Better or worse ETFs

library(tidyverse)
library(lubridate)
library(scales)
library(ggrepel)
library(tidyquant)
library(jsonlite)
theme_set(theme_minimal())
invisible(Sys.setlocale("LC_TIME", "en_US.UTF-8"))

library(knitr)
library(kableExtra)

mbb <- tq_get("MBB", from = as.Date("1900-01-01")) 
agg <- tq_get("AGG", from = as.Date("1900-01-01")) 
rem <- tq_get("REM", from = as.Date("1900-01-01")) 
usrt <- tq_get("USRT", from = as.Date("1900-01-01")) 
qclr <- tq_get("QCLR", from = as.Date("1900-01-01")) 
qqq <- tq_get("QQQ", from = as.Date("1900-01-01")) 
qqqm <- tq_get("QQQM", from = as.Date("1900-01-01")) 
xclr <- tq_get("XCLR", from = as.Date("1900-01-01")) 
roro <- tq_get("RORO", from = as.Date("1900-01-01")) 
jojo <- tq_get("JOJO", from = as.Date("1900-01-01")) 
tlt <- tq_get("TLT", from = as.Date("1900-01-01")) 
spy <- tq_get("SPY", from = as.Date("1900-01-01")) 
stnc <- tq_get("STNC", from = as.Date("1900-01-01")) 
rpar <- tq_get("RPAR", from = as.Date("1900-01-01")) 
boxx <- tq_get("BOXX", from = as.Date("1900-01-01")) 
sdhy <- tq_get("SDHY", from = as.Date("1900-01-01")) 
sgov <- tq_get("SGOV", from = as.Date("1900-01-01")) 
bil <- tq_get("BIL", from = as.Date("1900-01-01")) 
atacx <- tq_get("ATACX", from = as.Date("1900-01-01")) 

etfs <- bind_rows(mbb, agg, rem, usrt, qclr, qqq, qqqm, xclr, roro, jojo, tlt, spy, stnc, rpar, boxx, sdhy, sgov, bil, atacx)

saveRDS(etfs, "etfs/etfs.RDS")

etfs <- readRDS("etfs/etfs.RDS")

rets <- etfs |> 
  group_by(symbol) |> 
  transmute(symbol, date, r = adjusted/lag(adjusted)-1) |> 
  drop_na() |> 
  ungroup()

SPY: The SPDR S&P 500 ETF Trust (SPY) tracks the S&P 500 Index, which includes 500 of the largest U.S. companies by market capitalization. It offers broad exposure to the U.S. stock market, covering various sectors and industries, and is widely used as a benchmark for U.S. equity performance.
TLT: The iShares 20+ Year Treasury Bond ETF (TLT) tracks the ICE U.S. Treasury 20+ Year Bond Index, which includes long-term U.S. Treasury bonds with maturities of 20 years or more. It offers exposure to long-duration U.S. government bonds, which can be sensitive to interest rate changes and economic conditions.
SGOV: The iShares 0-3 Month Treasury Bond ETF (SGOV) tracks the ICE U.S. Treasury Short Bond Index, which includes short-term U.S. Treasury bills with maturities of 0-3 months. It offers exposure to short-duration U.S. government bonds, providing liquidity and stability to a portfolio.
MBB: The iShares MBS ETF (MBB) tracks the Bloomberg U.S. MBS Index, investing in U.S. agency mortgage-backed securities (MBS) issued by Fannie Mae, Freddie Mac, and Ginnie Mae. It provides exposure to high-quality MBS with low credit risk but is sensitive to interest rate changes. The ETF aims to offer stable income with moderate duration risk.
AGG: The iShares Core U.S. Aggregate Bond ETF (AGG) tracks the Bloomberg Barclays U.S. Aggregate Bond Index, which includes U.S. investment-grade bonds. It provides broad exposure to the U.S. bond market, including government, corporate, and mortgage-backed securities.
REM: The iShares Mortgage Real Estate ETF (REM) tracks the FTSE NAREIT All Mortgage Capped Index, which includes U.S. real estate investment trusts (REITs) that invest in residential and commercial mortgages. It offers exposure to the mortgage real estate sector, which can be sensitive to interest rate changes and economic conditions.
USRT: The iShares Core U.S. REIT ETF (USRT) tracks the FTSE NAREIT Equity REITs Index, which includes U.S. real estate investment trusts (REITs) that own and operate income-producing real estate. It provides exposure to the U.S. real estate market, offering potential diversification and income.
QCLR: The ClearBridge All Cap Growth ETF (QCLR) seeks to provide long-term capital appreciation by investing in U.S. large-cap growth stocks. It focuses on companies with strong growth potential and competitive advantages, aiming to outperform the broader market.
QQQ: The Invesco QQQ Trust (QQQ) tracks the Nasdaq-100 Index, which includes 100 of the largest non-financial companies listed on the Nasdaq stock exchange. It offers exposure to technology, consumer discretionary, and communication services sectors, with a focus on growth-oriented companies.
QQQM: The Invesco NASDAQ 100 ETF (QQQM) tracks the Nasdaq-100 Index, providing exposure to large-cap U.S. technology and growth stocks. It aims to replicate the performance of the index and offers diversification within the technology and growth sectors.
XCLR: The Invesco S&P 500 Equal Weight Communication Services ETF (XCLR) tracks the S&P 500 Equal Weight Communication Services Index, which includes communication services companies from the S&P 500 Index. It offers equal-weight exposure to communication services stocks, providing diversification within the sector.
RORO: The ATAC Rotation ETF (RORO) aims to provide long-term capital appreciation by investing in U.S. equity sectors with positive momentum. It uses a proprietary model to rotate between sectors based on market conditions, seeking to outperform the broader market.
JOJO: The ATAC Credit Rotation ETF (JOJO) seeks to provide long-term capital appreciation by investing in U.S. corporate bonds with positive momentum. It uses a proprietary model to rotate between credit sectors based on market conditions, aiming to outperform traditional fixed income benchmarks.
STNC: The iShares MSCI World Small Cap ETF (STNC) tracks the MSCI World Small Cap Index, which includes small-cap companies from developed markets around the world. It offers exposure to smaller companies with growth potential and diversification benefits, complementing large-cap investments.
RPAR: The RPAR Risk Parity ETF (RPAR) aims to provide risk-managed exposure to global asset classes, including U.S. stocks, bonds, gold, and real estate. It uses a risk parity strategy to balance risk across different asset classes, seeking to deliver consistent returns with lower volatility.
BOXX: The iShares iBoxx $ Investment Grade Corporate Bond ETF (BOXX) tracks the Markit iBoxx USD Liquid Investment Grade Index, which includes U.S. dollar-denominated investment-grade corporate bonds. It offers exposure to high-quality corporate bonds issued by U.S. companies, providing income and diversification benefits.
SDHY: The iShares 0-5 Year High Yield Corporate Bond ETF (SDHY) tracks the Markit iBoxx USD Liquid High Yield 0-5 Index, which includes U.S. dollar-denominated high-yield corporate bonds with maturities of 0-5 years. It offers exposure to high-yield bonds with shorter durations, aiming to reduce interest rate and credit risk.
BIL: The SPDR Bloomberg Barclays 1-3 Month T-Bill ETF (BIL) tracks the Bloomberg Barclays 1-3 Month U.S. Treasury Bill Index, which includes short-term U.S. Treasury bills with maturities of 1-3 months. It offers exposure to short-duration U.S. government bonds, providing liquidity and stability to a portfolio.
ATACX: The ATAC Credit Opportunities ETF (ATACX) seeks to provide long-term capital appreciation by investing in U.S. corporate bonds with positive momentum. It uses a proprietary model to rotate between credit sectors based on market conditions, aiming to outperform traditional fixed income benchmarks.

rets |> 
  group_by(symbol) |>
  summarise(min_date = min(date)) |> 
  ggplot(aes(x = min_date, y = reorder(symbol, min_date))) +
  geom_label(aes(label = symbol), size = 3) +
  scale_x_date(date_breaks = "2 year", date_labels = "%Y") +
  labs(title = "ETFs", x = "First available date", y = NULL)

Principal Component Analysis

rets_pca <- rets |> 
  pivot_wider(names_from = symbol, values_from = r) |> 
  arrange(date) |> 
  filter(date >= as.Date("2021-08-27")) |>
  select(-BOXX) |> 
  select(-date)

pca <- prcomp(rets_pca, scale. = T, center = T)

rots <- pca$rotation |> 
  as.data.frame() |> 
  rownames_to_column("symbol") |> 
  as_tibble() |> 
  mutate_if(is.numeric, round, 2)
# Get max absolute value for scaling
max_abs <- max(abs(rots[,-1]))

# Function to generate transparent-to-green gradient
get_color <- function(value, max_value) {
  intensity <- rescale(abs(value), to = c(0, 1), from = c(0, max_value))  
  colorRampPalette(c("white", "green"))(100)[round(intensity * 99) + 1]  # Pick color based on scaled intensity
}

# Apply bold & color gradient
rots_bold <- rots %>%
  mutate(across(-symbol, ~ cell_spec(as.character(.), 
                                     bold = abs(.) == max(abs(.)),  
                                     background = get_color(., max_abs))))


# Render the table
rots_bold |> 
  knitr::kable(format = "html", escape = FALSE) |> 
  kableExtra::kable_styling(full_width = FALSE, bootstrap_options = c("striped", "hover"))

symbol	PC1	PC2	PC3	PC4	PC5	PC6	PC7	PC8	PC9	PC10	PC11	PC12	PC13	PC14	PC15	PC16	PC17	PC18
MBB	-0.23	-0.38	0.02	0.1	-0.03	-0.05	-0.11	0.11	-0.02	0.22	0.25	0.13	0.5	0.1	-0.34	0.12	0.5	0
AGG	-0.22	-0.41	0.02	0.12	-0.05	-0.03	-0.1	0.06	0.06	0.09	0.15	0.09	0.25	0	0.05	-0.23	-0.77	0
REM	-0.27	0.06	0.03	-0.37	0.25	-0.25	-0.04	0.14	-0.54	0.23	0.04	-0.5	0.05	0.15	0.17	0.02	-0.04	0
USRT	-0.27	0.07	0.02	-0.36	0.21	-0.35	-0.19	0.26	0.14	-0.25	-0.32	0.37	0.1	-0.42	-0.04	0.08	0.01	0
QCLR	-0.24	0.2	-0.02	0.39	-0.25	0.12	-0.03	0.28	-0.55	-0.1	-0.34	0.34	-0.03	0.22	-0.02	-0.03	-0.02	0
QQQ	-0.29	0.26	-0.03	0.17	-0.24	0.05	-0.01	-0.01	0.12	0.05	0.11	-0.21	0.11	-0.29	0.09	0.28	-0.05	0.7
QQQM	-0.29	0.26	-0.03	0.17	-0.24	0.06	-0.01	0	0.11	0.05	0.11	-0.21	0.11	-0.29	0.09	0.27	-0.04	-0.71
XCLR	-0.14	0.15	-0.04	0.54	0.8	0.01	-0.04	-0.13	0.06	0.06	-0.01	0	0	-0.02	0.01	0	0	0
RORO	-0.25	-0.05	-0.01	0.05	-0.15	-0.44	0.21	-0.76	-0.11	-0.13	-0.18	0.05	0.03	0.07	-0.14	0	-0.01	0
JOJO	-0.22	-0.33	0.02	0.02	0.05	0.31	0.2	0.13	0.17	-0.34	-0.48	-0.49	0.01	0.04	-0.27	0.01	0.01	0
TLT	-0.18	-0.45	0.04	0.15	-0.05	-0.03	-0.04	-0.02	0.04	-0.09	-0.04	0.03	-0.19	-0.05	0.77	-0.02	0.31	0
SPY	-0.3	0.26	-0.01	0	-0.11	-0.03	-0.04	0.05	0.2	0.02	0.11	-0.11	-0.04	-0.04	-0.05	-0.84	0.23	0.01
STNC	-0.29	0.21	0.02	-0.12	0	-0.1	-0.07	0.09	0.44	-0.14	0.1	0.08	-0.06	0.74	0.1	0.23	-0.06	0
RPAR	-0.29	-0.22	0.05	0.03	-0.02	-0.04	-0.11	0.07	-0.04	0.15	0.23	0.06	-0.78	-0.11	-0.37	0.12	-0.02	0
SDHY	-0.23	0.04	0.01	-0.33	0.08	0.65	-0.43	-0.43	-0.1	0.05	-0.06	0.16	0.04	0	0.04	0	0	0
SGOV	-0.02	-0.05	-0.7	-0.03	0.05	0.02	-0.05	0.01	-0.18	-0.56	0.38	-0.04	0.01	-0.02	-0.03	0	0.01	0
BIL	-0.02	-0.05	-0.7	-0.05	-0.05	-0.04	0.03	0.02	0.18	0.56	-0.38	0.03	-0.05	0.04	0.03	0.01	0	0
ATACX	-0.25	0.03	0	-0.23	0.15	0.26	0.8	0.05	-0.04	0.09	0.19	0.3	0.02	-0.07	0.08	0.01	-0.01	0

library(factoextra)

pca %>% fviz_eig()

pca %>% fviz_pca_biplot()

Linear models

rets_lm <- rets |> 
  pivot_wider(names_from = symbol, values_from = r) |> 
  arrange(date) |> 
  filter(date >= as.Date("2021-08-27")) |>
  select(-BOXX) |> 
  select(-date)

model_mbb <- lm(MBB ~ SPY + TLT + SGOV, data = rets_lm)
model_agg <- lm(AGG ~ SPY + TLT + SGOV, data = rets_lm)
model_rem <- lm(REM ~ SPY + TLT + SGOV, data = rets_lm)
model_usrt <- lm(USRT ~ SPY + TLT + SGOV, data = rets_lm)
model_qclr <- lm(QCLR ~ SPY + TLT + SGOV, data = rets_lm)
model_qqq <- lm(QQQ ~ SPY + TLT + SGOV, data = rets_lm)
model_qqqm <- lm(QQQM ~ SPY + TLT + SGOV, data = rets_lm)
model_xclr <- lm(XCLR ~ SPY + TLT + SGOV, data = rets_lm)
model_roro <- lm(RORO ~ SPY + TLT + SGOV, data = rets_lm)
model_jojo <- lm(JOJO ~ SPY + TLT + SGOV, data = rets_lm)
model_stnc <- lm(STNC ~ SPY + TLT + SGOV, data = rets_lm)
model_rpar <- lm(RPAR ~ SPY + TLT + SGOV, data = rets_lm)
model_sdhy <- lm(SDHY ~ SPY + TLT + SGOV, data = rets_lm)
model_bil <- lm(BIL ~ SPY, data = rets_lm)
model_atacx <- lm(ATACX ~ SPY + TLT + SGOV, data = rets_lm)

library(stargazer)

stargazer(model_mbb, model_agg, model_rem, model_usrt, model_qclr, model_qqq, 
          model_qqqm, model_xclr, model_roro, model_jojo, model_stnc, model_rpar, 
          model_sdhy, model_bil, model_atacx, type = "html", 
          title = "Linear models", out = "lm.html")

**Linear models**

	Dependent variable:

	MBB	AGG	REM	USRT	QCLR	QQQ	QQQM	XCLR	RORO	JOJO	STNC	RPAR	SDHY	BIL	ATACX
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)	(10)	(11)	(12)	(13)	(14)	(15)

SPY	0.085^***	0.065^***	0.965^***	0.801^***	0.536^***	1.283^***	1.278^***	0.550^***	0.689^***	0.158^***	0.800^***	0.378^***	0.335^***	0.0003	0.719^***
	(0.008)	(0.005)	(0.036)	(0.028)	(0.018)	(0.014)	(0.014)	(0.043)	(0.033)	(0.015)	(0.013)	(0.012)	(0.017)	(0.001)	(0.034)

TLT	0.357^***	0.350^***	0.288^***	0.240^***	0.042^**	0.006	0.005	-0.010	0.524^***	0.525^***	0.054^***	0.522^***	0.115^***		0.270^***
	(0.008)	(0.005)	(0.035)	(0.027)	(0.017)	(0.014)	(0.014)	(0.042)	(0.032)	(0.015)	(0.013)	(0.012)	(0.017)		(0.033)

SGOV	0.384	0.743^**	0.085	0.404	0.832	1.026	1.034	3.889	1.719	0.970	-1.009	-0.670	1.039		1.250
	(0.524)	(0.333)	(2.394)	(1.846)	(1.170)	(0.962)	(0.955)	(2.890)	(2.184)	(0.993)	(0.888)	(0.826)	(1.162)		(2.232)

Constant	-0.00001	-0.0001	-0.0004	-0.0002	0.00000	-0.0002	-0.0002	-0.001	-0.001	-0.0001	0.00001	-0.00000	-0.0001	0.0001^***	-0.001
	(0.0001)	(0.0001)	(0.001)	(0.0004)	(0.0003)	(0.0002)	(0.0002)	(0.001)	(0.0005)	(0.0002)	(0.0002)	(0.0002)	(0.0002)	(0.00001)	(0.0005)


Observations	865	865	865	865	865	865	865	865	865	865	865	865	865	865	865
R²	0.733	0.864	0.488	0.525	0.525	0.902	0.903	0.159	0.469	0.631	0.810	0.776	0.336	0.0004	0.392
Adjusted R²	0.732	0.864	0.486	0.524	0.523	0.902	0.902	0.156	0.468	0.630	0.809	0.775	0.333	-0.001	0.390
Residual Std. Error	0.002 (df = 861)	0.002 (df = 861)	0.011 (df = 861)	0.009 (df = 861)	0.006 (df = 861)	0.005 (df = 861)	0.005 (df = 861)	0.014 (df = 861)	0.010 (df = 861)	0.005 (df = 861)	0.004 (df = 861)	0.004 (df = 861)	0.005 (df = 861)	0.0002 (df = 863)	0.011 (df = 861)
F Statistic	786.806^*** (df = 3; 861)	1,827.726^*** (df = 3; 861)	273.492^*** (df = 3; 861)	317.735^*** (df = 3; 861)	316.867^*** (df = 3; 861)	2,640.570^*** (df = 3; 861)	2,661.046^*** (df = 3; 861)	54.172^*** (df = 3; 861)	253.932^*** (df = 3; 861)	491.278^*** (df = 3; 861)	1,220.939^*** (df = 3; 861)	991.772^*** (df = 3; 861)	145.026^*** (df = 3; 861)	0.304 (df = 1; 863)	185.030^*** (df = 3; 861)

Note:	p<0.1; p<0.05; p<0.01

# annualized alphas
alphas <- rets_lm |> 
  select(-c(SPY, TLT, SGOV)) |>
  map(~ {
    model <- lm(.x ~ SPY + TLT + SGOV, data = rets_lm)
    (1+coef(summary(model))[1, 1])^252-1
  }) |> 
  unlist()

ggplot(data.frame(symbol = names(alphas), alpha = alphas), aes(x = alpha, y = reorder(symbol, alpha))) +
  geom_vline(xintercept = 0, linetype = "dashed", color = "red") +
  geom_point() +
  geom_label(aes(label = scales::percent(alpha, accuracy = 0.01)), nudge_x = 0.01) +
  scale_x_continuous(labels = scales::percent) +
  labs(title = "Annualized alphas (since 2021-08-27)", x = NULL, y = NULL)

rets_lm2023 <- rets_lm <- rets |> 
  pivot_wider(names_from = symbol, values_from = r) |> 
  arrange(date) |> 
  filter(date >= as.Date("2023-01-01")) |> 
  select(-date)
  
# annualized alphas
alphas <- rets_lm2023 |> 
  select(-SPY) |> 
  map(~ {
    model <- lm(.x ~ SPY + TLT + SGOV, data = rets_lm2023)
    (1+coef(summary(model))[1, 1])^252-1
  }) |> 
  unlist()

ggplot(data.frame(symbol = names(alphas), alpha = alphas), aes(x = alpha, y = reorder(symbol, alpha))) +
  geom_vline(xintercept = 0, linetype = "dashed", color = "red") +
  geom_point() +
  geom_label(aes(label = scales::percent(alpha, accuracy = 0.01)), nudge_x = 0.01) +
  scale_x_continuous(labels = scales::percent) +
  labs(title = "Annualized alphas (since 2023-01-01)", x = NULL, y = NULL)

rets_lm2024 <- rets_lm <- rets |> 
  pivot_wider(names_from = symbol, values_from = r) |> 
  arrange(date) |> 
  filter(date >= as.Date("2024-01-01")) |> 
  select(-date)
  
# annualized alphas
alphas <- rets_lm2024 |> 
  select(-SPY) |> 
  map(~ {
    model <- lm(.x ~ SPY + TLT + SGOV, data = rets_lm2024)
    (1+coef(summary(model))[1, 1])^252-1
  }) |> 
  unlist()

ggplot(data.frame(symbol = names(alphas), alpha = alphas), aes(x = alpha, y = reorder(symbol, alpha))) +
  geom_vline(xintercept = 0, linetype = "dashed", color = "red") +
  geom_point() +
  geom_label(aes(label = scales::percent(alpha, accuracy = 0.01)), nudge_x = 0.01) +
  scale_x_continuous(labels = scales::percent) +
  labs(title = "Annualized alphas (since 2024-01-01)", x = NULL, y = NULL)

Better or worse ETFs

Martin Geissmann

2025-02-07

Principal Component Analysis

Linear models