_object = ql.Period(int(Period[:-1]), unit)
return period_object
date = ql.Date(9, ql.February, 2017)
ql.Settings.instance().evaluationDate = date
data = pd.read_csv('C:/Data_JPY_Playing.csv').fillna('')
data_selected = data[['fdate', 'ptype' ,'maturity','fixing','values']].to_records(index=False)
Rate_Helper_Full_Disc = []
for fdate, ptype, maturity, fixing, values in data_selected:
if row['ptype'] == 'Deposit':
helper_disc = ql.DepositRateHelper(ql.QuoteHandle(ql.SimpleQuote(row['values']/100)),
Convert(row['maturity']),
int(row['fixing']),
ql.Japan(),
ql.ModifiedFollowing,
False,
ql.Actual365Fixed())
Rate_Helper_Full_Disc.append(helper_disc)
disc_curve = ql.PiecewiseCubicZero(date, Rate_Helper_Full_Disc, ql.Actual365Fixed())
disc_curve.enableExtrapolation()
```
## Answer by Antoine Conze (score 3, accepted)
https://quant.stackexchange.com/a/38089
A curve is used to do calculations (e.g. discounting of cash flows) as of a given trade date. Bootstrapping a single curve for two different trade dates does not make sense. With the first set of data you should bootstrap an OIS curve for the 2017-02-09 trade date, with the second set of data you should bootstrap an OIS curve for the 2017-02-10 trade date.Shown in full with attribution under the source's licence. Licence: CC BY-SA 4.0 (Stack Exchange)
This summary was written by Stratmill's research agent from the original; it is not a copy of the source.