A computer that buys and sells around the clock according to fixed rules, without emotion and without a break. That is how a crypto trading algorithm is usually pitched. In practice, though, it is harder than the pitch suggests. This article explains what such an algorithm does, why crypto seems suited to it and where the risks lie, then compares building your own, buying a bot and investing through a fund. Our knowledge centre also has a broader article on algorithmic investing.<\/p>\n
At its core, a trading algorithm is a set of rules carried out by a computer. Those rules decide when to buy or sell, for what amount and when to close a position again. A rule might look at the direction of the price over a given period, for instance. It knows nothing about news or sentiment, only the data it is given.<\/p>\n
The software connects to a trading platform through an API, a technical link over which it places orders and retrieves prices. The orders end up in the platform’s order book, the list of outstanding buy and sell orders. What sits in that book determines how well an order is filled. Technically, then, an algorithm for Bitcoin differs little from one for any other coin. The real difference lies in the market it trades in.<\/p>\n
The crypto market has no closing bell. Trading carries on overnight and at weekends, a rhythm no person can keep up for long. Historical price data for the larger coins is also easy to obtain, which makes rules simple to test. The market’s volatility plays a part as well. Where prices move sharply, longer trends sometimes form, and many strategies are designed to capture exactly that.<\/p>\n
Those same features bring extra risks, however. A market that never closes can move sharply on a Sunday night with nobody watching, and the system must then cope on its own. What a systematic approach involves in broad terms is covered in our article on a systematic crypto strategy<\/a>.<\/p>\n The same volatility that creates opportunities also makes an algorithm vulnerable. A rule that works well in a calm market can produce a string of losses in a turbulent one. The Dutch Authority for the Financial Markets (AFM) notes in its consumer information that the value of crypto rests largely on speculation, so prices can fluctuate sharply.<\/p>\n Liquidity also varies widely by coin and platform. Liquidity is the extent to which you can trade without moving the price yourself. For smaller coins the order book is often thin, so a large order gets filled at progressively worse prices. That gap between the expected and the actual price is called slippage. Add trading fees, and a strategy that trades frequently can easily earn less than the model implies.<\/p>\n In a flash crash, the price on a single platform plunges briefly and often recovers soon afterwards. An algorithm that sells at fixed price levels may exit right at the low. The loss is then realised, even if the price is higher again a quarter of an hour later.<\/p>\n Just as important is the security of the platform and of access to it. An outage or a change to the API can stop orders from going through. Anyone holding an API key can also trade on your behalf. So switch off withdrawal permissions on such a key and restrict access, for example to fixed IP addresses. More broadly, any form of automated trading should make clear who has access, what that access allows and where the crypto is held. Why safekeeping matters for larger sums is explained in our article on storing crypto safely<\/a>.<\/p>\n Almost every algorithm is first tested on historical data. Such a backtest shows how the rules would have performed in the past. That is useful, but it is also tempting. Anyone who adjusts the settings for long enough will eventually find a combination that looks excellent in hindsight. This is known as overfitting. The rules have then been tuned to chance patterns in that particular period, and such patterns rarely return.<\/p>\n David Bailey, Jonathan Borwein, Marcos L\u00f3pez de Prado and Qiji Jim Zhu described this problem in a study in The Journal of Computational Finance<\/a>. In their view, testing the same data many times over makes it almost certain that some results rest on chance alone. What a backtest can and cannot show is set out in our article on backtesting and what it proves<\/a>.<\/p>\n Search for a crypto trading algorithm and ready-made trading bots soon appear. Some are simple tools that carry out rules you set yourself, leaving you responsible for the strategy and the risks. Other providers promise a fixed or guaranteed return. That is a warning sign, because no rule can guarantee a profit in a volatile market.<\/p>\n The AFM describes exactly this pattern as a form of fraud on its page about the risks of crypto<\/a>. A website promises guaranteed, high profits through a so-called automatic trading platform. Once the consumer has deposited money or crypto, the provider disappears and the money cannot be recovered.<\/p>\n So always check who is behind a provider. Anyone who provides crypto-asset services to clients in the EU on a professional basis, such as holding crypto in custody or executing orders, generally needs authorisation under the Markets in Crypto-Assets Regulation (MiCA). Certain existing financial institutions, such as banks and investment firms, can instead notify their regulator. The AFM keeps a register of crypto-asset service providers<\/a>. If a firm is not listed, the AFM advises extra caution. Managers of investment funds are generally not included in this register. The AFM keeps a separate list of managers that fall under the registration regime, and Callisto Capital is on it. That also applies to Callisto Capital. In January 2025 the AFM also stressed that trading crypto remains very risky even under supervision.<\/p>\n Building your own crypto trading algorithm gives you full control, at the cost of a great deal of work. You maintain the software, manage connections and keys and watch the system even on holiday. A purchased bot removes some of the programming, but the risks around strategy, security and choice of platform still rest with you.<\/p>\n A bot is software that carries out the instructions it is given, whereas a fund also has an organisation around the strategy. A well-run fund usually has a manager who runs the strategy and handles risk management, a separate legal owner of the assets, a custodian and an independent administrator who calculates the value of the fund and keeps the books. The investor then owns no coins but participations, which entitle the investor to a pro rata share in the value of the fund’s assets. A fund does not by definition perform better or carry less risk than a bot. The difference lies in how it is organised and who is responsible for what. The trade-offs are discussed in our article on investing in a crypto fund or buying crypto yourself<\/a>.<\/p>\n Callisto Capital manages a systematic, trend-following crypto fund with an algorithmic strategy. It is a fund for joint account (fonds voor gemene rekening, FGR) whose assets are legally owned by a foundation (stichting). Amdax holds the crypto in custody in the name of the foundation, through a custody foundation that is separate from Amdax B.V. AssetCare, as an independent party, keeps the fund administration. AssetCare calculates the net asset value (NAV) each month, which the manager checks against its own estimate and approves. Investing in crypto remains risky in this form too. Fixed rules make it predictable how the fund trades, not what the outcome or the return will be.<\/p>\n If you would like to know how our fund works, we would be glad to welcome you at our office for an introductory meeting. In it we discuss our investment philosophy in broad terms, the fund structure, the risks and the terms for subscribing and redeeming. You can arrange a meeting via contact<\/a>.<\/p>\n Callisto Capital manages a crypto fund for high-net-worth individuals and entrepreneurs.<\/p>\n This article is general information, not investment advice. Investing in crypto is risky and you may lose your investment. Callisto Capital B.V. is registered with the AFM, but does not require a licence and is not subject to ongoing supervision by the AFM.<\/em><\/p>\n","protected":false},"excerpt":{"rendered":"A computer that buys and sells around the clock according to fixed rules, without emotion and without a break. That is how a crypto trading algorithm is usually pitched. In practice, though, it is harder than the pitch suggests. This article explains what such an algorithm does, why crypto seems suited to it and where…","protected":false},"author":3,"featured_media":2756,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"_acf_changed":false,"callisto_views":7,"footnotes":""},"categories":[22],"tags":[],"class_list":["post-2722","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-systematic-investing"],"acf":[],"_links":{"self":[{"href":"https:\/\/callistocapital.nl\/en\/wp-json\/wp\/v2\/posts\/2722","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/callistocapital.nl\/en\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/callistocapital.nl\/en\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/callistocapital.nl\/en\/wp-json\/wp\/v2\/users\/3"}],"replies":[{"embeddable":true,"href":"https:\/\/callistocapital.nl\/en\/wp-json\/wp\/v2\/comments?post=2722"}],"version-history":[{"count":6,"href":"https:\/\/callistocapital.nl\/en\/wp-json\/wp\/v2\/posts\/2722\/revisions"}],"predecessor-version":[{"id":2791,"href":"https:\/\/callistocapital.nl\/en\/wp-json\/wp\/v2\/posts\/2722\/revisions\/2791"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/callistocapital.nl\/en\/wp-json\/wp\/v2\/media\/2756"}],"wp:attachment":[{"href":"https:\/\/callistocapital.nl\/en\/wp-json\/wp\/v2\/media?parent=2722"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/callistocapital.nl\/en\/wp-json\/wp\/v2\/categories?post=2722"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/callistocapital.nl\/en\/wp-json\/wp\/v2\/tags?post=2722"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}Where a crypto trading algorithm runs into difficulty<\/h2>\n
A good backtest proves very little<\/h2>\n
Off-the-shelf trading bots and promises of fixed returns<\/h2>\n
Building your own, buying a bot or investing through a fund<\/h2>\n