Profits cannot be boosted by a method alone
The title of this article uses a phrase that appears often in trading marketing: boosting profits. It is worth saying plainly that no chart method can promise a gain. Prices are uncertain, results vary from period to period, and many people who trade cryptocurrency lose money, sometimes all of it.
What technical ideas can be given is a test. Backtesting is the practice of applying a clearly written rule to past price data to see how it would have behaved. Done carefully, it teaches you how a rule works and how fragile it is. Done carelessly, it produces impressive-looking numbers that mean very little.
What backtesting is
A backtest has four parts.
- A rule written precisely. For example: “buy at the close when the 10-day average rises above the 30-day average; sell at the close when it falls below.” There must be no room for judgment, because a computer or a spreadsheet has to follow it.
- A set of historical data. Daily or hourly prices for the coin and period you are studying.
- Assumptions about trading. Fees, the price at which trades are filled, how much is invested each time.
- Measurements of the result. Total return, the worst fall from a peak, number of trades, and how long the rule was out of the market.
The result describes only what would have happened in that sample. It says nothing certain about what will happen next.
A deliberately simple illustration
Imagine a rule that buys on day 1 at a price of 100 and sells on day 11 at a price of 110. Before costs, that is a gain of 10 on 100, or 10 percent. Now assume a trading fee of 0.1 percent of the trade value on each side. The buy costs 0.1 and the sale costs about 0.11, so about 0.21 in total, which brings the gain to about 9.79, or roughly 9.8 percent.
The numbers are made up purely for arithmetic. The point is that costs reduce results, and a rule that trades many times a month will pay them many times. A strategy that looks slightly profitable before fees can be unprofitable after them.
Pitfalls that make backtests misleading
Overfitting. If you try many rules and many settings and keep the one that did best on the past data, you have mostly found the one that fits that history’s accidents. Out of enough combinations, some will look good by chance. The more knobs a rule has, the greater the risk.
Look-ahead bias. This happens when a test accidentally uses information that was not available at the time, such as using a day’s closing price to decide a trade at that day’s open.
Survivorship bias. Testing only on coins that still exist today ignores those that collapsed or disappeared, which makes results look better than real experience.
Ignoring costs and price slippage. Fees, the bid-ask spread and the gap between the price you expected and the price you got all reduce returns. In thin markets they can be large.
Too little data or one market regime. A rule tested only during a long rise may look brilliant simply because prices went up. Crypto has seen very different conditions, and a few years of history can contain only one or two market cycles.
Ignoring the path. A rule with a good final return may have suffered a deep fall along the way. Looking at the maximum drawdown shows how much pain a trader would have had to sit through, and whether it would have been realistic to continue.
Better habits when testing
- Split the data. Design the rule on one part of the history, then check it once on a later part it has never seen. This is often called out-of-sample testing. If performance falls apart, that is useful information.
- Keep rules simple. Fewer settings means less room to fit noise.
- Include realistic costs. Use fees and assumed slippage that are on the pessimistic side.
- Compare with something plain. Ask how the rule did against simply holding the asset over the same period. A complicated rule that does worse than doing nothing has not added value.
- Look at many outcomes. One total-return figure hides the variation. Look at the ups and downs.
- Record what you tried. If you tested fifty variations, the best one is much less impressive than it looks.
What a good result still does not tell you
Even a careful test can only say that a rule would have worked in the past under the stated assumptions. Markets change. A pattern that was common when few people were watching may fade, and crypto markets have changed a lot as new venues, products and participants arrived. A good backtest is at most a reason to look further, not evidence that a rule will make money.
It is also easy to underestimate the human side. A test follows a rule perfectly. A person may hesitate, abandon the rule after a loss, or add size after a win. Emotional discipline is a large part of why results in practice differ from results on paper.
Where this fits
Testing is a way to learn about trading strategies, not a route to guaranteed returns. It also pairs with risk management: deciding how much you could afford to lose comes before any thought of how much a rule might gain, and position sizing affects outcomes at least as much as the entry signal does.
Cryptocurrency prices are volatile, past performance does not predict future results, and you can lose everything you invest. This article is for education and is not financial, tax or legal advice.