Choosing Quantitative Libraries for Fixed-Income Risk Analytics
Summary
The discussion concerns software choices for building fixed-income portfolio risk analytics when QuantLib and its Java interface are unavailable. The questioner reports problems with JQuantLib, including a bond-yield calculation error, and asks for pure Java alternatives. This points to a practical selection issue: language and deployment constraints must be weighed against coverage and reliability for bond analytics.
Responses identify OpenGamma Strata as a Java market-risk library derived from the OG-Platform codebase, and describe it as designed for use as a standalone library with examples and asset-class coverage. Another answer suggests TensorFlow Quant Finance for distributed computation and automatic differentiation, while QSTK is mentioned as a Python toolkit. These are suggestions rather than a systematic comparison: the discussion gives no benchmark results, validation study, or detailed feature-by-feature assessment, and one contributor discloses an affiliation with Strata's developer.
Key ideas
- The choice of a quantitative library can be constrained by permitted languages and deployment requirements.
- The questioner reports a bond-yield calculation issue with JQuantLib.
- OpenGamma Strata is presented as a standalone Java market-risk library.
- TensorFlow Quant Finance is suggested for distributed computation and automatic differentiation.
- The replies offer recommendations but no comparative benchmark or independent validation.
Tags
Full text
# What is the best alternative of Quantlib library # What is the best alternative of Quantlib library We need to build a Fixed Income Portfolio Risk Analytics solution. Somehow due to administrative reason we can't use Quantlib which is written in C++, even call it through SWIG via JNI. We have tried Jquantlib, but it seems not 100% replica of original Quantlib which is written in C++ have bugs( for e.g. root not bracketed error in bond yield calculation). So right now we can see two options opengamma and Maygard (the google archive) which is written in Pure Java. Can any experienced users share their views on this two library or if they know any better pure Java-based libary alternative. ## Answer by JodaStephen (score 10) https://quant.stackexchange.com/a/22458 The Strata project is the new pure Java market risk quant library from OpenGamma. For more information, see the documentation and GitHub. It is Apache v2 licensed. Strata takes the experience of the OG-Platform codebase referenced in the question and turns it into a library - no need for databases, servers or similar. Ease of use is a big focus and there are examples to allow easy evaluation. See this link for asset class coverage. Disclaimer: I work for OpenGamma, who develop Strata. ## Answer by lehalle (score 6) https://quant.stackexchange.com/a/50737 I did not tested it by now, but Google released a library similar to quantlib written in TensorFlow (tf-quant-finance). It may be worthwhile to test it (and to post here your views on it), because once you are in TF, you can - easily distribute your computations over a grid of computers (including GCP or AWS) - if you are a machine learning enthousiastic: use it to backprop a gradient descent over any parameter - and you should probably be able to use TF automatic differentiation features (see Mini-symposium on automatic differentiation and its applications in the financial industry) ## Answer by Mike El Jackson (score 4) https://quant.stackexchange.com/a/17769 QSTK is nice and open source , it is the QuantSciTookKit and it has some good functionality if you are interested in python programming. Here is the GitHub repo.
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.