The Nobel price laureate physicist Niels Bohr once said, “Prediction is very difficult, especially if it’s about the future.”
This quote captures the reality in the markets rather accurately.
Is Bitcoin breaking through the 50k mark this year? Or even the 100k mark? Will Ethereum outperform Solana between now and the end of the year? Will Bitcoin close higher or lower by the end of the week?
If one could make correct predictions reliably, one could get filthy rich in no time. But directional price predictions for stocks, currencies, or cryptos are seldom right. Even professional analysts get it wrong over and over again.
A helpful trick is to base trading on leading signals. Any information out there that had some predictive quality in the past – whether it is fundamentals, technical patterns, or any other piece of data – may be the building blocks for a successful trading strategy.
Besides that, there’s a less-known area where the financial math wizards, commonly known as quants, come into play. These guys develop complicated pricing models that people without a PhD in maths or physics struggle to understand.
But don’t worry, even top Hedge Fund managers (maybe besides Jim Simons) don’t do the maths themselves. They spend money for such analysis by hiring the right talent or by straightforward buying analysis from third-party research firms.
The good news is that the playing field in finance is more even than ever. What until recently was out of reach for people without a personal Bloomberg Terminal can nowadays be obtained by anyone.
The “Crypto Asset Price Forecasts and Price Driver Analysis” dataset by the Dutch crypto experts Blockchain Investments & Co. (data for sale here) is an example of quant-grade alternative data. But is it any good? Let’s find out.
These price forecasts use a method called Holt-Winters, named after their two inventors and published in the 1960s. In a nutshell, the Holt-Winters model applies a triple exponential smoothing to a time series of observations. It considers an average, a trend, and a cyclical (or seasonal) element. Initially, Winters used the model to forecast sales, but in reality, it can be used for anything from room temperature to asset prices.
The mentioned dataset provides price forecasts up to one week ahead. It also gives confidence intervals and classifies the price drivers according to the Holt-Winters model. The next chart shows the forecast and actual price for Bitcoin since 2021.
One could think of several ways to incorporate the price forecasts into trading. We impose the following simple rule:
- Every week on Monday, if the 1-week forecast is above the current price, buy the cryptocurrency. Stay out of the market otherwise.
This we call the Forecast Strategy (“FC Strategy”).
The six coins with the largest market caps (exculusing stable coins) as of January 2016 will be used: Bitcoin (BTC), Ripple (XRP), Litecoin (LTC), Ethereum (ETH), Dash (DASH), and Dogecoin (DOGE).
How did it do?
The strategy’s results are awful in all six coins. Only in Bitcoin the long-only variant of the strategy could come close to keeping up with a holder’s performance (Buy-and-Hold).
But why is that? Well, the forecasts miss the required precision. Actually, in only 51.5% of all weekly estimates, the direction was correct (in Bitcoin, while in some other coins, it was even below 50%). That low accuracy is unacceptable for a long-short strategy. Also, it will cause one to miss tons of opportunities in a long-only approach.
No strategy can match the returns of holding. Or is there a way of trading successfully?
An Easy Cryto Strategy
Strategies based on moving averages are among the most successful in amplifying returns and controlling risks. When using a single simple moving average (SMA), one buys if the price is above the SMA and sells if the price is below the SMA.
The main reason why this (often) works is momentum. And momentum can be found in most assets, in stocks, bonds, commodities – and cryptos.
This time the rule will be:
- Every week on Monday, if the price (from Sunday) is above the 30-day SMA, buy the cryptocurrency. Sell short otherwise.
As we can see, the SMA strategy outperformed the buy-and-hold strategy in all of the six coins between 2016 and 2022. In Etherium, one could 20-thousand-X the initial investment. An incredible number.
Of course, in retrospect, that’s easy to claim. But the fact that momentum is a thing in all markets makes the strategy more robust. Better, definitely, than the quants’ forecasts that we saw before.
To give a fair trial of the forecast data by Blockchain Investments & Co., we also assessed the predictions in a linear model setup. The forecasts are insignificant on every common level of significance. The current price is generally the best estimate for tomorrow’s (or next week’s) price.
And that unfortunately, is not enough to build a robust trading strategy with.
We hoped to find better results. But as seen with the SMA – a really simple strategy relying on trends – anyone with enough dedication can learn to create successful crypto strategies crushing performance of the hodlers. Check out our courses for more details on successful trading strategies.