I've been meaning to port over a bit of my R code to Python for quite some time. Finally I mustered up the strength to do so and thought I'd share the first bit with you guys as this comes in handy from time to time.

What the code does is download select interest rates (H15) from the fed then bootstraps and interpolates the zero curve with the data. My primary use for this is selecting the appropriate discount factor for options pricing given a current date and tenor. There are plenty of other uses for it though.

code:

#!/usr/bin/env python

import numpy as np  
from scipy import interpolate  
from pandas import *  
from datetime import *

class zeroCurve:

    def __init__(self):  
        self.zero_curve = None  
        self.__h15Url = 'http://www.federalreserve.gov/datadownload/Output.aspx?rel=H15&series=bf17364827e38702b42a58cf8eaa3f78&lastObs=&from=&to=&filetype=csv&label=include&layout=seriescolumn&type=package'

        self.__tenors = [1, 3, 6, 12, 24, 36, 60, 84, 120, 240, 360]  
        self.__cash_names = ["1m", "3m", "6m", "12m"]  
        self.__cash_tenors = [1, 3, 6, 12]  
        self.__bond_names = ["24m", "36m", "60m", "84m", "120m", "240m", "360m"]  
        self.__bond_tenors = [24, 36, 60, 84, 120, 240, 360]  
        self.__zero_times = map(lambda x: x/12.0, [1, 3]+[i for i in xrange(6, 361, 6)])  
        self.__start_date = datetime(2002, 1, 1)

    def load_curve(self):  
        df = read_csv(self.__h15Url, skiprows=70)  
        df.columns = ['date'] + [str(tenor)+"m" for tenor in self.__tenors]  
        df['date'] = map(lambda x: datetime.strptime(x, '%Y-%m-%d'), df['date'])  
        df = df.set_index("date")  
        filter_keep = list(df['1m']!='ND')  
        df = df[filter_keep][self.__start_date:]  
        t360_nD = list(df["360m"]=='ND')  
        df['360m'][t360_nD] = df['240m'][t360_nD]  
        df = df.astype(float)/100  
        return df

    def get_zeros(self, cash_names, bond_names, cash_rates, bond_rates, cash_tenors, bond_tenors):  
        df_all = 1/(cash_rates*cash_tenors/12+1)  
        md = np.append(df_all[['6m', '12m']].values, np.arange(18,361,6))  
        dv = md.cumsum() / 2  
        fe_bonds = np.array([2*(1-md[i-1])/md[0:i].sum() for i in xrange(1,3)])  
        bond_times = np.arange(6, 361, 6)  
        by = interpolate.interp1d(np.append([6,12], bond_tenors),np.append(fe_bonds, bond_rates), kind=1)(bond_times)  
        for i in xrange(2, len(by)):  
            md[i] = (1 - dv[i-1] * by[i])/(1 + by[i] * dv[i-1])  
            dv[i] = md[0:(i+1)].sum() / 2  
        all_times = np.append(cash_tenors,bond_times[2:])  
        all_factr = np.append(df_all.values, md[2:])  
        return -12*np.log(all_factr)/all_times

    def strip_all(self, df):  
        zeros = map(lambda x: (x[0], self.get_zeros(self.__cash_names, self.__bond_names, x[1][self.__cash_names], x[1][self.__bond_names], self.__cash_tenors, self.__bond_tenors)), df.iterrows())  
        df = DataFrame.from_items(zeros, orient="index", columns=["USD"+str(j) for j in self.__cash_tenors+[i for i in xrange(18,361,6)]])  
        return df

    def zero_data(self):  
        df = self.load_curve()  
        self.zero_curve = self.strip_all(df)

    def zero_rate(self, asof, tenor):  
        zero_rates = self.zero_curve[self.zero_curve.index < asof][-1:]  
        return float(interpolate.interp1d(x=np.array(self.__zero_times), y=zero_rates.values[0], kind="linear")(tenor))  

zeroCurve.py

output example:

In [2]: zC = zeroCurve()

In [3]: zC.zero_data()

In [4]: zC.zero_rate(datetime(2014,1,27), 21.0/252.0)  
Out[4]: 9.999958333595127e-05  

If anyone would like to move it into quantopian's platform or speed my code up a bit :-), that would be very nice.