Holiday Effects on S&P 500 Index Volatility and Return Rates for
Market Efficiency Validation
Zihan Li
School of ACCOUNTING& FINANCE, Hong Kong Polytechnic University, Hong Kong, China
Keywords: Holiday Effect, Market Effectiveness, GARCH.
Abstract: The holiday effect in the financial landscape is important, especially in understanding how market behaviour
changes around holidays. This phenomenon has been under-appreciated in the past. This article provides
additional evidence for the holiday effect. It analyses whether the volatility and returns in the S&P 500 Index
are significantly different between the five trading days before the holiday, and the rest period. Next, the
effectiveness of the market is determined through the 'efficient market hypothesis'. The adjusted closing price
for the S&P 500 index was calculated for each trading day between December 23, 2013 and December 29,
2023. This work will build a Generalized Autoregressive Conditions Heteroskedasticity model (GARCH) to
calculate the volatility of the index through the index returns. Both datasets are tested using the F Test method
and T Test method. The results show low volatility and high returns for the five-trading day of the pre-holiday
time period. As a result, it indicates that the holiday effect is predominant in S&P 500 Index, which represents
by the 500 U.S. companies that are under the control of the S&P 500. Due to arbitrage opportunities, the U.S.
Stock Market has low efficiency.
1 INTRODUCTION
In financial market analysis, the "holiday effects"
refers to an anomaly in the market (Gama, 2013;
Brockman, 1998), a pattern of unusual volatility and
returns on the stock market, which often occurs just
before a holiday. The stock market performs better
than usual in the days leading up to holidays, and
rising prices of stocks imply high returns. This
suggests that stock returns before holidays are
different from other time periods. Short-term returns
are usually higher before holidays, which is contrary
to the efficient market hypothesis. According to the
EMH, stock prices already contain all available
information, so investors cannot outperform the
market. The data was selected for each trading day
from December 23, 2013, to December 29, 2023. This
work will build a Generalized Autoregressive
Conditions Heteroskedasticity model (GARCH) to
calculate the volatility of the index through the index
returns. Both datasets are tested using the F-Test
method and T-Test method. The results show low
volatility and high returns for the five-trading day of
the pre-holiday time period. These results indicate
that the holiday effect is predominant in U.S. Stock
Market, which represents by S&P 500 Index. Due to
arbitrage opportunities, the U.S. Stock Market has
low efficiency.
In other words, no investor can make exceed
profit in an efficient market. This makes sense to a
large extent, but everything has two sides. The
discovery of the holiday effect challenged EMH by
suggesting that the behavior of market participants
could be influenced by irrational factors such as
investor sentiment or behavioral biases. Such unusual
stock performances have attracted the interest of
many financial analysts and market participants
because they allow investors to earn additional
returns on both short-term and long-term trades. Such
unusual phenomenon is represented by several
markets around the whole world and it is worth
exploring whether such phenomenon could have a big
impact on major stock indices like S&P 500. Because
the U.S. stock market is representative of global
equity markets. Investigating the influence of the
holiday effect on the S&P 500 and analyzing in depth
its role on volatility and returns will greatly assist the
understanding of market anomalies in the U.S. stock
market. Examining the holiday effect not only
challenges traditional financial theories, but also
provides insight into the correlation between the S&P
500 and related financial contracts through data
Li, Z.
Holiday Effects on S&P 500 Index Volatility and Return Rates for Market Efficiency Validation.
DOI: 10.5220/0013270300004568
In Proceedings of the 1st International Conference on E-commerce and Artificial Intelligence (ECAI 2024), pages 547-552
ISBN: 978-989-758-726-9
Copyright © 2025 by Paper published under CC license (CC BY-NC-ND 4.0)
547
comparisons that reveal the efficiency of the U.S.
market as a whole.
Market volatility and returns are widely
recognized as important indicators of economic
performance and the investment performance of the
global financial system. Volatility refers to the
magnitude and speed of change in asset prices over
time. High-volatility financial products experience
significant price changes, reflecting the
unpredictability and instability of markets, while low-
volatility financial instruments are relatively stable.
The performance of the Standard & Poor's 500 Index
(S&P 500) is closely related to the volatility of the
stock market, especially in the United States (Ang,
Hodrick, Xing, & Zhang, 2006). Therefore, its
performance can serve as a reliable indicator of
market volatility in the United States. In addition, the
S&P 500 index represents 500 of the largest and best-
known publicly traded companies in the United
States, covering different industries such as energy,
consumer goods, and technology, reflecting the
market capitalization performance of these industries
and thus the overall performance of the U.S.
economy. As a result, changes in the index can also
reflect the overall economic conditions in the United
States. In order to effectively explore the anomalies
of the U.S. market and help financial market
participants manage risks and formulate investment
strategies; the selection of the S&P 500 Index is
reasonable. In addition, volatility and yield, as
important parameters for assessing investment
performance, are sensitive to changes in market
efficiency. Studying how they are affected, and their
intrinsic relationship can make important predictions
about market trends and help financial institutions
and individual investors make informed decisions.
Since this study more focus on analyzing in depth
whether holiday effects exist. This essay will provide
a comprehensive understanding of U.S. stock market.
It will also examine whether they have an effect on
S&P500 volatility and returns. A detailed analysis of
this holiday effect will help market participants, risk
managers and traders to better understand the market
and optimize their trading and risk management
strategies. it will also provide richer insights on
patterns of market behavior in periods of unusual
volatility. This will be useful for understanding and
predicting future market direction. The article
predicts that the S&P 500 will show a holiday effect
in the follow-up experiments. This manifests as lower
volatility, higher returns and higher risks during
holiday periods.
2 LITERATURE REVIEW
2.1 Holiday Effect Verification
The EMH theory is one of the cornerstones in
financial economics. This theory assumes market
prices reflect all information available, thus rejecting
the possibility that speculators could exploit
information to generate excess returns (Fama, 1970).
The EMH theory has been widely accepted by
professional financial practitioner as applicable to the
financial markets. However, there are persistent
anomalies that have been observed in actual markets
which challenge this theory. Investors may notice
anomalies when they look at data and see that a stock
has outperformed the market over time in a particular
characteristic. One such anomaly is the holiday effect.
Before delving into the volatility and returns of the
S&P 500 over time, it's important to first determine if
the holiday effect exists.
Lakonishok & Smidt (Lakonishok, 1988), in order
to explore market anomalies and test for persistent
seasonal patterns, conducted pioneering research to
test for 90 years of U.S. closing price data for Dow
Jones Industrial Average. Lakonishok & Smidt
(Lakonishok, 1988), by analysing the data collected,
pointed out that there were persistent and abnormally
higher returns around certain times, such as
weekends, the end of the month, the end of the year,
and holidays. Stock returns are significantly higher at
these times than they are on normal trading days,
which indicates that there may be systematic
anomalies in this market. The holiday effect was
brought into focus by the challenge to traditional
market models.
Kim and Park (Kim, 1994), comparing data from
July 1, 1973, to June 30 1987, explore the holiday
effect on stock returns. They found that stocks on the
three major U.S. stock exchanges (the New York
Stock Exchange, the American Stock Exchange, and
the NASDAQ) had unusually high returns in the days
leading up to the holidays. The study also verified the
holiday effect in the UK and Japanese stock markets.
The results show that the holiday effect in the UK and
Japanese markets is independent of the US market,
and that this phenomenon persists even after taking
into account the impact of the US holiday effect. This
indicates that there is an international linkage to the
holiday effect. By comparing stock returns between
the holiday season and the next, it was confirmed that
all 14 CEE financial market had abnormally high
stock return (Gakhovich, 2011). The holiday effect is
also present in this case. The holiday effect is
ECAI 2024 - International Conference on E-commerce and Artificial Intelligence
548
confirmed in multiple locations as Lakonishok &
Smidt's findings have been verified.
The holiday effect on the Swedish stock market is
missing, according to Eidinejad & Dahlem's
(Eidinejad, 2021) analysis of daily price data for the
Swiss stock exchange over a 40-year period between
1980-2019. Their results show a significant post-
holiday effect in the Swedish Stock Market over the
entire sample time period. This is evidenced by
significantly higher stock market returns than other
times. This is not the same as the holiday effect,
which indicates abnormally volatile returns before the
holidays. When the sample was broken down by
decade, researchers found that the holiday effects
were primarily present in the 1990s & 2000s. The
effect was not seen in the 1980s & 2010s. The study
also found no pre-holiday effects in any period. The
article explains why this phenomenon occurs. The
rapid development of Internet in 1990s may well have
caused investors to pay more attention to the market
during holidays. This may have been a reason for the
post-holiday effects in 1990s. By 2010, the market
might have adjusted to this anomaly. The findings of
the Swedish stock exchange support the idea that
holiday effects are not always present on the market.
2.2 S&P 500 Volatility Effect
Financial market analysis is not complete without
understanding the impact of market volatility on
investment returns. The traditional view of
investment emphasizes that low-volatility stocks on
the market are typically associated with less price
fluctuation and may therefore represent lower risk
and lower return. This intuitive concept implies
higher risk is required for investors to achieve higher
returns.
A leading academic quarterly in finance called
The Journal of Portfolio Management published an
article that suggests a different view. Blitz & Van
Vliet (Blitz, 2007) created quartile portfolios based
upon historical volatility, and compared performance
of different volatility-based portfolios using stock
market data from 1986 to 2006. Their research
revealed that low volatility portfolios are not only
associated with lower risks, but also have higher
adjusted returns than high volatility portfolios. It is
important that they also analyzed portfolios from
other markets. This phenomenon is not limited to one
market but is present in many markets around the
world, including the U.S.A., Europe, China and
Japan. To verify that this phenomenon was not a
fortuitous circumstance, the article also explored
whether other factors, such as size, value, or
momentum, which are known to have financial
market effects, could have an effect on this low-risk,
but high return, compared to the return of a portfolio
with low volatility. The results show that the low
volatility is still significant after controlling for all of
these factors. This indicates that it is a separate and
economically significant effect. This unusual
phenomenon brings up an important point: in certain
situations, the traditional association between equity
market volatility and risk may need to be reevaluated.
The article provides new perspectives on practical
risk management and asset allocation.
In this context, it is important to analyze changes
in volatility and returns for the S&P 500, under the
influence of holiday effects. The performance of S&P
500 can be a good indicator of the state of the U.S.
stock market. Its performance during a certain time
period such as pre-holidays, for example, can give
investors important information to make investment
decisions. A more detailed analysis of the S&P 500's
risk and reward patterns can reveal the nuances in
market dynamics when influenced by specific
holidays. This can help investors to not only
capitalize on trading opportunities over the short-
term, but also give them insight into long-term
investment strategy development.
3 METHODOLOGIES
3.1 Data Preparation
In contemporary era background, the stock market
provides a fair, transparent and cost-effective
environment for the investors to make investment on
nearly equal condition to foster the healthy
development of economic. investors can easily search
lots of stock information from some financial website
such as wind, yahoo finance and Bloomberg. This
article chooses the yahoo finance as data resource.
Because this website can be used by public investors
to find stock price. Therefore, this article used the
adjust closed price of S&P 500.
S&P 500 Index Adj Close: This dataset represents
the closing price of a stock at the end of a given
trading day, after taking into account factors such as
ex-rights and dividends. Its price reflects the true
value of the stock. Historical data from 2013-12-23 to
2023-12-29 is included here.
In addition, the calculation of index return rate is
the key factor of this research. The volatility this
article discuss is also generated from the return rate.
Holiday Effects on S&P 500 Index Volatility and Return Rates for Market Efficiency Validation
549
It is essence to use precise quantitative date for the
calculation of return rate of S&P 500.
3.2 Model Construction
Since volatility is not directly observable, this paper
requires a reliable proxy variable. If the conditional
mean is zero, the square of the returns can be used as
an unbiased estimate of the underlying volatility
process. (Awartani, 2005). Therefore, creating the
Specialized models to make prediction by using
return rate is the key factor of this experiment.
At the beginning of the experiment, this article
wanted to depend on the Autoregressive Conditional
Heteroskedasticity (ARCH) which can be used to
catching volatility characteristics in financial time
series data. However, Zhang (Zhang, 2016)
mentioned that this model needs the high amount of
data which can lead to a great number of problems in
the process of calculate the volatility. For instance,
the problem of multicollinearity of the explanatory
variables and the inability to ensure that the
restriction α is always not less than zero.
Therefore, this article chose the GARCH model
(Xu, 2011) to calculate the volatility. Because this
model can easily find heteroskedasticity, which
suggest that the volatility of the index rate is not
always constant in time series. It always changes over
the time. In addition, the basic factor of the GARCH
model is predicting the future volatility by using the
past volatility.
3.3 Test Method
Since this article must judge the significant difference
between two data set, this article will use some test to
help determined. the F Test and T Test is the best
option to make judgement.
3.3.1 F Test
For the F test, it usually is used to compare whether
the variances were significantly different between
two or more samples.
The basic set of steps is to ensure the null
hypothesis H0: The variance of two samples is equal;
alternative hypothesis Ha: The variance of two
samples is not equal.
The formulas are shown below:
𝑆
𝑛
𝑥
𝑥^2

𝑘1
(1)
𝑆
𝑛
𝑛
1𝑆

𝑁𝑘1
(2)
Throughout the experiments, this work decided to
use the EXCEL to help simplify the process of
calculating the F statistic.
Finally, this article will determine the
corresponding p value based on the F statistic. The
acceptance of the original hypothesis will be
determined by the comparing the p value to the
magnitude of 0.05.
3.3.2 T Test
T-tests are commonly used to determine whether the
means of two or more samples are significantly
different. The basic steps include setting the null
hypothesis H0: The means of two samples are equal;
alternative hypothesis Ha: The means of the two
samples are not equal. If the p-value is less than 0.025,
the null hypothesis is rejected.
4 RESULTS AND DISCUSSION
4.1 Calculate The S&P 500 Volatility
For the validation of the existence of the holiday
effect. this article first wanted to use the average daily
volatility of the S&P 500 based on all of the 2013 to
2023 period shown in Figure 1 to see if there were any
unusual fluctuations in it. This essay imported the
data in to RStudio and obtained the GRACH model
diagram shown in the figure after code programming.
Figure 1: 10 Years Index Volatility (Photo/Picture credit:
Original).
4.2 Return Rate Calculation
"The most widely-used statistics in finance are
expected return and volatility" (Sakr & Sherif, 2017).
The calculation of volatility is derived from the
returns. Therefore, this study calculated the returns
first. This article can easily calculate the daily returns
by using a formula on an EXCEL sheet. The formula
ECAI 2024 - International Conference on E-commerce and Artificial Intelligence
550
is the ratio of adjusted closing prices of the current
trade, minus the adjusted close of the previous day's
trading, to the adjusted close of the previous day's
trading. The same formula can be used to calculate
daily returns after obtaining the daily returns of Index
and Futures over the ten-year period.
4.3 F Test
Excel software is used to calculate the F statistics,
with significance set at 95%, corresponding to a p-
value of 0.05. If the p-value is below 0.05, the null
hypothesis is rejected, indicating a significant
variance difference between the two groups.
4.3.1 S&P 500 Return Rates
The null hypothesis for this F test is the following: the
variance in return rates of the S&P 500 index in the
pre-holiday periods equals the variance in return rates
of the index in the non-holiday periods.
The alternative hypothesis is that the variances of
the S&P 500 return rate in pre-holiday periods are not
equal to the variances of the index return rates during
non-holiday periods.
H0: b1=b2 (3)
Ha: b1b2 (4)
The graph below shows that the F statistic is
1.37493. This means that the p-value is 1.74E-04
which is less than 0.05
There is strong evidence that this article rejects the
null hypotheses and accept the alternative hypotheses.
proving a significant difference in the variances of the
S&P 500 Index return rates between pre-holiday and
non-holiday periods.
4.3.2 S&P 500 Volatility
The null hypothesis for this F test is the following:
the variance in Volatility of the S&P 500 index in
the pre-holiday periods equals the variance in
Volatility of the index in the non-holiday periods.
The alternative hypothesis is that the variances of
the S&P 500 Volatility in pre-holiday periods are
not equal to the variances of the index Volatility
during non-holiday periods.
H0: b3=b4 (5)
Ha: b3b4 (6)
The graph below shows that the F statistic is
2.24308. This means that the p-value is 1.05E-23
which is less than 0.05
There is strong evidence that this article should
reject the null hypotheses and accept the alternative
hypotheses. This proves that there is a significant
difference between the variances of the S&P 500
Index Volatility during Pre-holiday and the variances
of the index Volatility during the non-holiday
periods.
4.4 T Test
The F Test shows that both return rates and
volatility are significantly different between the
holiday period and the non-holiday periods. This
article chose the T test for two sample
heteroskedasticity. Then, this essay will calculate
the confidence intervals of both and determine if
there is a significant difference in the mean.
4.4.1 The T Test about S&P 500 Return
Rates
The null hypothesis for this t-test is that the index
returns rates during non-holiday times are equal to the
S&P 500' return rates in pre-holiday times. The
alternative hypothesis is the S&P 500' return rate in
the pre-holiday period does not equal that of the index
during the non-holiday period.
H0: m1=m2 (7)
Ha: m1m2 (8)
Since this article set the significance at 95% it
means that a=0.05. When p value (single tailed
critical mass), is less than a/2 =0.025the null
hypotheses was rejected. In this test, p value (single
tailed critical mass), is 0.008292. This is less than
0.025. there is strong evidence for rejecting the null
hypothesis, and it can be concluded that there are
significant differences between those two types of
mean.
This article established a confidence interval by
the information which is gathered from the internet.
This article put the follow formula about confidence
interval into RStudio
(m1-m2) ± 1.96 * (σ / sqrt(n)) (9)
In the end, this article gets the 95% confidence
interval for m1-m2 which is between-0.009473-
0.008196359
It represents that there is a 95% probability that
any S&P 500 return rate from non-holiday period
minus any S&P 500 return rate from pre-holiday will
be within this interval. Overall, the SP500 return rate
from the non-holiday period is less than the SP500
return rate from pre-holiday.
Holiday Effects on S&P 500 Index Volatility and Return Rates for Market Efficiency Validation
551
4.4.2 The T Test about S&P 500 Volatility
This t test's null hypotheses H0 are that the index
volatility in non-holiday times is equal to the
S&P500's mean volatility before holidays. The
alternative hypothesis Ha is the S&P 500's mean
volatility in the pre-holiday period does not equal
that of the index during the non-holiday period.
H0: m3=m4 (10)
Ha: m3m2 (11)
Since this article set the significance at 95% as its
regulation, it means that a=0.05. When p value (single
tailed critical mass), is less than a/2 =0.025, the null
hypotheses was rejected. In this test, p value (single
tailed critical mass), is 0.003443. This is less than
0.025. there is strong evidence for rejecting the null
hypothesis, and it can be concluded that there are
significant differences between those two types of
mean.
This article established a confidence interval by
the information which is gathered from internet. This
essay put the follow formula about confidence
interval into RStudio
(m3-m4) ± 1.96 * (σ / sqrt(n)) (12)
In the end, the 95% confidence interval can be
obtained for m3-m4 which is between 0.0002820763
and 0.0012361062.
It represents that there is a 95% probability that
any S&P 500 volatility from non-holiday period
minus any S&P 500 volatility from pre-holiday will
be within this interval. Overall, the SP500 volatility
from the non-holiday period is higher than the SP500
volatility from pre-holiday. This is consistent with the
domain knowledge about the index holiday effect.
5 CONCLUSIONS
This article analyzes whether the holiday effects
exist in the S&P 500. It also verifies the validity of
the market. This analysis is used both to determine if
the holiday effect has an impact on the stock market,
and to verify its validity. This analysis gives investors
food for thought as they decide on trading strategies
and risk-management strategies. This study used a
variety of data analysis techniques throughout the
paper, including the GRACH model and T-Test. The
GRACH models have been applied to more than 5000
data sets. The predictions are largely confirmed by
this research.
The model which is used in this article allowed to
calculate volatility of the S&P 500 based on its
returns. It verified the holiday effect by
experimenting with the volatility and returns of S&P
500. This article analyzed the holiday effect using
data analysis concluded that it manifests itself as
lower holiday volatility. This is contrary to what this
paper initially predicted. The expectations are higher
returns and lower volatility. This study can provide a
better option for the risk-averse investor who wants
to trade stocks. There are also some limitations. For
instance, this article successfully verified the
existence of holiday effect by studying the return rate
and volatility of sp500 index. However, it ignores
studying the components. In addition to that, this
article cannot make definition about which vacation
day is most affected by the holiday effect, which
make the analysis of the impact of holiday effect is
not precise as expected. This research hope to adopt a
more comprehensive approach to analysis in the
future to assist investors choose the best investments
strategy.
REFERENCES
Awartani, B. M. A., & Corradi, V. 2005. Predicting the
volatility of the S&P-500 stock index via GARCH
models: the role of asymmetries. International Journal
of Forecasting, 21(1), 167–183.
Blitz, D., & Van Vliet, P. 2007. The volatility effect: Lower
risk without lower return. Journal of portfolio
management, 102-113.
Bollerslev, T., Engle, R. F., & Nelson, D. B. 1994. ARCH
models. Handbook of econometrics, 4, 2959-3038.
Brockman, P., & Michayluk, D. 1998. The persistent
holiday effect: Additional evidence. Applied
Economics Letters, 5(4), 205-209.
Eidinejad, S., & Dahlem, E. 2021. The existence and
historical development of the holiday effect on the
Swedish stock market. Applied Economics Letters,
29(19), 1855–1858.
Fama, E. F. 1970. Efficient Capital Markets: A review of
theory and Empirical work. the Journal of Finance/the
Journal of Finance, 25(2), 383.
Gama, P. M., & Vieira, E. F. 2013. Another look at the
holiday effect. Applied Financial Economics, 23(20),
1623-1633.
Kim, C., & Park, J. 1994. Holiday effects and stock returns:
Further evidence. Journal of Financial and Quantitative
Analysis, 29(1), 145.
Lakonishok, J., & Smidt, S. 1988. Are seasonal anomalies
real? A Ninety-Year perspective. Review of Financial
Studies/the Review of Financial Studies, 1(4), 403–425.
Xu, J., Zhang, Z., Zhao, L., & Ai, D. 2011. The application
review of GARCH model. International Conference on
Multimedia Technology.
ECAI 2024 - International Conference on E-commerce and Artificial Intelligence
552