# Backtesting the week: Which days are best for trading crypto?

In traditional stock market investing, it’s a well-documented phenomenon that certain days of the week tend to be more profitable than others.

The __Weekend Effect__, for example, is a market phenomenon which posits that stock returns on Mondays are often lower compared to previous Friday, as companies will often try to ‘bury’ bad news by releasing them on Friday evenings, after the market shuts down.

And while crypto never sleeps, the same ‘day bias’ has been creeping into digital asset markets for a good while.

To date, there hasn’t been much research into the ‘day of the week’ effect for crypto. Do the same rules apply? Buy coins on Monday + unload before the weekend = lambo? What are the ‘days of the week’ axioms that rule the crypto market?

**Testing the ‘Days of the Week’ Strategy**

To find out, we decided to take a closer look at the average BTC price returns for each day of the week, including weekends. As we often do in our analyses, we use BTC as a proxy for the entire crypto market, as it still very much dictates the price action of most altcoins over time.

To start, let’s take a look at the average log returns for Bitcoin from 2017 until present day:

*Note for stats geeks: When you average over returns (changes in percent) you probably want to use log returns (or the cumulative returns). When averaging regular returns you run into problems with summing up positive and negative returns. Imagine an asset dropping 50% and then going up 50% the next day. The average would be a return of 0% when in reality you would be at -25% because you first lose half of your money and then only get 50% on that half of your money. Using log returns solves that problem.*

Voila! We can immediately see there are some big differences between BTC returns on different week days:

**Monday and Saturday emerge as clear winners**, with average returns of +0.69% and +0.5%, respectively. If you bought BTC every Saturday from the start of 2017 until today and sold it the same night each time, your current Bitcoin portfolio would be sitting comfortably in the green.**Wednesday, Thursday and Sunday**all returned negative values for the past 2 years. To date, Wednesday proved to be the worst performing day.

Armed with this information, we can now try and backtest a day-sensitive BTC strategy. Let’s see how our BTC portfolio would have performed if - starting in 2018 - we only bought Bitcoin on Mondays and Saturdays and sold at the end of the day:

The blue line shows our ‘days of the week’ strategy; the orange line - which we’re using as a benchmark - shows HODLing BTC for the same time frame.

In the observed time frame, our strategy would pocket us a cool **68% profit**, while HODLing Bitcoin resulted in about **38% loss**.

Unfortunately for our model, the infamous drop at the end of 2018 happened to be on a Monday, making the comparison a bit tighter than it would’ve been otherwise. Still, even with that (massive) outlier, our day strategy recorded a relatively stable growth during most of the 2018 bear market, and seamlessly beat the benchmark.

Looking further in the past provides interesting results as well. Backtesting the same day strategy, only this time from 2017 onwards, gives the following outcome:

This time, our strategy does not in fact outperform HODLing, which is not too surprising considering that there were relatively few bad days for Bitcoin in 2017. In this extended time frame, our day strategy would’ve returned 469% profit - or about 71% of our benchmark’s (HODLing) returns.

Although that may sound like a failed experiment, it is worth noting that we get these 71% of the returns of only **two** days of the week. Also we have a **massive **difference in volatility exhibited by the two strategies.

While HODLing Bitcoin took market participants on a non-stop rollercoaster ride - skyrocketing their risk exposure - our day strategy successfully removed much of BTC’s inherent volatility, charting a slow-but-steady, almost continuously upward trajectory across the observed time frame.

In numbers, the annualized standard deviation of our day strategy for the observed time frame was 0.43, while the annualized standard deviation of holding was 0.84. The annualized Sharpe ratio of our strategy was 1.65; for holding, it was 1.35

As we can see, the strategy **significantly reduces volatility and enances risk-adjusted returns**, providing a better risk-reward balance for investors.

Perceived volatility of the crypto market, spearheaded by BTC, has been a huge barrier to entry for both hobby and institutional investors. Even when underperforming compared to HODLers, our strategy proves to be a very effective ballast for Bitcoin’s price action - sacrificing some of its returns for a substantial reduction in associated risk.

**Eliminating hindsight bias from our results**

The backtesters among you may be thinking: isn't testing this strategy **on the same time frame **as the one that we already calculated the average returns for a classic case of hindsight bias?

It is indeed! Backtesting a strategy on a certain time frame after we already know what worked in that same frame is pretty self serving. After all, hindsight is always 20-20, right?

So instead of testing our day strategy back, let us now try and test it ‘forward’ - starting in 2018.

Let’s imagine that we are at the beginning of 2018, and want to start trading based on the most profitable days for Bitcoin so far.

As 2018-onwards hasn’t happened yet, we can’t rely on previously mentioned data. Instead, we go even further back, and calculate the average log returns for BTC between 2016 and 2017, so we can use those results in our model.

Here are the 2016-2017 average log returns for Bitcoin:

In the end, we actually get a pretty similar output as we did when calculating 2017-2019 log returns, if you consider Bitcoin’s bullish tendencies throughout 2016.

The best performers are once again Monday, Thursday and Saturday, although Wednesday and Sunday (which will become red days for BTC in the future) are also returning positive values for our new time frame.

With these results in hand (we’re still in 2018), we might have decided to trade Bitcoin only Mondays, Thursdays and Saturdays - and sell at the end of each.

Here are the results of our day strategy from January 2018 onwards:

With only past data to work with, we again outperform holding while making only a small loss of -4% compared to the -38% of holding. Not bad!

**Calculating best performing days for ANY coin**

So our analysis strongly suggests that some days of the week are more auspicious for crypto traders than others. And while BTC is a decent proxy for the whole crypto market, the weekly pattern might be different for other coins.

This is why we created a ‘daily performance’ template, where you can determine the best trading days for any coin in the Santiment database.

Download the __Sansheets plugin__ so you can import Santiment data into Google Spreadsheets, then use __this pre-made template__ to understand which days have had the best historical returns for any coin.

**Applying the ‘Days of the Week’ Strategy**

As outlined above, this strategy presents an interesting model for trading Bitcoin, comfortably outperforming HODLing when applied forward starting in 2018, by relying on previous 2 years of data.

That said, it needs mentioning that this analysis doesn’t take into account one obvious downside of any active trading strategy - transaction costs. These would certainly chip away a bit at the projected returns, narrowing the gap between the outlined strategy and our benchmark in reality. So trading the days alone will probably not give you the holy grail of trading strategies. It should rather be seen as an addition to other strategies that might give you an edge by trading more bullish on certain days of the week.

If you’re already an active trader, this information could help offset some of the inherent volatility of the crypto market, and warn you of the times when the odds tend to be stacked against you - no matter what the charts say.

*Disclaimer - THIS IS NOT FINANCIAL ADVICE. The above analysis is for informational purposes only. Do your own research and due diligence before making any investment decisions.*