• Contact
  • Register
    (If you're a human, don't change the following field)
    Your first name.
    (If you're a human, don't change the following field)
    Your first name.

    Enter your email above, then click "Sign Up" to join the STAC mail list and (optionally) register to access materials on the site. Click for terms.

Posted February 14, 2019 by GuoXiaoXiao

Quality test of Early Exercise

I run the R code in A2 to generate the output of Early Exercise and compare it with the reference value in Quality tool. I found the error is 0.1. Why the error is so large? 

Comments

Posted on February 14, 2019 by peterlankford

Hi, GuoXiaoXiao. The R code uses a very small number of timesteps and paths in order to allow it to run quickly on a laptop. When you ran quality tests, did you set timesteps to 252 and paths to 25,000 (the baseline problem size for STAC-A2)?

In general, please note that the purpose of the R code is to describe the algorithms in STAC-A2. It is not a complete reference implementation (for example, it does not input problem sizes from the STAC-A2 config files).

Cheers,

Peter

Posted on February 14, 2019 by GuoXiaoXiao

Hi Peter,

      I read the spec. I think for quality test of Early Exercise. It uses the single-asset, timesteps  to 100, paths to 10000 and read the model parameter from input_HESTON_1ASSET.csv.  I did the same problem sizes setting.

Thanks, GuoXiaoXiao

Posted on February 14, 2019 by peterlankford

You're right,  I forgot to check the params for that test.  So it sounds like you set them correctly.  Did you set the strike to 1? If so and you're still getting a lot of error, I'll have to punt to someone else on this list who actually understands the code, since I'm out of my depth.

Thanks,

Peter

Posted on February 14, 2019 by GuoXiaoXiao

yes, the strike is 1. The output for R code is about 0.22 and the reference code is about 0.104. 

Thanks, GuoXiaoXiao

Posted on June 5, 2019 by peterlankford

I just checked the specs, and they say the output of the SUT is compared to a Black-Scholes binomial approximation. The input parameters for the SUT are:

input_HESTON_1ASSETQUAL.csv
input_MATRIX_1ASSETQUAL.csv
input_HESTON_2ASSETQUAL.csv
input_MATRIX_2ASSETQUAL.csv

If you're asking what the input parameters are for the B-S approximation, the best place to look is in the source to the QA tool.

Thanks,

Peter

Posted on June 5, 2019 by GuoXiaoXiao

Hi Peter,

where can I find  "STAC-A2 Quality Assessment Algorithm Specifications (C code), latest revision" ? Thanks.

Posted on June 5, 2019 by peterlankford

I've dropped you a note to source out acces to that repo.

Thanks

Posted on June 5, 2019 by GuoXiaoXiao

Hi Peter,

I have the access to the repo. However, I always cannot pass the exerly exercise quality test. Where can I find a detail descirption about this test? Now I use R algorihtm as my golden. 

Posted on June 5, 2019 by peterlankford

GuoXiaoXiao, let's set up a time to talk.  Please respond to the email I sent so we can take this off the group forum.

Thanks!

Peter

The STAC-A2 Benchmark suite is the industry standard for testing technology stacks used for compute-intensive analytic workloads involved in pricing and risk management.

Read more