I have to build an operational risk model and to be compliant with Basell II and III regulations I thought of using AMA (Advanced measurement approach) or COM (change of measurement). We have no historical data in the sense that what my company does is spreading around a questionnaire where the heads of the departments have to indicate for each risk category the amount of money (in their opinion) was wasted for that category and how many times that loss occurred in 1 year. They fill out a table in which there is MIN, MAX, MEDIAN..in Euros.

The goal is to estimate the total losses so I want in the end a number in Euros, and in addition also the distribution of this loss.

Do you agree with using the approaches AMA or COM? I noticed they want in input recordings of the past losses per each category but I do not have such data.

Have you got any reference where it is shown the code to implement those?

Many thanks everybody in advance

  • $\begingroup$ anecdotally, a common theme I have seen with op risk models is that they underestimate or fail to analyse black swan events (managerial risks) (aml risks) that lead to losses that far exceed any of the other op risks that you might spend a large amount of resources calculating. just FYI $\endgroup$
    – Attack68
    Sep 24, 2020 at 13:21
  • 1
    $\begingroup$ 1 your employer should (definitely) subscribe to the ORX events database and (possibly also) to "IBM FIRST Risk Case StudiesRisk event database". 2 many large organizations are abandoning AMA (even if they have had it running for a few years) and switching to standardized model. 3 I found this COM paper helpful onlinelibrary.wiley.com/doi/abs/10.1111/… $\endgroup$ Sep 24, 2020 at 13:29
  • $\begingroup$ @DimitriVulis thanks for your feedback. Actually I have read in many websites that AMA model is better than Basic Indicator Approach and the Standardised Approach. So would you specify why is that better? regards $\endgroup$
    – Luigi87
    Sep 24, 2020 at 14:29
  • $\begingroup$ Sorry I don't want to name anyone concrete, but multiple large firms who have run AMA for some time have now decided that it's too much trouble/expense for them, not worth the savings over Standardized, and that Standardized bis.org/basel_framework/chapter/OPE/… is the right balance. So they scrap their AMA (or their plans for AMA) and aim for Standardized. (Basic would be too expensive for most firms, so using Standardized instead of Basic is worth the work.) $\endgroup$ Sep 24, 2020 at 14:39
  • $\begingroup$ Did you find anything concrete on this as to how to define a concrete number by AMA. $\endgroup$ Sep 6, 2022 at 14:08

1 Answer 1


I googled and found https://github.com/TommasoBelluzzo/BaselTools . It says:


The tool can be run by executing the BaselOP.m script. The underlying calculations are based on the SMA model defined within the BCBS 356. The application offers the opportunity to compare the SMA capital requirements with those produced by the obsolete Basel II approaches defined in BCBS 128: the Basic Indicator Approach, the Standardised Approach and the Alternative Standardised Approach.

It looks like readable matlab code implementing the SMA. I haven't tried running it, but it looks like it would be helpful to someone tasked with implementing SMA.

  • $\begingroup$ I had a look and it is a matlab gui..it can be useful but I need something to be done in R or python, anyway I check it to see if I can traslate it to other languages. Also I am not sure it can work with the input I have. Thanks anyway $\endgroup$
    – Luigi87
    Sep 25, 2020 at 8:26
  • 1
    $\begingroup$ I took a closer look at Tommaso Belluzzo's contributuion. Some of the underlying calculations for SMA (and also for SA-CCR) are in baseltools.java, which is found inside baseltools.jar (jar files are just zip files). $\endgroup$ Sep 25, 2020 at 12:22
  • 1
    $\begingroup$ ORX: Why operational risk measurement remains important despite SMA $\endgroup$ Sep 30, 2020 at 9:08

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.