Research Article #29 - Bitcoin Pre-Holiday Effect
Is the anomaly also present in cryptocurrencies?
👋 Hey there, Pedma here! Welcome to the 🔒 exclusive subscriber edition 🔒 of Trading Research Hub’s Newsletter. Each week, I release a new research article with a trading strategy, its code, and much more.
Join over 1K+ readers of the newsletter!
In today’s issue of Trading Research Hub we study a strategy with the following historical performance:
Total Unleveraged (%) Returns: 226%
Win Rate: 55%
Annualized Volatility: 14.46%
Annualized Sharpe: 0.52%
Average Drawdown: -11%
Introduction
Hey everyone, Pedma here!
Today, we are going to analyze another calendar effect, that has produced some interesting historical anomalies in stock returns.
This anomaly has not only been detected in U.S. equities but also in international markets.
As usual we will apply this idea to the cryptocurrencies market, as that is the focus of this newsletter.
Just to give a brief context, on what calendar effects are, I am going to give you a few examples.
There’s been studied anomalies in calendar effects such as:
Day of the week
Week of the month
Month of the year
Turn of the month
Holidays
Seasons
Today we will focus on one group of these anomalies… holidays.
The strategy’s rules will be given in the Strategy Parameters section below.
I had to use Bitstamp data for this because there’s a limited amount of holidays per year and we need as much data as possible.
We will also be delving into other altcoins where the anomaly should also be present, depending on its robustness.
Let’s dive into the strategy below!
Index
Introduction
Index
Thesis for the Strategy
Strategy Parameters
Strategy Performance and Results
Python Code Section
Sponsor Of Today’s Article
Thesis for the Strategy
There’s been a lot of effort devoted to research on calendar effects.
But as with all market anomalies, there’s always questions about their statistical significance.
Many researchers, that usually do not allocate capital to their own strategies, resort to blind data mining.
Data mining is traditionally done by feeding a machine learning model with a bunch of parameters, and forcing it to find the best possible parameters, given a set of objectives.
This leads to a lot of problems, because price series are filled with noise, and market dynamics are always changing.
However we can do some work to mitigate the possibility of that happening to us by focusing on robust ideas.