Moving averages and Bitcoin

A study of trend following BTC using simple moving averages

An Epic Run

Bitcoin started 2017 at a price of 963. Today, 4 and a half years later, it is at 31,800, a gain of 33x. To put this in perspective, Amazon went from 757 to 3573 during that same period, about a 5x gain. There is simply no doubt that Bitcoin has been the top performing asset class of the last decade.

But with this extreme performance, comes extreme volatility. Amazon stock went down 30% from its peak in 2018, but overall the ride was a smooth ascent. Bitcoin dropped 80% peak to trough in that same year, and again 50% so far this year.

So the question arises: can moving averages help you decide when to jump on the train and when to jump off? The answer is mixed: in the short term (typically) yes, but in the long term HODLing still outperforms.

The idea of Trend following using Moving Averages

Moving averages are the core of most trend following techniques. If Bitcoin is lower than it’s 20 day moving average it is falling. If it’s higher than the moving average, then it is rising. The rule is to wait until you get the confirmation that BTC is rising, buy, and then sell when you get the confirmation that the trend has reversed.

Unfortunately, as can be seen in the above graph, this doesn’t always work. In 2021, so far, you would be up 8% by just buying and holding BTC. But if you had used the MA to time the market, you would be down a massive 46%. The 20 day moving average, which was supposed to protect you on the downside, actually did the reverse, causing massive damage from March 1 on.

Impact of Trading Fees

A big chunk of the underperformance is trading fees. In 2021 alone, the strategy generated 28 trades. At a cost of 30 bp (typical for a pro trader), that is an 8% headwind. Trade on Coinbase or similar platform, and you are looking at 20% just in fees.

2019 and 2020

Looking back a few years, we see slight underperformance of the strategy in both 2019 and 2020. Overall, the strategy does not do much — it just tracks Bitcoin in those years (and is less tax efficient, as we will discuss later)

2018 — the one year it really worked

The one year where the MA trading strategy definitely did work was 2018. You still ended up losing money (10%) but you survived the carnage of the 70% drop in BTC.

The performance of the 20MA strategy in 2018 really

2017, similar to 2019 and 2020

The real massive bull market year for Bitcoin was 2017. In that year, the market was mainly trending, and the MA strategy tracked BTC pretty closely.

The impact of taxes.

Because trend following trades so frequently, it results in all trading profits for each year being taxed at ordinary income. Which means that what ever is left after bad signals and trading fees gets sliced in two.

Compare what happens to a buy and holder who held thru today, and just cashed out versus the trader who held throughout the entire period.

Amazingly, the HODLR still beats the Trader after 5 years, even with the trader completely skipping the carnage of 2018!

What about the 10 day Moving average?

A good question is what if we used the 10 day Moving average instead of the 20 day? To look at that in detail, consider 2018

In this case, the MA strategy now loses 24% versus 10%. A lot of this is increased trading fees: the number of trades doubles in the year from 27 to 62. In addition, the 10 MA strategy gets more false signals, and is in the market for longer periods in this epic bear market.


Buying and holding BTC has outperformed simple trading with moving averages in every year except 2018. And even in 2018, you would need to use a 20 day MA and not a 10 day to really avoid the bear market. After taxes, it’s really not a great tool to play BTC.