Translating ratings¶

When working with ratings, you pretty fast come to the point where you need to do some computations, e.g. calculating the average rating of a portfolio.

In these cases, you can't simply use human-readable ratings. You first need to translate these ratings into numerical rating scores. And when you are done with your computations and want to present your results, you have to translate rating scores back into human-readable ratings.

When translating ratings into scores and vice versa, there's a difference between long-term and short-term ratings and pyratings actually uses different translation tables.

Ratings vs. numerical scores¶

pyratings provides a number of functions that help to translate ratings into scores and vice versa. The respective API Reference holds a number of concrete examples of how to apply these functions.

• get_scores_from_ratings: Convert regular ratings into numerical rating scores.
• get_ratings_from_scores: Convert numerical rating scores into regular ratings.

For example, we can convert ratings_df into scores as follows:

In [1]:

import pandas as pd
import numpy as np

import pyratings as rtg

import pandas as pd import numpy as np import pyratings as rtg

In [2]:

ratings_df = pd.DataFrame(
    data=(
        {
            "rating_S&P": ['AAA', 'AA-', 'AA+', 'BB-', 'C'],
            "rating_Moody's": ['Aa1', 'Aa3', 'Aa2', 'Ba3', 'Ca'],
            "rating_Fitch": ['AA-', 'AA-', 'AA-', 'B+', 'C'],
        }
    ),
    index=[f'security_{i}' for i in range(5)]
)
ratings_df

ratings_df = pd.DataFrame( data=( { "rating_S&P": ['AAA', 'AA-', 'AA+', 'BB-', 'C'], "rating_Moody's": ['Aa1', 'Aa3', 'Aa2', 'Ba3', 'Ca'], "rating_Fitch": ['AA-', 'AA-', 'AA-', 'B+', 'C'], } ), index=[f'security_{i}' for i in range(5)] ) ratings_df

Out[2]:

	rating_S&P	rating_Moody's	rating_Fitch
security_0	AAA	Aa1	AA-
security_1	AA-	Aa3	AA-
security_2	AA+	Aa2	AA-
security_3	BB-	Ba3	B+
security_4	C	Ca	C

In [3]:

scores_df = rtg.get_scores_from_ratings(
    ratings=ratings_df,
    rating_provider=["SP", "Moody", "Fitch"]
)

scores_df

scores_df = rtg.get_scores_from_ratings( ratings=ratings_df, rating_provider=["SP", "Moody", "Fitch"] ) scores_df

Out[3]:

	rtg_score_rating_S&P	rtg_score_rating_Moody's	rtg_score_rating_Fitch
security_0	1	2	4
security_1	4	4	4
security_2	2	3	4
security_3	13	13	14
security_4	21	20	21

Converting the numerical scores back into human-readable ratings is as easy as:

In [4]:

rtg.get_ratings_from_scores(
    rating_scores=scores_df,
    rating_provider=["SP", "Moody", "Fitch"]
)

rtg.get_ratings_from_scores( rating_scores=scores_df, rating_provider=["SP", "Moody", "Fitch"] )

Out[4]:

	rtg_SP	rtg_Moody	rtg_Fitch
security_0	AAA	Aa1	AA-
security_1	AA-	Aa3	AA-
security_2	AA+	Aa2	AA-
security_3	BB-	Ba3	B+
security_4	C	Ca	C

Ratings vs. WARF¶

pyratings also offers the possibility to translate ratings (or even rating scores) into WARF's and vice versa. Again, take a look at the individual function's references and study the examples.

• get_warf_from_ratings: Convert regular rating(s) to numerical WARF(s).
• get_ratings_from_warf: Convert WARFs into regular ratings.

For example, we can convert ratings_df into WARFs as follows:

In [5]:

rtg.get_warf_from_ratings(ratings=ratings_df)

rtg.get_warf_from_ratings(ratings=ratings_df)

Out[5]:

	warf_rating_S&P	warf_rating_Moody's	warf_rating_Fitch
security_0	1	10	40
security_1	40	40	40
security_2	10	20	40
security_3	1766	1766	2220
security_4	9999	9998	9999

Again, when not specifying rating_provider, pyratings tries to extract the appropriate rating provider automatically.