Constructing a Bond Discount Curve with Rateslib’s Global Solver

# Bootstrap a Bond Curve in Rateslib


# Bootstrap a Bond Curve in Rateslib












Is it possible to get a Zero coupon curve from a set of FixedRateBond in rateslib?

I tried looking in the documentation but didn't find a way of doing it.

There's something called CurveSolver but it seems like you need to supply a Curve to it.

From a set of FixedRateBond and their prices or YTM observed in the market I would like to derive a zero coupon curve.

## Answer by Attack68 (score 1, accepted)

https://quant.stackexchange.com/a/80656

"Bootstrapping" is a name of one specific method to construct a curve from a given set of instruments. Bootstrapping works by sequentially solving values on the curve usually in ascending order of maturity of the instruments. To bootstrap a curve a few conditions must be met, which is usually why some software allow a curve to be created directly from bond instruments, becuase that software automatically creates an inflexible but suitable curve from the maturity of the instruments and then sequentially computes the necessary discount factors.

Rateslib does not bootstrap. It uses an alternate method to construct curves which is much more flexible. That being a "global optimiser".

Any curve that can be bootstrapped can be solved by a global optimiser, but it is not true that any curve solved by a global optimiser can be bootstrapped becuase bootstrapped curves require those aformentioned conditions. Thus, rateslib can indirectly bootstrap your curve, by solving it globally.

In order to solve curves in rateslib you need to do 2 things:

- Define the Instruments whose prices will instruct the Curve.

- Define the structure of the Curve which will be solved to suit those Instrument prices.

Since you just want a basic 'bootstrap' curve, the default structure you are looking for is to put nodes at the maturity of each bond, and use log linear interpolation.

```
from rateslib import *  # Python 3.12, Rateslib 1.4.0

bonds = [
    FixedRateBond(dt(1999, 9, 26), "3y", spec="us_gb", curves="bond", fixed_rate=2.5),
    FixedRateBond(dt(1998, 3, 15), "7y", spec="us_gb", curves="bond", fixed_rate=2.75),
    FixedRateBond(dt(1999, 5, 22), "10y", spec="us_gb", curves="bond", fixed_rate=3.0),
]

curve = Curve(
    nodes={
        dt(2000, 1, 1): 1.0,  # <-- TODAY
        dt(2002, 9, 26): 1.0, # <-- Bond Maturity
        dt(2005, 3, 15): 1.0, # <-- Bond Maturity
        dt(2009, 5, 22): 1.0, # <-- Bond Maturity
    },
    interpolation="log_linear",
    id="bond"
)

solver = Solver(
    curves=[curve],
    instruments=bonds,
    s=[100, 100, 100],  # <-- Clean price of each bond
)
```

The `Curve` constructed contains discount factors. You can extract a discount factor for any date using Pythons `__getitem__`, i.e.

```
curve[dt(2003, 4, 5)]
# 0.919685
```

Or you can plot overnight forward rates from the curve:

```
curve.plot("1b")
```

To calculate zero rates and plot them, you have to loop through all the dates you want and perform the calculation for a zero rate from a discount factor manually.