Automating Performance: Can You Beat the Algorithms?

This is a platform focused on autonomous (fully independent of any human input) investing algorithms, generating BUY or SELL signals over a particular period of time. The platform aims to stimulate research and simulation ofinvesting algorithms, starting initially with a focus on the liquid US equities. It aims to add new algorithms and assets/markets every 3~6 months going forward.

Why investing algorithms? The two main reasons are lack of emotional bias and transparency. Next question: can machines invest without any human interference? The answer is “YES” nowadays, and the current platform is here to prove it.

MarkSignAlgo aims to be a platform that encourages the research and development of investing algorithms, which focus on the decision aspect of investing, as opposed to the trading algorithms, which focus on how to trade a live order. If trading algorithms have been competing for the same job with traders and sales traders, investing algorithms would compete for the same job with investors and fund managers. Here is a short graphical introduction.

Algorithms are not necessarily the best at investing, but they do fix two fundamental issues mentioned above:

• Emotional bias -- Algorithms are not emotional. How many mistakes haven’t we all made by not following well established rules because of various reasons? Algorithms don’t have this problem: once coded and launched, they maintain the discipline and stick to the rules.
• Transparency -- Algorithms results can be simulated. Even without access to the internal details of an algorithm, we can always see and check its simulation results. And more importantly, the way such results are being created is consistent. Such consistency is unique to algorithms and does not have an equivalent with human beings, because humans grow and change continuously, for good or for worse.

This platform is focusing on autonomous investing algorithms that do not require any human touch or input once they are launched. Everything from data update to signal computation is done automatically. Human intervention may only be necessary to make sure the infrastructure is up and running.

There is no one best algorithm, as they all trade each ticker or each financial instrument (stock, crypto or something else) differently. The same algorithm can deliver excellent performance for one stock, but relatively poor performance for another. That is why each available ticker has its own page for every algorithm that applies, where subscribers can check the investment simulation results. Simulations do not account for most of the real life variables, like trading costs, potential slippage costs and so on, and as such they aim to be a reference at best.

The key criteria for all algorithms on the platform is consistency under any market conditions. There are 2 specific high level conditions that the algorithms aim to avoid:

• the so-called “momentum trap” behaviour. This occurs when algorithms do very well when the market is moving in one direction, but fail when such direction changes .
• the crowd effect (for lack of better naming). The algorithms “logic” is designed to work even if every participant on the street, the “whole crowd”, uses them.

Note: most algorithms are either focusing on identifying Buy Signals or Sell Signals. They are not combined because the underlying logic for identifying such signals is different. We leave this as an experiment to the readers.

Each individual asset/ticker has its own page, where subscribers can see the results of available algorithms simulations on such an asset, including the detailed simulated Buy and Sell trading over the recent years.

Each algorithm will also have its own page, where the reader is only presented with the description of inputs and outputs of such algorithms, as well as the philosophy/ideas underlying them, but not with the internal details of such algorithms - such information remains proprietary. The page will flag at least the following items:

The current platform does not offer financial advice. The goal here is to stimulate people’s idea generation and testing. Even with the algorithm internals not being public, there are plenty of variables that can change the outcome of any investment. Each algorithm page is trying to suggest further areas of research for people to improve their results.