Quantitative trading is not only for institutional traders; retail traders are also involved. If you want to generate algorithms, programming skills are recommended, but even these skills are not always necessary. You can use programs and services that write programming codes for strategies based on the input you provide. Then insert the code generated by the program/service into the trading platform and start trading. But before all this happens, people who want to become algorithmic traders will go through a few steps to determine what they want to accomplish with algorithms and how to do it.
Time frame and limitations
Although well-programmed algorithms can run on their own, some manual supervision is recommended. Therefore, please select the time frame and transaction frequency that you can monitor. If you have a full-time job and your algorithm is programmed to make hundreds of trades per day on a one-minute chart while you are working, then this may not be ideal. You may wish to choose a slightly longer time frame for your transactions and reduce the frequency of transactions so that you can pay close attention to it.
The profitability of the algorithm testing phase does not mean that it will continue to generate these returns forever. Sometimes, if the result shows that it is no longer functioning normally, you will need to step in and change the trading algorithm. This is also a time commitment that anyone engaged in algorithmic trading must accept.
Financial constraints are also a problem. With high-frequency trading strategies, commissions will increase quickly, so make sure you work with the lowest cost broker available, and the potential profit of each transaction is guaranteed to pay these commissions, possibly multiple times a day. Start-up funding is also a consideration. Different markets and financial products require different amounts of capital. If you are trading stocks within a day, you need at least US$25,000 (more is recommended), but trading foreign exchange or futures may require less capital.
Market restrictions are another issue. Not every market is suitable for algorithmic trading. Choose stocks, ETFs, foreign exchange pairs or futures with sufficient liquidity to process the orders that the algorithm will generate.
Develop or fine-tune strategy
Once you understand the financial and time constraints, you can develop or fine-tune a programmable strategy. You may have a manual trading strategy, but is it easy to code? If your strategy is highly subjective and not rule-based, programming the strategy may not be possible. Rule-based strategies are the easiest to code-entry, stop loss, and price target strategies based on quantifiable data or price changes.
Since rule-based strategies are easy to copy and test, if you don’t have your own ideas, you can get a lot for free. Quantpedia is such a resource, providing academic papers and trading results on various quantitative trading methods. You can code the outlined rules and test the profitability of past and current data. The coding algorithm requires programming skills or access to software or someone who can code for you.
Testing the trading algorithm
The most important step is testing. Once the trading strategy is coded, do not use it to trade real capital before testing. Testing includes running the algorithm on historical price data, showing the performance of the algorithm in thousands of transactions. If the historical testing phase is profitable and the statistics generated are acceptable for your risk tolerance-such as maximum drawdown, win rate, bankruptcy risk, then continue to test the algorithm under real-time conditions on the demo account. Again, this stage should generate hundreds of transactions so that you can access performance.
If the algorithm is profitable on historical price data and trading real simulated accounts, please use it to trade real capital, but be vigilant. Real-time conditions are different from historical or demo tests because the orders of the algorithm will actually affect the market and may cause slippage. Before verifying whether the algorithm is effective in the real market, as it did in the test, please stay vigilant.
As long as the algorithm runs within the statistical parameters established during the test, do not use the algorithm. The advantage of the algorithm is that there is no emotion when trading, but a trader who constantly revises the algorithm is offsetting this benefit. However, the algorithm does require attention. Monitor performance. If the market conditions change so much that the algorithm no longer works properly, you may need to make adjustments.
Algorithmic trading is not a one-time effort to make you rich overnight. In fact, there are as many quantitative transactions as manual transactions. If you choose to create an algorithm, be aware of how time, financial, and market constraints may affect your strategy, and plan accordingly. Transform the current strategy into a rule-based strategy so that it is easier to program, or choose a quantitative method that has been tested and researched. Then, run your own test phase using historical and current data. If the verification is passed, the algorithm is run with real money under surveillance. Adjust if necessary, otherwise let it work.