策略回测思维

How to determine whether a strategy is reliable using the "sample size"

Do you think a strategy can be concluded after testing 5 times?
In the highly volatile and noisy cryptocurrency market, I generally believe that:

Without 50 to 100 samples, the conclusion of the strategy should be questioned.

Simple suggestions:
1) Start with a small position or a simulation to run samples, rather than directly investing real money with a large position;
2) Record each trade's: entry conditions, exit conditions, profit and loss R value, holding time;
3) After reaching 50 or 100 samples, evaluate overall performance instead of getting hung up on one or two “particularly outrageous examples.”

Remember:
There will always be “peculiar market trends”,
but what we want to use is the "statistical positive expectation," not just a story that happened to be right this time.
#样本量 #策略回测思维