Quantitative Trading

What is quantitative trading?

Quantitative trading, also known as algorithmic trading, is a trading method that strictly follows the buying and selling decisions given by computer algorithm programs. It replaces human subjective judgments with advanced mathematical models, and uses computer technology to audition a variety of "high probability" events that can bring excess returns from huge historical data to formulate strategies. It greatly reduces the impact of investor mood swings and avoids irrational investment decisions in the case of extreme frenzy or pessimism in the market.

Quantitative trading is not a simple process

It is easy to confuse quantitative trading with technical analysis. In fact, quantitative trading is much richer in content. Many quantitative trading systems use fundamental data, such as valuation, market value, cash flow, etc. when modeling and calculating, and some algorithms use news as variables for calculation. Technical analysis basically only needs to use the volume and price data of the trading target.

Quantitative Trading Features:

1. Discipline

Make decisions based on the results of the model's operation, not on feelings. Discipline can not only overcome human weaknesses such as greed, fear and fluke, but also overcome cognitive biases and can be tracked.

2. Systematic

The specific performance is "three more". One is multi-level, including models at three levels: major asset allocation, industry selection, and selection of specific assets; the other is multi-angle, the core idea of quantitative investment includes macro cycle, market structure, valuation, growth, profit quality, analyst profit forecast, market sentiment, and other angles; the third is multi-data, that is, the processing of massive data.

3. Arbitrage thinking

Quantitative investing uses a comprehensive and systematic scan to capture the opportunities presented by mispricing and misvaluation, in order to identify valuation depressions and profit by buying undervalued assets and selling overvalued assets.

4. Probability wins

One is that quantitative investment continuously mines and exploits patterns that are expected to repeat from historical data; the other is to rely on portfolio assets to win, rather than individual assets to win.

Due to the above advantages of quantitative trading, many investment institutions will have quantitative strategy reports, including fundamental quantitative, technical indicators quantitative, sentiment indicators quantitative, etc. Some of these reports are done in a very rigorous manner, and the reliability of the conclusions is very high.

Seeing that many novice investors must have developed a strong interest in quantitative trading, what quantitative communities do we have for learning?

The current popular domestic communities include Youkuang, Jukuan, and Rice Basket, which bring together many miners to share and discuss quantitative trading strategies. The community also provides free backtesting platforms for students who lack data to use.

In addition, the quantitative community abroad is richer. If you have good enough English, you can find it by going to Bing or Google. By then, you will be able to access the vast majority of strategic ideas in the field of quantitative trading.

What are the basic steps for formulating a quantitative strategy?

The first step is to build logic using ready-made metrics.

There are many technical indicators built into the software, take one, write it into the buying and selling point, and test the historical market, so that you can get a simple strategy. With the accumulation of strategy experience, the logical choices here will become more and more diverse.

Of course, such strategies are generally not profitable.

Therefore, our second step is to perform parameter optimization.

Choose a parameter traversal to see how different parameters affect the strategy. In general, we will get several sets of parameters that seem more profitable.

Then, our third step is to perform out-of-sample detection.

For example, the parameters we traversed before were several parameters that performed well from the data in 2016, so we will use the 2014/2016 data to test these parameters. Generally speaking, this parameter will be very dismal outside the sample, and it will not be as powerful as the optimization in the sample at all.

At this point, the fourth step is to observe and determine the reasons for the failure of the strategy.

Suppose we find that the reason for the failure of the strategy is that one or two special markets outside the sample lead to large losses, then we can set a hard stop loss to avoid this risk; if we find that the failure of the strategy is due to too few transactions, then we will slightly relax the trading logic, such as requiring the place of x to be changed to = x or even = x-1. And so on, this modification is the experience of the strategy.

After setting up the new logic, we return to step 2 and repeat the above steps.

We ended up with a sample strategy that made money both inside and outside.

The fifth step is to track the real deal.

Examine the strategy over time in a completely unknown market in the future and observe how the strategy actually performs. If the performance is in line with expectations, then the strategy is effective.

The sixth step is to conduct transactions.

As the trade progresses, we also observe the effectiveness of the strategy when it is found that the strategy has lost more than expected.

The seventh step is to adjust or terminate the strategy.

Sharing experiences in specific development

To some extent, strategy is a thinking-trying-thinking-trying cycle.

② When selecting metrics, it is necessary to avoid using future functions.

③ Parameters should not be excessive.

④ In parameter optimization, the more parameter groups available, the better, indicating that the strategy is highly effective.

⑤ In the real offer, the expectations of the strategy are generally discounted, and 50% of the expected value is qualified.

The strategy of trading too few times is usually luck.

To test a super profitable curve, you must have written the wrong logic.

As can be seen from the practical steps, quantitative trading is not suitable for any investor. If you are interested in this trading strategy, then you need to have some programming skills. But this does not mean that quantitative trading is out of reach.

If you don't have the corresponding programming skills, you can choose to buy other people's strategies or go to social trading platforms to copy orders, which is a more reliable way.

If you find it difficult to control your emotions during trading and want to use a computer to perform systematic trading, but you are afraid of mathematics or technology, it is recommended that you first read the entry-level masterpiece "Open the Black Box of Quantitative Trading", written by a quantitative financial expert on Wall Street. The book combines a large number of cases and true stories to explain for you that quantitative trading is actually easier to control and understand than you think.