AT1 vs Credits Historical OAS per Rating¶

This notebook shows how to use polars-bloomberg to pull BQL history and compare option-adjusted spreads (OAS) between AT1 (CoCo) bonds and other (non-subordinated) credit, bucketed by rating. The BQL query is constructed to pull historical OAS by date, rating, and credit type (CoCo vs Credit). By controlling for credit risk via rating, we can isolate the “AT1 premium” and show that CoCo investors have historically earned additional spread even after adjusting for credit quality.

We:

Pull historical OAS for AT1 (CoCo) and senior credit across rating buckets
Chart the spread gap through time to highlight the persistent CoCo premium after rating adjustment

Out[1]:

In [2]:

Copied!





# Get historical median OAS spreads for IG, HY and AT1 bonds

from polars_bloomberg import BQuery

query = """
    let(
        #rng = range(2014-12-31, today(), frq='CM');
        #sprd = spread(dates=#rng, fill=PREV, spread_type='OAS');
        #is_coco = capital_contingent_security();
        #month = MONTH(#sprd().DATE);
        #year = YEAR(#sprd().DATE);

        #oas = median(group(#sprd(), by=[bb_composite, #is_coco, #month, #year]));
    )
    get(#oas)
    for(
        filter(
            members(['I31095 Index', 'LP01STAT Index', 'LP05STAT Index']),
            bb_composite() in ['BBB', 'BBB-', 'BB+', 'BB']
        )
    )
    preferences(dropCols=['ORIG_IDS'])
"""

with BQuery() as bq:
    df = bq.bql(query).combine()
df.tail(3)
# Get historical median OAS spreads for IG, HY and AT1 bonds

from polars_bloomberg import BQuery

query = """
    let(
        #rng = range(2014-12-31, today(), frq='CM');
        #sprd = spread(dates=#rng, fill=PREV, spread_type='OAS');
        #is_coco = capital_contingent_security();
        #month = MONTH(#sprd().DATE);
        #year = YEAR(#sprd().DATE);

        #oas = median(group(#sprd(), by=[bb_composite, #is_coco, #month, #year]));
    )
    get(#oas)
    for(
        filter(
            members(['I31095 Index', 'LP01STAT Index', 'LP05STAT Index']),
            bb_composite() in ['BBB', 'BBB-', 'BB+', 'BB']
        )
    )
    preferences(dropCols=['ORIG_IDS'])
"""

with BQuery() as bq:
    df = bq.bql(query).combine()
df.tail(3)

Out[2]:

shape: (3, 7)

ID	#oas	DATE	BB_COMPOSITE	#IS_COCO	#MONTH	#YEAR
str	f64	date	str	bool	i64	i64
"BBB:true:9.0:2023.0"	437.833167	2023-09-30	"BBB"	true	9	2023
"BBB:true:9.0:2024.0"	303.500856	2024-09-30	"BBB"	true	9	2024
"BBB:true:9.0:2025.0"	246.992669	2025-09-30	"BBB"	true	9	2025