Decoding Markets and Marwood Research
Download 1.87 Mb. Pdf ko'rish
|
101 Lessons For Aspiring Traders PDF
01
02 Combining trading systems together is one of the best ways to reduce drawdown and improve consistency. For example, combining 16 uncorrelated signals, each with a Sharpe ratio of 0.5, would generate a portfolio Sharpe ratio of 2.0. That is a very profitable system. 03 04 Don’t believe anything you read until you have tested and tried it out for yourself. What used to work ten years ago may not work today and you only find out what works by rigorous testing and analysis. Follow a system and pay attention to statistical analysis. But don’t be afraid to skip a trade that doesn’t feel right. Always do a post-trade diagnosis to see if you made the right decision. 05 06 07 08 09 10 Don’t give up on a system after a few losing trades because that might be the time the system turns around. Learn about the law of large numbers and make sure you are able to cope with the expected system drawdown before you start trading. Trading systems are crucial for traders. But they are complex to create and 90% of traders will lose money trying to create their own due to the many mistakes that creep into the development process. Trading strategies need to be built with careful attention to robustness and real world dynamics. Most quantitative trading systems only have a very small edge but that’s OK. Virtu Capital only wins 51-52% of its trades but it went 6 years with only one losing day. Virtu harnesses the power of large numbers to exploit it’s tiny edge. There are many different ways to validate a trading system, none of them are foolproof. Large sample sizes, cross validation, walk-forward, Monte Carlo. These methods can provide confidence but nothing is guaranteed. That’s because the stock market is a dynamic, complex system that changes all the time. Always be wary of the role of luck in trading. As the saying goes ‘don’t confuse brains with a bull market’. A trading system can always break when a bull market ends. When backtesting a trading system, always leave some data out-of-sample to verify your strategy with (preferably five years or more). Only backtest the out-of-sample once and keep a record of your findings. A good OOS performance is essential for a trading system but it doesn’t guarantee anything. What you want to see is a similar performance out-of-sample to what you saw in-sample Download 1.87 Mb. Do'stlaringiz bilan baham: |
Ma'lumotlar bazasi mualliflik huquqi bilan himoyalangan ©fayllar.org 2024
ma'muriyatiga murojaat qiling
ma'muriyatiga murojaat qiling