1. OKex's BTC futures strategy, using two cycles of data 1D and 1H; 2. retest K-line cycle selection 1D, bottom K-line cycle selection 1H, retest 2018.1-2020.9 retest results are good; If the K line cycle chooses 1D, the bottom K line cycle chooses 15M or 5M, the result is not ideal.
I don't know where the problem is, can you point me in the direction?
Is it a problem with the data, a problem with the code, or a problem with your own retrieval parameters?
The grassThe impact of the bottom-line K-line cycle, see the Community Top-up Primary Tutorial on feedback
The Little DreamYes, the linear retrieval system first needs to retrieve data, however, the parameters for retrieving the K-line in the code are uncertain, so the retrieval system settings specify the K-line cycle, the bottom K-line cycle. It is used to start the request data, and multi-cycle data is supported, not every cycle data is loaded into one of the retrieval system, for performance, only one cycle data is loaded, the big cycle data is composed of small cycles, so the K-line cycle, the bottom K-line cycle is set differently, the number of KBAR lines obtained varies, that is why.
The Little DreamYes, the number of K-lines will have an impact.
The Little DreamDetermine if the underlying K-line cycle is the same, when you set the default cycle, the underlying K-line cycle changes automatically.
The Little DreamIf your strategy is based on a K-line indicator, and you use different K-line cycles depending on the indicator, the signals will certainly be different, and the results will be different. The underlying K-line cycle affects the tick data (real-time price at the time of the tick), and if you use too large a K-line cycle, the data generated by the tick will be large in size (rough, but the tick rate is fast, so you can choose how many K-line cycles you need) to some extent.
The grassAnd if you look at the regression principle, the smaller the underlying cycle, the more accurate it is.