Surprisingly Profitable Pre-Holiday Drift Signal for Bitcoin
Introduction
Cryptocurrency markets have matured into a distinct asset class characterized by extreme volatility, deep liquidity pools, and worldwide retail participation. Traditional equity and commodity markets exhibit a well-documented pre-holiday effect, where returns on trading days immediately preceding public holidays tend to outperform other days. Given that Bitcoin is often described as the archetypal absolute risk asset, it is natural to hypothesize that any calendar-driven anomalies observed in equities should manifest—or even amplify—in crypto markets.
However, unlike equity markets, where institutional investors and marketing calendars drive collective behavior, crypto markets are more dispersed, retail-dominated, and influenced by nontraditional information flows. This article investigates whether the classic pre-holiday effect applies to Bitcoin and assesses the extent to which it can be amplified by an attention-grabbing momentum filter based on local price highs.
Background
The pre-holiday effect was first documented by Ariel (1990), who showed that U.S. equity markets earn abnormally high returns on trading days immediately preceding public holidays. Subsequent work by Kim and Park (1994) confirmed this pattern across the NYSE, AMEX, NASDAQ, FT30 (U.K.), and Nikkei Dow (Japan), demonstrating the anomaly’s persistence under differing institutional arrangements. Quantitative research platforms, such as us at Quantpedia, classify the pre-holiday anomaly as one of the most substantial calendar effects, reporting that average returns on pre-holiday days can be more than ten times larger than on regular trading days. Hansen and Lunde (2003) developed a rigorous testing framework that conditions on the complete set of possible calendar dummies to avoid data-mining biases, and they show this approach retains good power by exploiting the specific correlation structure of calendar effects.
Behavioral finance casts calendar anomalies as departures from market efficiency driven by sentiment and attention biases. Fama (1998) characterizes what is now known as the anomalies literature, highlighting how phenomena like the pre-holiday effect contradict the semi-strong form of the efficient market hypothesis.
Methodology & Data
Data span daily closing prices of Bitcoin in the form of BITO ETF from January 2018 to June 2025, sourced from our internal database (2018-2021 as futures proxy, 2021 – 2025 as ETF itself). Why did we use BITO ETF (which tracks the bitcoin futures) and not spot Bitcoin ETFs like IBIT, GBTC, or FBTC, or the spot Bitcoin itself? As we repeatedly stated, we consider only data from 2018 onward to be relevant for the backtesting purposes (since the introduction of the regulated Bitcoin futures) and combination of the Bitcoin futures data and BITO ETF gives the long historical window over which we can run our tests.
Public holiday dates are defined by a consensus U.S. calendar, as per Time and Date (see example for 2025), acknowledging that U.S. holidays often set the tone for all global equity markets, thus hypothetically influencing global crypto sentiment.
While ProShares Bitcoin ETF (BITO) is traded on U.S. exchanges, it cannot be traded on D0, which falls on the day of a holiday. Hence, we always hold long Bitcoin positions during the holidays, hoping to capitalize on attention-grabbing events.
Public Holiday Window Analysis
To start, we examine trading period starting five trading days before each holiday (D–5) and ending five trading days after (D+5).
Following is the histogram (bar chart) showing the daily distribution of returns, in the mentioned days preceding and after the holiday:
Picture 1: Holiday drift in Bitcoin
This charts visualizes whether there is any pre-holiday drift and can help us to prepare basic seasonal strategy that will span multiple days.
Following that preliminary analysis, we can define a simple trading strategy that buys crypto on the day preceding the holiday (D-1), holds through the holiday, and liquidates at the close of the day after the holiday (D+1). And here is the equity curve, which shows the appreciation of the initial investment for such a strategy:
Picture 2: Holiday drift in Bitcoin – D+1 do D-1, equity curve
Nothing extraordinary, right? While this strategy yields some positive returns, it is not very satisfactory and tends to be flat for most of the time. Therefore, we need to look for another component to enhance the strategy’s returns.
Attention-Augmented Strategy
Based on the first simple strategy, we hypothesize that coupling the simple holiday drift with another attention-grabbing anomaly will yield significantly better results. And what could these 2nd anomaly be? Our favorite 10-day high strategy. Why should we combine these two anomalies and hope for better returns? Around holidays, retail traders typically have more free time and are more prone to engage with financial markets out of curiosity or even boredom. If, during this period, crypto prices are already breaking to new short-term highs, the additional attention and speculative activity from retail participants can act as fuel for further price increases. In essence, the combination of a bullish technical signal and heightened retail activity around holidays can exacerbate upward moves, creating a repeatable and potentially profitable trading pattern.
We thus define an N-day high filter: a day on which Bitcoin’s closing price exceeds the maximum closing price of the preceding N trading days (N ∈ {5, 10, 20}). Then we identify dates that are simultaneously within the holiday window (D–5 to D+5) and satisfy the N-day high condition. For each qualifying date, one enters a long position at the close and liquidates at the next trading day’s close.
The following are the results from such an investigation, first in the form of histograms depicting the distribution of results for each considered day alone:
Picture 3: 5-day high filter + Holiday drift
Picture 4: 10-day high filter + Holiday drift
Picture 5: 20-day high filter + Holiday drift
That’s a significant difference compared to the histogram showing just the pre-holiday effect without the 10-day high filter. It really seems that the boredom of the holidays helps cryptocurrency speculation. The whole period between D-1 and D+1 has significantly positive average daily returns.
Strategy Evaluation
Let’s evaluate they equity curves themselves. We ultimately chose window D-1 to D+1 and here are the results:
Picture 6: Cumulative equity curves for each N-day high filter + Pre-holiday Drift (D-1 to D+1)
Performance and risk metrics (compound annual return, annualized [yearly] volatility, Sharpe ratio, maximum drawdown, and CAR / maximum drawdown) can be found in the table below:
The best-performing 5-day high variant version comes at a cost with a tradeoff of the worst risk terms, including the highest volatility and maximum drawdown from the tested sample. Still, the Sharpe ratios over 0.6 and Calmar ratio over 0.7 look really attractive for us. It really pays off to join the speculating crowd from time to time…
Discussion & Conclusion
Our findings confirm that Bitcoin exhibits a pre-holiday drift akin to that observed in equity markets, but only when coupled with a short-term momentum trigger. The N-day high filter serves as a proxy for heightened market attention, capturing instances in which retail and institutional participants simultaneously face discretionary trading time and positive feedback loops. This synergy yields robust returns with attractive risk-adjusted profiles.
These results underscore the importance of combining calendar-based signals with behavioral proxies in absolute risk assets. They suggest that crypto markets are not immune to the herding and sentiment biases that characterize traditional markets; in fact, those biases may be magnified given the asset’s speculative nature and retail concentration. Once the N-day high filter is applied, the strategy’s performance improves markedly. The filtered holiday strategy is performing better than the unfiltered one, demonstrating the compounding effect of calendar timing and momentum screening.
Author: Cyril Dujava, Quant Analyst, Quantpedia
Are you looking for more strategies to read about? Sign up for our newsletter or visit our Blog or Screener.
Do you want to learn more about Quantpedia Premium service? Check how Quantpedia works, our mission and Premium pricing offer.
Do you want to learn more about Quantpedia Pro service? Check its description, watch videos, review reporting capabilities and visit our pricing offer.
Are you looking for historical data or backtesting platforms? Check our list of Algo Trading Discounts.
Would you like free access to our services? Then, open an account with Lightspeed and enjoy one year of Quantpedia Premium at no cost.
Or follow us on:
Facebook Group, Facebook Page, Twitter, Linkedin, Medium or Youtube
Share onLinkedInTwitterFacebookRefer to a friend
