Example of python code?

Quantlab API:s and Integration
1

Ok. I have managed to install Quantlab and connect it to my python dev. How do I call an instrument or a list of instruments (i.e. a curve)?

Can anyone give a short example.

2

Hi,
sure, I have a simple demo example that shows how to:

  1. load quantlab into python
  2. use date functions
  3. extract a quote from a list of instruments coming from a “curve”
  4. plot the quote vs maturity

Will try to attach the python example code below.

#!/usr/bin/env python
# coding: utf-8

# In[30]:


import matplotlib
import matplotlib.pyplot as plt
import numpy as np
import ql

#print(sys.version_info)

cn = "SEMTG="
s = list()
m = list()
d = ql.move_bus_days(ql.today(), -30)
qs = "Mid"
t = ql.curve(cn, d,qs, True, True).n_instruments()
r = range(0, t-1)

for n in r:
    s.append(ql.curve(cn, d, qs, True, True).instruments()[n].quote())
    m.append(ql.curve(cn, d, qs, True, True).instruments()[n].maturity())
    
    
fix, ax = plt.subplots()
ax.plot(m,s)
ax.set(xlabel='maturity', ylabel='Yield%',title='ML experiments')
ax.grid()

plt.show()


# In[24]:


names = list()
dur = list()
mat = list()
yld = list()
for n in range(0,t-1):
    names.append(ql.curve(cn, d,qs, True, True).instruments()[n].name())
    dur.append(round(ql.curve(cn, d,qs, True, True).instruments()[n].mac_dur(),1))
    yld.append(ql.curve(cn,d,qs, True, True).instruments()[n].quote())
    mat.append(ql.curve(cn, d,qs, True, True).instruments()[n].maturity())
    
print(np.array([names, yld,  dur, mat]).transpose())


# In[10]:


print(ql.curve(cn, d).get_cashflows(20))
1 Like
3

Adding to the above example.
A five parameter interest rate model referred to as Nelson-Siegel-Laguerre can be fitted to the yields using a L2 optimizer (Levenberg Marquardt).
(Note that this example needs the parameters and imports of the previous examples to work.)

r2 = ql.range_vector(1,30)
s2 = list()
d2 = ql.move_bus_days(ql.today(), -2)

for myt in r2:
    s2.append(ql.fit_l2(ql.curve(cn,d,qs, True, True), ql.ns_laguerre(), ql.std_weights.WS_EQUAL)
             .effective_rate_spot(myt)*100)
    
fix, ax = plt.subplots()
ax.plot(r2,s2)
ax.set(xlabel='time (years)', ylabel='Yield%',title='Quantlab to the people!')
ax.grid()

plt.show()

Result should look something like this when run in a Jupyter notebook:
image

2 Likes
4

That’s a nice example Robert! But it seems a bit excessive to recalibrate the curve again and again when plotting the graph. Might I suggest something like this instead:

fitted = ql.fit_l2(ql.curve(cn,d,qs, True, True), ql.ns_laguerre(), ql.std_weights.WS_EQUAL)
for myt in r2:
    s2.append(fitted.effective_rate_spot(myt)*100)
5

Well spotted and of course you are right. I had an ambition to plot the history of the fit but then opted for one date only. Then one fit should be enough. :grinning:

1 Like