Brazil EMBI directed graphs

Monthly — is the current spread priced?

Current EMBI	175 bp (2026-09-02)
Macro fair value	170 bp (month-end 2026-08-31)
Rich / cheap	fair
Gap (current − fair)	+5 bp (0.435σ)
Month-end residual	+4 bp
Residual σ	11 bp
In-sample R²	0.849

4 monthly print(s) may not yet have been tradable at the last EMBI date.

Parent	Δ last month	Contrib bp	Typical print lag (d)
dxy (t)	-0.956	-1.9	0
tot_gap (t)	0.000	+1.1	45
us_hy_oas (t)	-0.220	-0.3	0
tot (t)	0.000	+0.0	45
debt_gap (t)	0.000	+0.0	35
current_account (t)	0.000	+0.0	25

Has EMBI had time to trade the prints?

Input	Est. release	Traded after print?	Days since
us_hy_oas	2026-08-31	yes	2
tot	2026-10-15	pending	-43
tot_gap	2026-10-15	pending	-43
dxy	2026-08-31	yes	2
debt_gap	2026-10-05	pending	-33
current_account	2026-09-25	pending	-23

Monthly DAG

Narrative edges (strongest first)

From		To	Lag	β (z)	Child R²
public_debt_gdp	→	debt_gap	t	+1.000	-0.901
fx	→	reer	t	-0.948	-7.642
nominal_deficit	→	primary_balance	t	+0.930	0.888
dxy	→	fx	t	+0.916	-803.235
sp500	→	equity	t	+0.898	-15.794
selic	→	policy_fed_gap	t	+0.875	-7.905
dxy	→	reer	t	-0.839	-7.642
fx	→	equity	t	+0.814	-15.794
us_hy_oas	→	equity	t	-0.708	-15.794
oil	→	primary_balance	t	-0.704	0.888
fed_funds	→	equity	t	+0.699	-15.794
reer	→	equity	t	-0.677	-15.794
ipca	→	policy_fed_gap	t	+0.647	-7.905
oil	→	embi_bra	t	-0.621	-0.933
debt_gap	→	policy_fed_gap	t	-0.607	-7.905
equity	→	policy_fed_gap	t	-0.603	-7.905
current_account	→	equity	t	+0.587	-15.794
reer_gap	→	embi_bra	t	-0.579	-0.933
copper	→	tot	t	+0.574	-30061.986
primary_balance	→	debt_gap	t	+0.572	-0.901
primary_balance	→	public_debt_gdp	t	+0.572	-0.134
tot_gap	→	equity	t	+0.567	-15.794
us_hy_oas	→	fx	t	-0.564	-803.235
nominal_deficit	→	public_debt_gdp	t	+0.561	-0.134
current_account	→	fx	t	+0.535	-803.235
copper	→	embi_bra	t	-0.514	-0.933
copper	→	equity	t	+0.512	-15.794
oil	→	tot	t	+0.505	-30061.986
ust_short	→	selic	t	+0.505	-71.186
current_account	→	ext_debt_exports	t	-0.496	0.241
tot	→	embi_bra	t	-0.488	-0.933
reer	→	embi_bra	t	-0.484	-0.933
copper	→	exports_growth	t	+0.483	-1.010
tot_gap	→	primary_balance	t	+0.481	0.888
tot_gap	→	current_account	t	+0.480	0.324
ipca	→	selic	t	+0.480	-71.186
industrial_prod	→	primary_balance	t	-0.472	0.888
primary_balance	→	selic	t	-0.467	-71.186
global_sentiment	→	equity	t	+0.465	-15.794
us_hy_oas	→	reer	t	+0.460	-7.642
oil	→	reer	t	+0.454	-7.642
fx	→	policy_fed_gap	t	-0.447	-7.905
tot_gap	→	fx	t	+0.440	-803.235
reer	→	policy_fed_gap	t	+0.436	-7.905
tot_gap	→	reer	t	-0.428	-7.642
us_hy_oas	→	reer_gap	t	-0.417	-19.889
sp500	→	policy_fed_gap	t	-0.416	-7.905
reer	→	reer_gap	t	+0.416	-19.889
dxy	→	embi_bra	t	+0.411	-0.933
tot	→	exports_growth	t	+0.408	-1.010
tot	→	reserves	t	-0.404	-103.401
us_hy_oas	→	policy_fed_gap	t	+0.401	-7.905
oil	→	fx	t	-0.392	-803.235
copper	→	tot_gap	t	+0.384	-24.901
reer_gap	→	selic	t	+0.372	-71.186
primary_balance	→	embi_bra	t	+0.367	-0.933
reserves_growth	→	embi_bra	t	-0.357	-0.933
fx	→	embi_bra	t	+0.351	-0.933
reserves	→	reserves_growth	t	-0.331	-1.327
dxy	→	policy_fed_gap	t	-0.321	-7.905

ΔEMBI engine (holdout score)

Ridge on ΔEMBI: 1 risk (HY or VIX or S&P) + 1 rates/USD + 1 commodity + 1 local + 1 stock. Ranked by holdout R², not quintile lift. Best: theory|lags=[1] · parents: us_hy_oas, dxy, oil, equity, reer_gap · 182 train / 19 test through 2026-09-30.

From		To	Lag	β (z)	Child R²
us_hy_oas	→	embi_bra	t−1	-0.116	-0.003
equity	→	embi_bra	t−1	-0.080	-0.003
dxy	→	embi_bra	t−1	+0.042	-0.003
reer_gap	→	embi_bra	t−1	+0.036	-0.003
oil	→	embi_bra	t−1	-0.000	-0.003

Method	Lags	OOS R²	σ-quintile	Lift vs naive	Naive middle	±1 bin	3-class	Sign	RMSE Δ	Edges	Parents
theory|lags=[1] ← best	[1]	-0.726	47.4%	-0.105	57.9%	89.5%	47.4%	57.9%	0.107	24	5
pc|lags=[1]	[1]	-0.726	47.4%	-0.105	57.9%	89.5%	47.4%	57.9%	0.107	50	5
pc_strict|lags=[1]	[1]	-0.726	47.4%	-0.105	57.9%	89.5%	47.4%	57.9%	0.107	44	5
tan|lags=[1]	[1]	-0.726	47.4%	-0.105	57.9%	89.5%	47.4%	57.9%	0.107	11	5
hybrid|lags=[1]	[1]	-0.726	47.4%	-0.105	57.9%	89.5%	47.4%	57.9%	0.107	52	5
theory|lags=[1, 2]	[1, 2]	-0.777	47.4%	-0.105	57.9%	89.5%	47.4%	57.9%	0.109	24	10
pc|lags=[1, 2]	[1, 2]	-0.777	47.4%	-0.105	57.9%	89.5%	47.4%	57.9%	0.109	50	10
pc_strict|lags=[1, 2]	[1, 2]	-0.777	47.4%	-0.105	57.9%	89.5%	47.4%	57.9%	0.109	44	10
tan|lags=[1, 2]	[1, 2]	-0.777	47.4%	-0.105	57.9%	89.5%	47.4%	57.9%	0.109	11	10
hybrid|lags=[1, 2]	[1, 2]	-0.777	47.4%	-0.105	57.9%	89.5%	47.4%	57.9%	0.109	52	10
theory|lags=[1, 3]	[1, 3]	-1.064	42.1%	-0.158	57.9%	89.5%	42.1%	63.2%	0.117	24	10
pc|lags=[1, 3]	[1, 3]	-1.064	42.1%	-0.158	57.9%	89.5%	42.1%	63.2%	0.117	50	10
pc_strict|lags=[1, 3]	[1, 3]	-1.064	42.1%	-0.158	57.9%	89.5%	42.1%	63.2%	0.117	44	10
tan|lags=[1, 3]	[1, 3]	-1.064	42.1%	-0.158	57.9%	89.5%	42.1%	63.2%	0.117	11	10