Distributed cryptographic currencies
A cryptocurrency is a peer-to-peer distributed, digital currency.
For some cryptocurrencies the entire amount of available units is issued before the cryptocurrency is made public, but for some cryptocurrencies the units are created / mined in time (although some can be mined before the cryptocurrency is made public). In both cases, the software which manages the cryptocurrency can be modified by its issuer to issue more cryptocurrency later.
The most important feature of a cryptocurrency is that there is no central authority to control it, and no central authority to keep track and validate the transactions. The cryptocurrency's underlying peer-to-peer network keeps track and validates the transactions in a distributed manner.
Cryptocurrencies, unlike traditional assets, are not owned by companies or governments, so while their creators can disappear, the software and the peer-to-peer network can continue to run, and can be maintained by others. Currently, there is some centralization in most cryptocurrencies; some of them have just a few miners, some have most of their validating nodes owned by their creators, but things will decentralize more in time.
Some states want to, and will, create their own cryptocurrencies. A state created cryptocurrency is not a cryptocurrency in the practical sense because it's controlled by one entity, the state, so it's in fact digital fiat currency owned by the state, not by the people. It's not decentralized currency, it's centralized.
Ignore any cryptocurrency whose creators don't make public the source code which manages the cryptocurrency.
Most cryptocurrencies have a finite supply (= amount of available units), or have a very small inflation of the supply. Such cryptocurrencies can act as a store of value, like gold, meaning that they can preserve or increase their purchasing power over time; this means that in the future you will need to pay an equal or smaller number of units in order to buy the same thing; this is called "deflation of prices".
In contrast, a currency whose circulating supply increases in time, like fiat currency, has a decreasing purchasing power over time; this means that in the future you will need to pay a larger number of units in order to buy the same thing; this is called "inflation of prices".
Since the cryptocurrencies with the largest market capitalization have a fixed circulating supply, it means that their prices (relative to fiat currency) will increase in time, in the absence of bad news. This means that: future price = current price * future market capitalization / current market capitalization. This means that the value of the cryptocurrencies that you hold today will increase in time. Obviously, there is no guarantee that a specific cryptocurrency will survive for years or decades.
Since you can usually own and use fractions of a cryptocurrency's units, with many decimals, a high price per unit is not a problem.
In the first half of 2010, Bitcoin was worth less than 1 cent / unit. 7 years later it was worth over 1'000 USD / unit. This means an increase in price of over 100'000 times, so someone who has invested 10 USD in 2010 and kept it until 2017 became a millionaire. This kind of growth will never be achieved again, not even by far. This has happened because Bitcoin was the start of a financial revolution. As the market capitalization of a cryptocurrency raises, its potential percentual growth decreases because the total possible market capitalization is economically limited.
The World Economic Forum predicts that the market capitalization of the entire cryptocurrency industry could reach 8 trillions USD until 2027. The problem is that you don't know which cryptocurrencies will still have value then, nor which of them will grow faster than most of the others.
These cryptocurrency platforms have potential to be used as store of value, and as general purpose currencies.
Listing is by market capitalization, descending.
Bitcoin is the cryptocurrency which has started the entire industry of cryptocurrencies. Its proof of work consensus algorithm has performance issues. Bitcoin is used as a settlement currency, that is, it's the currency in which most trades are being denominated and provides the highest transactional liquidity. Its price drops the least when the entire ecosystem has a price correction. When its price drops significantly, the price of the others drop even more. When its price raises significantly, the prices of the others follow after a while.
CoinMarketCap = Statistics for cryptocurrencies.
IcoStats = Statistics for cryptocurrency ICOs.
Bitcoin Magazine = News.
Coindesk = News.
Binance = Exchange.
Bittrex = Exchange.
Kraken = Exchange. Accepts fiat.
The most fundamental principle of investment, "buy low and sell high" / "buy the dips", passes through people's minds like a ghost through the real world: with no effect.
The overwhelming majority of people start buying when they see the price raising, and start selling when they see the price lowering, the opposite of what they should logically and mathematically do, that is, they "buy high and sell low". The greater the raise, the more fiat currency they invest. Then, they complain that they are not making or are even losing money, and start searching for that magical recipe (like charts) that's going to quickly make them a lot of money, just like any other betting addict who believes that he's going to beat the market with his "system".
When a cryptocurrency's price raises, nobody (charts included) knows whether that trend will continue or the price will decrease, so you have to budget your investment.
Sadly, most people will continue to ask questions like "Is this a good time to buy?" Why is it sad? Because it's the wrong question to ask. If the answer were "yes", most people would use it as an excuse to invest a lot of money and then blame the person who has answered, for any drop of the price, even though the answer is correct in the context of a long-term investment.
It's also sad because if people ask someone with investment experience whether it's a good time to buy or not, and they are told that nobody can say what the price will do next, they will insist by asking "I understand that nobody can say what the price will do, but what is your opinion?" So, they ask for advice but ignore the actual advice and, since they only accept a "buy" or "don't buy" answer, they are trying to shift the blame for the aftereffect on someone else.
The purpose of a very long term investment strategy is to maintain, and possibly grow, the value of the investment over years or even decades.
It's important to follow an investment strategy which increases the potential profits and decreases the potential losses, balancing them, while still betting / trusting on a rise of the price in the long term.
The investment value may be maintained even if, after years, the price of the bought cryptocurrency is below the price that it had when you started investing. This is because even though you have originally bought at a higher price, you have probably also later bought at much lower prices, prices which were probably lower than the price that the cryptocurrency had when you had to sell (meaning that you have both profits and losses).
Some people say that you should invest a lot of money when the price is low. The problem is that you can never know if a price will raise or drop after you start investing. Even for an emergent market, while you can bet that it will keep on growing, you can't know for sure, it's just a bet.
This is why investing all the money that you have on a single transaction is a sure way to lose money, and why you have to follow an investment strategy.
When the price drops dramatically, like half in a few days, when everybody panics and starts selling, in the absence of worse news, treat this as a buying opportunity because the buyers are inclined to accept the higher prices easier in the future, since they were recently in that area.
State regulation is not bad news. Some investors perceive state regulation to be bad, but in fact state regulation brings stability and predictability for big investors and for businesses. It's important to note that state regulation and banning are different things; regulation means the acceptance and the ordering of the business environment, banning means not accepting and forbidding business. When a rumor about impending state regulation (not banning) causes a dramatic drop of the price, as it usually does, treat that as a good opportunity to buy, but keep in mind that you don't know when the drop will stop.
Invest in emergent markets, if the business model seems sound, because the growth potential (in percents) is greater than in mature markets, and once an emergent market becomes mature, the investment value will have grown much more than it could have grown in a mature market.
For example a cryptocurrency with a market capitalization of 1 billion (at investment time) requires 99 more billions in order to grow your investment 100 times, however, a cryptocurrency with a market capitalization of 10 billions requires 990 more billions in order to grow your investment 100 times, and obviously, it's far easier for a cryptocurrency to grow with 99 billions than 990 billions. Of course, a small cryptocurrency is also riskier for investments since it could disappear easier than a large cryptocurrency, so you should invest only a small amount of fiat currency in one.
However, it's important to understand that the price doesn't depend on the amount of fiatcurrency which gets into the cryptocurrency, it depends on the direction in which sellers are willing to provide liquidity, and on the direction in which buyers are willing to consume the available liquidity.
A market capitalization X USD doesn't mean that an amount of X USD has changed hands from trader to trader, in exchange for cryptocurrency. An amount of USD that's several times smaller has changed hands, but the desire of the traders to consume the available liquidity has driven the price up. This happens with any asset which is in a limited supply, like it's the case in the equity / stock market.
A similar capitalization inflation happens with the supply of fiatcurrency, where the total supply is several times larger (globally, about 5 times) than the amount of available cash, but here the supply is increased by the state through the central bank, not through a free market.
Some people say that cryptocurrencies are in a bubble, meaning that they are valued too high and the bubble will pop. Say this is true and the prices of all the cryptocurrencies drop to 10% of what they are now. What does that mean? It means that it's a buying opportunity because the buyers are inclined to accept the higher prices easier in the future, since they were recently in that area. Even if the price were to go to virtually 0, so long as it goes back up the value of your investment is preserved, except if you do margin / leveraged trading or you are manually using stop-losses. Why would the price recover? Because cryptocurrencies (unlike traditional assets) are not owned by companies, so while their creators can disappear, the software can continue to run and can be modified by others. In the absence of a fundamental cause that can bring cryptocurrencies to a 0 valuation, the prices should recover.
The profits and losses, expressed in fiat currency, of a cryptocurrency are realized only when the cryptocurrency is sold in exchange for fiat currency. Until then, its value is like Schrödinger's cat, in multiple states at the same time. While the cryptocurrency is kept, its price could change dramatically during the next few days, weeks, months or years. This means that you should not despair when your investment's value is lower than originally, nor should you gloat when it's higher than originally.
While a drop of the price of a cryptocurrency A (that you own) decreases your investment's value, if you believe that the price of a cryptocurrency B will raise before the price of the cryptocurrency A raises, you can move your investment there without a loss. Even if you do sell, if you later buy back at the sell price then it's as if you haven't sold.
In metaphorical terms, if an elevator gets you down a few floors, you can take back up any other elevator, not necessarily the same. Still, you don't know in which direction any elevator goes, so randomly jumping from one to another might in fact bring you all the way down to the ground floor. In fact, you will most likely jump to a cryptocurrency which had a recent price raise, so you think it will continue to raise without limits, but that's a bad moment to jump to it.
Ultimately, even if you were to buy at the top price every time, you could still make a profit, like in this example.
Invest in several cryptocurrencies because nobody knows which of the current cryptocurrencies will still be around in a few years, or how much will each grow. Diversification lowers the potential losses and lets you profit from the growth of several cryptocurrencies.
How do you decide which cryptocurrency to buy? A cryptocurrency with a small market capitalization can grow percentually much more than one with a large market capitalization, but it can also disappear much easier. A high volume per market capitalization ratio indicates a cryptocurrency which is of more interest than one with a small ratio. Never take a buy decision based the price, on the number of units that you can buy, or on the total circulating number of units; they are irrelevant as indicators for potential growth.
Look at the cryptocurrency's team, at its (real world) connections, at what problems it's trying to solve, at its potential technical performance, at the geopolitical context. Is the cryptocurrency an internationalized effort? Is the emerging cryptocurrency designed for a market which is difficult to enter by an established cryptocurrency?
Don't waste your time and energy trying to predict, intuit or guess future prices. Money is made with statistics not gut feelings.
If you want to invest based on rumors and news, realize that the people with lots of money, the market movers, will find out the rumors and news before the vast majority of people, and will buy or sell before the rest, leaving the others to buy at the top or sell at the bottom. Then, they'll do the opposite (sell or buy) before the others realize what's happening, leaving everyone else trapped at a loss-producing price.
Never do margin / leveraged trading; this is reserved for trading in traditional markets, it's neither for trading cryptocurrencies (because the volatility is already high), nor for investments.
Sell only when it's necessary, like when you want to move to another cryptocurrency, or you want to buy a home.
Selling a cryptocurrency increases the potential losses because the price could keep on raising, and if you were to later buy at a higher price, you would lose the difference between the sell and the buy, compared to what would have happened had you not sold.
Traders always try to make a profit before others can, and, therefore, the future can't be predicted by the past. The price, the volume, the order book, the indicators, the overlays, they all show what was and what is, but never, ever, can they show what will be.
Profit is not made with predictions since there is absolutely no way to predict the future price, it's made with a statistical advantage over a long time, and this advantage is determined through simulations.
Technical analysis can only describe what has happened in the past. When it's trying to predict the future, technical analysis is like astrology.
Technical analysis fits the patterns to the data, that is, it works for the past because the analyst changes the described pattern to fit the analyzed time frame.
If technical analysis tells you that in the future the market will go up... unless it goes down, it's useless. To be useful, it must tell you only one of the following signals: wait, buy, or sell.
Why does TA appears to work sometimes? Because it's designed to show the lows and highs (on a chosen time frame), that is, it's meant to tell the traders to "buy low and sell high", not to predict the future. In order to make a significant, sustainable profit, the exact algorithm must be simulated with past data, and then used for the future in the hope that the future will be similar enough, even though not identical.
There are lots of recognizable patterns on charts. The mass behavior of people creates patterns. The problem is that all these patterns are transformed continuously, that is, they are scaled, sheared and rotated in unpredictable ways. Recognizing them afterwards is utterly useless, and in fact it's dangerous because it makes you think that recognizing them is useful in making a profit. It's like seeing shapes in coffee grounds; sure, you can see shapes in coffee grounds, but they have absolutely no effect on the future.
To make a profit, the investor must buy low and sell high using a variety of tools which help determine the prices for the buys and sells in a way which increases the probability to win. These tools are: an investment strategy, money and risk management, statistical simulations, gathering and interpreting the rumors and news (even using machine learning), and understanding the mass behavior of investors.
For example, in technical analysis there is a pattern called "pole and pennant" which is supposed to predict that a significant change in price will follow. This pattern looks like a triangle with the base on the left side and a vertex on the right side. This triangle is the result of mass behavior, that is, it's the effect of a cause. A significant change in price does follow, but not because of a geometric shape on a chart.
The reason why this triangle appears on the chart is that, firstly, the price moves in cycles, meaning that significant changes (up or down) are followed by small changes (called "consolidation"), then followed by significant changes, and so on. Secondly, in a period of consolidation a lot of cryptocurrency is bought at continuously increasing prices (= with an upward slope) and sold at continuously decreasing prices (= with a downward slope) , with the buys and sells centered around the middle of the base of the triangle. Once the price flatlines and the triangle gets its right vertex, meaning that profit can no longer be made with small price changes, a significant price change follows (possibly only after a while). The direction of this change is determined by fundamental causes, not by geometric shapes, and this direction is the only thing that matters when trying to make a profit.
This article does NOT constitute investment advice!
Investing money is risky: you might lose all the invested money!
Never invest more money than what you can afford to lose!
This algorithm is specifically designed for cryptocurrency markets, which have a very high volatility, not for traditional markets (which require margin / leveraged trading because their volatility is too small).
This algorithm considers that the cryptocurrency prices are expressed in fiat currency, prices which in the long term should be raising.
This algorithm is just an example that you can build on. All the values specified here have to be adjusted for each cryptocurrency, to the real context, possibly often since the cryptocurrency markets are very dynamic.
No matter how safe your investment strategy is supposed to be, you have to be prepared for recession periods when there is a huge price drop followed by months or years of prices which are much lower than those at which you have invested. If external circumstances force you to sell cryptocurrency during a recession, before the price recovers, you will have to be able to support the incurred losses. This is why you should never invest more money than what you can afford to lose.
The most important tools that you can use to balance the potential profits and losses are budgeting (= splitting your budget among your buys) and buying during price dips.
The algorithm described below minimizes the possibility of using your entire investment budget to buy cryptocurrency at what might be the highest price, while at the same time allowing you to profit if the price keeps on raising.
As the market capitalization of a cryptocurrency raises, its volatility decreases, so this algorithm might not work anymore when the changes of the market capitalization become small.
Put all the fiat currency that you want to invest in something called "investment budget", IB. Every time you receive your regular income, put 5% of it in the IB.
Plan in advance to split your IB in several parts that you can later use to set up buy orders at different moments, for different cryptocurrencies that you want to invest in, proportional with how much you think each cryptocurrency's price will raise in the long term.
Let P be the current price of the cryptocurrency that you want to invest in.
The general idea is to set up buy orders starting from price P and going down to 0, for each order using a part of the IB which is reserved for the cryptocurrency.
You should set up a buy order at price P because you don't know whether the price will move up or down from that moment on, but the important orders are the ones which are set up below P.
You should set up buy orders at prices lower than P because the price might decrease (you don't know how much), trigger the buy orders as it reaches them, and then might increase (you don't know how much, but hopefully above P). During recessions, this could happen even years later. In a way, this rewinds the time back into the past when the price was much smaller than its recent high.
Setting up in advance the buy orders below price P ensures that you don't miss the buying opportunity (like during sleep).
The more parts you split the IB in, the lower the potential losses (and profits) are. With all these parts of the IB you would have to set up an equal number of buy orders either at smaller differences from one another, or at lower prices.
The currency with which you set up an order is locked and is unusable for other orders until you close the locking order.
A fixed amount of fiat currency, like the 5% of your monthly income, can buy a decreasing number of units of cryptocurrency as its price increases, and an increasing number of units of cryptocurrency as its price decreases.
Many people will not buy when the price decreases because they fear that it will continue to do so. But you have to ask yourself why would the price continue to decrease and not recover in the long term? Bad news are a reason, fear is not.
People always find justifications about why they should buy a cryptocurrency whose price is raising. They'll tell themselves that this time will work (they feel it), that they can see the trend, that this time there is good news, that this has worked in the past, that they know that this cryptocurrency is not like the others and has a history of rebounding quicker than the rest, that this time they will score big to compensate for all the losses, and, finally, that they've heard (on the Internet) of people who do this and make a lot of money. The thoughts of a betting addict.
Some people compare buying when the price decreases with catching a falling knife. This comparison is fundamentally flawed because the knife is drawn by the ever present gravitational force, but the price of a cryptocurrency bounces around (unless there are long lasting bad news).
If the price experiences a very large increase in a very short time frame, increase which is called "parabolic raise", this will be followed by a correction which can easily put the price back at the level from which the parabolic raise has started, so don't get greedy in such moments, don't put all your IB in the cryptocurrency.
Set up buy orders
When there is fiat currency in the IB, pick a cryptocurrency that you want to invest in.
Let P be the cryptocurrency's current price.
Let H be the highest price from the past 2 months. When trying to determine H, exclude any spikes which appear to be technical anomalies, that is, which shoot out of the normal price range.
It may be best to pick a cryptocurrency in which to invest only if it's currently in a significant dip, like when P <= 0.7 * H. This technique is called "buying the dip".
Calculate a set of potential buy prices, using the following equations: 0.7 * H, 0.4 * H, 0.2 * H, 0.1 * H, 0.05 * H, and so on.
Split the IB reserved for the picked cryptocurrency in 3 equal parts and, using a part for each order, set up buy orders at price P and at the first 2 calculated prices which are below P.
Every time H increases visibly (like 5%), close all the buy orders and use the unlocked fiat currency to set up new buy orders at the calculated prices (not at price P). This technique is called "trailing the top". Using this technique will trigger at some point, with a high probability, the buy orders which were set up at the calculated prices, at least the one with the highest buy price.
Always use limit orders because these will be executed at the specified price or one which is more profitable for you. This is very important in a market with a limited liquidity, a market in which your orders may need a long time to be executed.
Here is a simple example of why it's important to budget, to buy as low as possible, and to still bet that the price will recover, at least in part, after a significant drop.
Let's say that you invest the amount of fiat currency F1 = F to buy the amount of cryptocurrency C1 = C at price P1 = P. This means that C1 = F1 / P1 = F / P = C, which means that F = P * C.
Later, the price decreases to P2 = P / 4, so you invest the same amount of fiat currency F2 = F to buy the amount of cryptocurrency C2 = 4 * C. This means that C2 = F2 / P2 = F / (P / 4) = F / P * 4 = 4 * C.
Then, let's say that external circumstances force you to sell all the cryptocurrency, although the price is only P3 = P / 2. This means that you get an amount of fiat currency F3 = P3 * (C1 + C2).
How much fiat currency did you gain or lost, compared to your investment?
That is R = F3 - F1 - F2 = P3 * (C1 + C2) - F1 - F2 = (P / 2) * (C + 4 * C) - (F) - (F) = (P / 2) * C * 5 - 2 * P * C = P * C * 5 / 2 - 2 * P * C = F * (5 / 2 - 2) = F / 2. This is a profit of 25% from the total investment (both times), a pretty good outcome.
If P3 = P * K3 then R = F * (5 * K3 - 2).
You would break even (that is, R = 0) if when you had to sell you had K3 = 2 / 5, so the price would P3 = P * 0.4. Had you only bought at price P both times, at sell time you would have R = - 1.2 * F, a rather bad outcome. This is a loss of 60% from the total investment (both times), a rather bad outcome.
Trading can increase the potential profits (compared to investing), but at the same time increases the potential risks.
Trading is like a giant game of poker, that is, the participants try to make a profit in a zero-sum game. There is a slow increase of the total value of the markets which have a limited number of shares, because the economy increases in time, just not enough to matter for trading, only for long term investing.
In the same time frame, it's likely that you could make the same amount of fiat currency, or more (especially if you buy during dips), by simply holding your original investment in the cryptocurrency. This way you would not have to go through the effort and stress that trading requires, and you would not risk losing everything due to a bad algorithm or parameters. Trading can become interesting once you add leverage into the algorithm, since the potential profit (and potential loss) will be multiplied compared to the investment strategy. Note that due to the very high volatility of cryptocurrencies, leverage is much riskier than it is in the classical markets.
A trader must be very determined to succeed. You might think that you are too, but you are competing against the world's best traders, you are competing against organizations which get the news before they become public, and which simulate and execute their trading with farms of computers. How good are your chances? Some informal research says that, in the traditional markets, 1 in 30 (male) professional traders can make a profit that will keep them in the business for many years. That's 1 in 30 people who try trading professionally, not of all people, not of home traders! So maybe stick to investing?
Trading requires the trader to "buy low and sell high". Don't interpret this as "buy the bottom, sell the top"; you don't need to know where either the bottom or the top are.
Beginners ask themselves what this means in practice, they ask how they can possibly know when the "market turns", that is, when the price stops decreasing and starts increasing. The answer is simple: you don't know, nobody knows. This is the wrong question to ask, a dangerous question which will keep you stuck in the belief that you have to know for sure.
The correct question is: how do you make a profit? To make a profit you don't need to know the exact movements of the price, you only need a statistical advantage. You need the probability to be in your favor. The winning trades must bring more profit than the losing trades bring losses, so that they can cover the stop-losses, the exchange's fees, and the price slippage (because the available liquidity might not cover your orders, in their entirety, at precise prices).
You can use any algorithm which signals you to buy and sell, but, as detailed below, you must simulate the algorithm's result.
If you buy low and sell high, the amount of fiat currency you have is multiplied (compared to the original) with the sell price divided by the buy price. If you repeat the process including the profit, the profit is compounded in time.
Compounding is limited by the liquidity which is available during the execution of the orders, so don't image that you can (for example) compound 10% at infinity and become a millionaire in no time. The liquidity can be inferred (with good precision) from the volume. This means that beyond a certain amount of currency, the trading budget must be split among several orders and executed at different times, and even for different cryptocurrencies; even so, there still are limits.
A trade's closing price can be a combination of:
Making a profit from trading depends on the price going both down and up. It is however irrelevant when this happens, it's irrelevant if the price is trending up or down, it only matters if you find opportunities when you can buy low and sell high, no matter what the price is.
Here is an example. Let's say that you have an amount of fiat currency F1. You wait for the price to drop to half from the maximum price from the past 2 months. Then you buy. Then you wait for the price to double, and sell. At this point you have amount of fiat currency F2 = 2 * F1. After 3 such trades you have F4 = 8 * F1.
How do you chose your trading algorithm? You simulate each potential algorithm, with various parameters, on the past prices for the chosen cryptocurrency, whether manually or in software. The more detailed the simulation is, the higher the profit can be, and the smaller the risk can be. You only need to know the open, high, low and close prices of each candle. Also, the volume is useful to get an idea about how much you can trade during a candle.
It's best to simulate in software because once you implement your algorithm, you can instantly simulate it over any period of time, with any set of parameters, for any cryptocurrency. You can also continuously simulate the algorithm using new data, to see if the parameters have to be changed because the market has changed its structure. You can even implement an optimizer which finds the most profitable simulation; do keep in mind that you have to strive for repeatability, not profitability.
You must simulate your trading algorithm! There is absolutely no way around this. Slight variations can mean the difference between ruin and richness, and an algorithm which works for one cryptocurrency might not work for another. Any strategy that works in simulations is a potentially profitable way to trade, but a simulation can only give you the optimal algorithm for the past. A simulation's result is no indication of what the used algorithm will do in the future.
If you find an algorithm which shows an enormous potential profit, you should not gloat. You have just overfitted the solution to the data, so the algorithm would not yield a similar profit on other time frames. You could, in theory, find an equation which describes every ebb and flow of the price, for a given time frame in the past, equation which shows an astronomic potential profit, but will the price movements be exactly repeated in the future? Absolutely not, so this equation will bring, most likely, zero profit in the future, and possibly ruin you. The best algorithm is not the one which shows the highest potential profit, but the one with the most repeatable pattern (which you can't know in advance).
One you simulate the algorithm on a specific time frame and get a good profit, simulate it on several other time frames which are close to the initial time frame; they have to be close because it's presumed that the market has the same characteristics for nearby time frames, but could have other characteristics for time frames which are farther in time.
The simpler the pattern you follow is, the better it will fit across multiple very different types of price movement, even though the profit will be modest. There will always be price drops followed by price raises, but the exact same ebbs and flows of the price will never be repeated. So, strive for repeatability, not profitability.
The shorter the trading time frame is, meaning the time between the opening and closing of an order, the faster and more reliable the Internet connection has to be. Retail traders should not try trading on time frames shorter than 15 minutes because they would be unable to compete with trading farms, and the resulting price slippage could lead to ruin.
Price movements are trend-based, not random. You might think that this is a good thing, that you need trends to make a profit, but random movements actually create more trading opportunities, and therefore increase the potential profits compared to the case of trend-based movements.
While artificial intelligence can be used to find profitable solutions which elude humans, it normally works when there is a target to find, that is, when it's possible to iteratively improve on potential solutions until an optimum solution is found. However, trading is a moving target, a target which tries to evade whoever or whatever attempts to catch it, a target which plays poker with its participants. Artificial intelligence works best when it has access to and can interpret news before they are released to the public.
Stop-losses are extremely expensive and can quickly decrease your budget to a fraction of the original one. For example, 3 stop-losses of 20% will bring your budget down to 50% from the original one; it's presumed that margin / leveraged trading isn't used.
Do you need stop-losses? If you were to do margin / leveraged trading, the stop-losses would be automatic. If you trade on short time frames, you need them; without them, you would periodically end up unable to trade for a long time, and unable to sell (for a profit) because the sell price would be well below the buy price. If you invest on the long term, if you are sure that you can endure years of recession without selling cryptocurrency, if the cryptocurrency will still be around in years, so long as the price is going to continue to raise (because the industry grows), stop-losses are not mandatory.
To minimize the probability of losing your entire budget in stop-losses, you have to split your budget in several parts and only risk a single part in a trade, at any given time. For example, split your trading budget in 3 parts. If you have a 20% stop-loss (because the volatility is high) and you use no margin / leveraged, after 12 consecutive stop-losses (equally distributed among the budget parts) your budget becomes 40% of the original. The more careful you choose when to trade, the smaller is the probability for a stop-loss to occur.
To compensate for possible changes in the market behavior over the long term, you can use several different algorithms which yield a profit in simulations.
Implementation details are far more important than what algorithm you use for entering / buying and exiting / selling, especially the details related to how you close the trade.
When you simulate a trading algorithm on past data, you have to be very careful about how you use the data from the moment of each simulated trading signal (which tells you to buy or sell).
The simulation algorithm has to iterate through all the candles, from the oldest to the newest, and for each candle (let's call it the reference candle) it has to go back a number of candles in order to gather the information required to compute the trading signals (which tell you to buy or sell during the reference candle).
A simulation will use the OHLCV (open, high, low, close, volume) of the candles. However because in real time trading the newest / current candle is being developed, its proper values (other than O) are unknown, so they can't be used until the candle is complete, which means that you can only buy and sell at the candle's closing price. It would be safe to use the values of the reference candle, but execute all the simulated buys and sells at the candle's closing price.
If a simulation shows a very high profit, it's most likely that you have used the values of the reference candle that you can't know during the reference candle.
Be pessimistic about your algorithm and presume that the buying price is going to be less favorable to you for most of the time.
It's best to display your trading signals on a chart, so that you can see how they all cluster, and how the losing trades cluster. If there is significant clustering, the repeatability of your algorithm is low, so you will likely not make money with it, and you might even lose money.
Leveraged trading allows you to amplify the potential profit and potential loss by a number, and is normally used in markets where the small volatility doesn't provide enough profit potential.
More interestingly, leveraged trading allows you to make a profit when the price of a cryptocurrency decreases, by short short selling the cryptocurrency.
Basically, when the price of the cryptocurrency is P1, you borrow (from the broker) cryptocurrency CC1 with which you buy fiatcurrency FC1 = CC1 * P1, hoping that the cryptocurrency's price will decrease.
The prices are expressed as P units of fiatcurrency for 1 unit of cryptocurrency (fiatcurrency / cryptocurrency), written as BTCUSD (so the opposite of the division).
In order to get this loan, you need to reserve a collateral deposit (called margin) at the broker, in the shorted cryptocurrency CCOL = CC1 / L, where L is the leverage that you are using. L is an integer number, at least 2.
When the price of the cryptocurrency decreases to P2, where P2 < P1, you buy cryptocurrency CC2 = FC1 / P2 = CC1 * P1 / P2 with the fiatcurrency that you have obtained from the earlier sell (FC1).
However, you buy only enough to cover the loan from the broker, so only CC1 = FC2 / P2. This means that you only have to sell fiatcurrency FC2 = CC1 * P2.
The rest of the fiatcurrency is your profit FCP = FC1 - FC2 = CC1 * P1 - CC1 * P2 = CC1 * (P1 - P2) = CCOL * L * (P1 - P2).
Keep in mind that your margin must be able to cover, at all times, your loss, so CCOL * (1 - MLL) >= LOS. LOS = CC1 - FC1 / Px = CC1 - CC1 * P1 / Px = CC1 * (1 - P1 / Px), where Px > P1. MLL is called margin liquidation level and it usually is around 0.4...0.5, but can be even 0.
Brokers automatically close a short position before its loss reaches the margin, in order to maximize your ability to sell the fiatcurrency that you've bought, and buy the cryptocurrency that you've borrowed, ability which may be hindered by a limited liquidity when the price increases quickly.
The broker will automatically close your short position if the cryptocurrency's price Px >= P1 / (1 - (1 - MLL) / L).
If you want more information on this subject, search the Internet for "backtest trading software", "algorithmic trading software" or "algorithmic trading platform".
If you were to build your own simulation software, you would implement and know every detail of the simulation process, and you would be able to keep your algorithm private.
Quant Connect (integrates with GDax)