Any QuantLib experts here?

earth_imperator · Sep 7, 2023

M.W. said:
You don't need QuantLib for that. In fact, you can easily simulate stock prices with an underlying GBM engine within excel.

https://investexcel.net/geometric-brownian-motion-excel/
More...

But as said I need to do it w/ QuantLib. GBM is just a minor part.
Excel is the biggest BS for such a task, b/c slow. I need to generate multi-millions of such GBM generated prices for a Monte Carlo simulation.

earth_imperator · Sep 7, 2023

Folks, just admit, none of you here is any expert in QuantLib as it looks like...

Overnight · Sep 7, 2023

earth_imperator said:
But as said I need to do it w/ QuantLib. GBM is just a minor part.
Excel is the biggest BS for such a task, b/c slow. I need to generate multi-millions of such GBM generated prices for a Monte Carlo simulation.
More...

Why bother? You'll find the math will come out all wrong and you'll start a new thread about it, whining to anyone who will listen that you have a better way.

earth_imperator · Sep 7, 2023

Overnight said:
Why bother? You'll find the math will come out all wrong and you'll start a new thread about it, whining to anyone who will listen that you have a better way.
More...

As posted some weeks ago here, I'm doing a research. I now just wanted to include also QuantLib into my research.
It indeed could be that the GBM in QuantLib is as buggy as most of the GBMs I so far have tested.

earth_imperator · Sep 7, 2023

Just an update:
Using the above said QuantLib GBM code I get very very weired results.
IMO the GBM in QuantLib can never ever be correct.
WIll continue my tests tomorrow.
Will also post the code here for verification & reproduction of the results by others.

Overnight · Sep 7, 2023

earth_imperator said:
...
IMO the GBM in QuantLib can never ever be correct...
More...

*Ahem* Who called it? Moi?

Naaaaaah

M.W. · Sep 7, 2023

Exactly, he could not even articulate why it must be in QuantLib. Probably some homework, lol. Not worth the effort.

Overnight said:
Why bother? You'll find the math will come out all wrong and you'll start a new thread about it, whining to anyone who will listen that you have a better way.
More...

ph1l · Sep 7, 2023

earth_imperator said:
Just an update:
Using the above said QuantLib GBM code I get very very weired results.
IMO the GBM in QuantLib can never ever be correct.
WIll continue my tests tomorrow.
Will also post the code here for verification & reproduction of the results by others.
More...

Another example for you to "prove" incorrect.
Bing Chat said:
Sure! There are many resources available online that provide examples of how to use QuantLib. For instance, Gouthaman Balaraman has a collection of tutorials on his website that cover various topics, including modeling fixed rate bonds, bootstrapping yield curves, and valuing interest rate swaps using QuantLib and Python². You can also find documentation and examples on the official QuantLib website¹.

Here is an example of how to create a Geometric Brownian Motion (GBM) model using QuantLib and Python:
Code:
```python
from QuantLib import *

# Set up the model parameters
riskFreeRate = 0.05
dividendYield = 0.02
volatility = 0.2
initialValue = 100

# Set up the time grid
maturity = 1.0
dt = 0.01
timeSteps = int(maturity / dt)

# Set up the random number generator
seed = 42
rng = GaussianRandomSequenceGenerator(UniformRandomSequenceGenerator(timeSteps, UniformRandomGenerator(seed)))
seq = GaussianPathGenerator(GBMProcess(QuoteHandle(SimpleQuote(initialValue)), YieldTermStructureHandle(FlatForward(0, TARGET(), riskFreeRate, Actual360())), YieldTermStructureHandle(FlatForward(0, TARGET(), dividendYield, Actual360())), BlackVolTermStructureHandle(BlackConstantVol(0, TARGET(), volatility, Actual360()))), maturity, timeSteps, rng)

# Generate the path
path = seq.next().value()

# Print the path values
for i in range(len(path)):
    print(f"Time: {i * dt:.2f}, Value: {path[i]:.2f}")
```
This code sets up a GBM model with specified parameters for the risk-free rate, dividend yield, volatility, and initial value of the asset. It then uses QuantLib's `GaussianPathGenerator` class to generate a sample path of the asset price over time. The resulting path is printed to the console.

I hope this example helps you get started with using QuantLib to create GBM models! Let me know if you have any further questions or if there's anything else I can help with.

Source: Conversation with Bing, 9/7/2023
(1) QuantLib Python Tutorials With Examples - G B. http://gouthamanbalaraman.com/blog/quantlib-python-tutorials-with-examples.html.
(2) QuantLib Documentation. https://www.quantlib.org/docs.shtml.
(3) Modeling assets with QuantLib | modelx. https://modelx.io/blog/2022/02/13/modeling-assets-with-quantlib/.
(4) Writing Your First QuantLib Program in C++. https://benjaminwhiteside.com/2018/12/07/writing-your-first-quantlib-program-in-c/.
(5) QuantLib Installation in Visual C++. https://www.quantlib.org/install/vc10.shtml.
(6) Introduction to Selected Classes of the QuantLib Library I. https://www.quantlib.org/slides/dima-ql-intro-1.pdf.
More...

earth_imperator · Sep 8, 2023

@ph1l, thx, can you or your BingChat perform the said test here?
The devil is in the detail: timeSteps (in above link it's called N) must be as small as possible to see the error, ie. 1 to about 25 or so.
In QuantLib the practical minimum possible seems to be 2 since the first of the generated values is always the passed initial price.

Ie. the challenge is: Does your GBM algorithm pass the "68–95–99.7 rule test"?
They are 3 tests, but the first (ie. the 68.27% test) just suffices to pass.

earth_imperator · Sep 8, 2023

M.W. said:
Exactly, he could not even articulate why it must be in QuantLib. Probably some homework, lol. Not worth the effort.
More...

BS, just read again.