How to predict cryptocurrency prices: All about algorithms
How do people work out their crypto price forecasts?
Contents
- What are algorithms and what are they used for?
- Who uses algorithms?
- How to predict cryptocurrency prices
- FAQs
If you want to predict cryptocurrency prices, there are lots of places that give their opinions. If, on the other hand, you want to know how to predict cryptocurrency prices, that is somewhat trickier. Whether it is a person using the data available to them to make a crypto price prediction, or whether it is the power of technology that brings you forecasts, it’s a concept well worth taking a look at if we are considering how to predict cryptocurrency trends and prices.
What are algorithms and what are they used for?
While the majority of publicly-traded companies’ share price predictions are often made by market experts, which means that a stock prediction algorithm is not as much of a go-to tool as it otherwise could be, cryptocurrency is a newer thing. Many cryptos have not really come across the radar of the stock experts. This means that, while you will find analysts’ price forecasts for bitcoin and, occasionally the likes of ETH or the major altcoins, the overwhelming majority of crypto price forecasts are based on a computer program called an algorithm.
An algorithm is something that allows something to do something automatically. For instance, social media websites often use algorithms to decide what things will appear on our feed. Likewise, a lot of internet organsations aim to crack the apparently ever-changing code that puts things higher up in Google searches than others. Algorithms can also be used for other methods, though, especially when it comes to making price forecasts.
We do need to point out here that predictions, whether they are made by a human being or by a machine, are very often wrong. For a stock prediction algorithm, a company could introduce a new product or service – or remove one from the market – that the code fails to predict. When it comes to a cryptocurrency prediction algorithm, it has to deal with an incredibly volatile commodity and a market that is the subject to forces including, but not limited to, regulation, changes in availability and the overwhelming dominance of bitcoin. Let’s not forget that when BTC does well, the market does well, and when BTC does badly, the market performs poorly.
Who uses algorithms?
Plenty of sites give out price predictions based on algorithms. We asked Gov Capital, PricePrediction.net, DigitalCoinPrice and CryptoNewsZ – among others – how their algorithms worked, but they did not immediately respond to our request for comment.
However, WalletInvestor does give some clues, noting that it is always trying to improve its algorithm, writing: “Machine learning is a type of artificial intelligence (AI) that allows software applications to become more accurate in predicting outcomes without being explicitly programmed. The basic premise of machine learning is to build algorithms that can receive input data and use statistical analysis to predict an output value within an acceptable range.
“The cryptocurrency, stock, commodity, fund, and forex rates are influenced by many things: economic news, trader opinions, natural disasters, wars, investor groups, and so on. It is, therefore, very difficult to give an accurate estimate of the future. WalletInvestor's cryptocurrency and other forecasts are based on changes in the exchange rates, trade volumes, volatilities of the past period and other important economic aspects.
“The accuracy of the prediction depends on the quantity and the quality of the data, so it is also difficult to anticipate anything in the case of newer cryptocurrencies.”
There are some prediction websites, though, that do appear to show how they work out their crypto price predictions. For instance, CoinCodex has a range of forecasts based on how much certain other parts of the tech world are expected to grow. The site makes predictions based on Google, Facebook, mobile phone usage and the growth of the internet in general.
The site also uses a range of technical analytic tools – such as the daily simple moving average, daily exponential moving average, weekly simple moving average and weekly exponential moving average – to make predictions. That said, it is worth pointing out that the site does hedge its bets somewhat, giving a range of prices for each indicator based on how long they are used for. Also, the site’s forecasts based on its crypto prediction algorithm do not go that far into the future, looking forward just one week.
Nevertheless, CoinCodex does give some hints and advice, answering a question about how to predict cryptocurrency prices on its dogecoin prediction page by saying: “You can predict cryptocurrency prices by considering both on-chain and off_chain metrics, as well as technical indicators that take into account dogecoin’s past price performance. Another tool you can use is to gauge the market sentiment to see whether investors are optimistic or pessimistic about dogecoin. Keep in mind that there is no way to predict the future with 100% success, and past performance is no guarantee of future results.”
How to predict cryptocurrency prices
So, how about if you want to create your own cryptocurrency price prediction algorithm? That is certainly possible, although you will need a not inconsiderable amount of technological knowhow. For instance, AI expert Venelin Valkov has created his own system to predict the price of bitcoin. Put relatively simply, he has taken 3,201 previous BTC prices and sorted them by time, split into equal intervals. Then, he says, he uses something called a deep neural network. This is – again, put extremely simply – a program that operates on several layers. The idea behind it is for the network to recognise past patterns and then come up with its own prediction based on those patterns.
Anyway, Valkov suggests using the network, which in effect you make yourself, to create your own bitcoin price forecast. Theoretically, this should also work for other cryptocurrencies. This does assume, of course, that you have the advanced computing skills that you will need to create the algorithm in which to feed the relevant data. Making a cryptocurrency prediction algorithm is a complex task and one which requires a certain amount of skill.
Valkov’s method is not the only one out there, however. As far back as 2014, computer scientists Patrick Jaquart, David Dann and Christof Weinhardt teamed up to carry out research, which found that two particular computing tasks, called recurrent neural networks and gradient boosting classifiers, were, in their opinion, best suited to cryptocurrency predictions.
Again, though, you will need a pretty high level of expertise to make the system work for you – and it is also worth pointing out that the research carried out by Jaquary, Dann and Weidhardt covered a prediction that looked forwards nine months. This, in itself, might just make you a bit cautious about taking crypto price forecasts that cover a longer period at face value.
So there we are. Let’s remember that price forecasts, whether human made or AI-generated, can very often be wrong. As with all things, if you want to make a crypto price prediction and then apply it, you will need to do your own research, remember that prices can go down as well as up and never invest more money than you can afford to lose.
FAQs
Is it possible to predict cryptocurrency prices?
It is, although how accurate these predictions are is open to some doubt. It is always possible that something can happen to the crypto market that no one has foreseen, which can send prices lifting into the stratosphere – or tumbling to the depths.
Are crypto price predictions accurate?
Very often, they are not. We don’t know the future and it would be wrong to think that we can know exactly what is going to happen. Crypto predictions – like pretty much all financial predictions – are very often wrong, especially as they go further and further into the future.
Which algorithm is best for crypto prediction?
This is a question that, honestly, we cannot answer. You will need to do your own research and come up with your own decision when it comes to which algorithm, if any, you should trust.
