Without a proper understanding of algorithmic hypotheses, it can be costly to waste time in subsequent stages without meaningful return. Finding a good algorithm may take a few years, even though forming one only takes minutes. A beginner in algorithmic trading will fail 99.5% of the time. Increasing the success rate from 0.5% to 10% is already a great success. This section demonstrates how to improve the success rate of an algorithm hypothesis.

Master Foundation Knowledge

• Data. Input data is a prerequisite to any algorithm. It plays an important role in forming an algorithm hypothesis. For instance, limiting input data to price and volume will force algorithmic traders to have technical analysis as the only choice.
• Strategy. The investment strategy must be concretized into a set of principles that must be followed when making trading decisions. It decides what securities to buy or sell, what type of order, when and what price, and in how much volume. For any investment strategy, the overall goal is to generate the highest possible return with the lowest risk. Understanding the risk-return trade-offs will enhance your chance of building a successful trading algorithm. Don’t focus too much on a single strategy. In fact, the majority of algorithmic traders only focus on the price momentum strategy because it is easy to program and execute. Imagine what would happen if 99% of traders share the same strategy? 

  Multiple strategies will limit risks in the long term. Articles 12 to 23 detail common trading strategies. Algorithmic traders can refer to these strategies to develop their own trading algorithms. They can be tailored to individual tendencies to become more effective.

• Individual tendencies. Every trader has their own personality tendencies. Some prefer safety to high risks with high returns. Some are willing to bet their entire portfolio on a few positions, instead of dividing them into smaller orders to lower risks. Some focus only on high-frequency trading strategies, expecting to find profits in the short term while ignoring long-term strategies. 

  Algorithmic traders may suffer from emotional discomfort when downplaying their individual tastes when forming algorithm hypotheses. It’s because trading algorithms can operate wildly differently from their own personality. Understanding emotional tendencies will make the journey much simpler. For example, an algorithmic trader with low-risk tolerance who follows a price momentum strategy will find it difficult to stay calm with temporary losses. They often elect to intervene at unnecessary moments.

Real Market Experience

Proposing a good algorithmic hypothesis often comes from an already effective approach in discretionary trading. Personal trading experience with both wins and losses is always valuable for algorithmic hypotheses. This source of information is irreplaceable since it comes from real experience. Knowledge from books and training courses is helpful but still cannot replace real trading. Thus algorithmic traders should take experience from decision-based trading, especially in the beginning. Ten years of market ups and downs will form a solid foundation for any investor.

Learn Proven Successful Trading Strategies 

There are countless publicly available trading examples and strategies for free. Compared to a brand new, unproven high-tech strategy, investors can increase success rate based on proven trading strategies.  Here are some examples:

• Warren Buffett’s investment philosophy on value investing on undervalued stocks.

• Passive investment in SPY.

• International arbitrage business.

• Market makers in commodities and forex markets.

Examining the potential of an algorithm hypothesis before the start of the testing phase will save a lot of time and effort. It also increases the success likelihood at the next stage of algorithm development.