What are different types of back testing strategies?

There are three mainstream approaches to back test strategies: the use of actual price data divided into three groups; Bootstrap that uses the actual price of the price but exceeds them; and Monte Carlo simulation. There are theoretical problems that divide the system builders over which methods is the best. It is important for a merchant to properly employ at least one of the re -testing strategies to his system before he trusts his business capital. The number of generated trades is a critical problem in choosing a backward testing strategy; At least 1,000 trades are required at each stage of the system's work.

The use of actual price data, divided into three parts, is the usual initial point for most system builders. The system is created using the first third of the data. At this point, the builder found algorithms, which seems to create sufficient profit with a sufficiently low risk to offer good prospects. The second third of the data is used to optimize the systhe same.

was optimized after the tension, it will be used for the remaining third of the data. This is called testing outside the sample, and there is a place where most systems fail. If the system still has good results across at least 1,000 stores, the system creator has a viable system. If the system generates less than 1,000 trades in testing outside the sample, the creator should consider a different backward testing strategy.

Bootstrapping is a method of drawing some data from the overall set, testing, putting data back and drawing more data or resampling and repetition. The ideal number of resample is n n or n n to n th where n is the number of data in the original sample. For a merchant who is likely to deal with at least 2,500 data points - 250 days a year in 10 years - this is not practical. Fortunately 100 rewrites will provide a high level of bootstrap trustEdky. If it takes 100 Resample, it does not provide the necessary 1,000 stores, the trader must continue to withdraw until this objective is met if he expects the system to be reliable rather than just a resampling of data.

The latest method of back testing strategy is Monte Carlo (MC) simulation. This method uses a computer to generate simulated data and the system is then tested on this data. The advantage of MC simulation is that one can create an unlimited amount of data, allowing to generate 10,000 stores or any other number of stores. Another advantage is that every new set of data is out of sample. This offers the opportunity to perform repeated optimization and test runs; Simply optimize in this data set and then use these system parameters for additional data that generates the computer.

The disadvantage of the MC simulation is that data may not have exactly the same function of the probability distribution as the trade, which could distort the results. In the best of all possible withThe sentences should be used all three backward testing strategies in the process of examining the system. Success in all three should offer a very high probability of success in the real world trading.

IN OTHER LANGUAGES

Was this article helpful? Thanks for the feedback Thanks for the feedback

How can we help? How can we help?