Why using the daily charts is giving you false signals

Many people avoid using the smaller time frames (15 min, 5 min, 1 min) because it adds noise to the data and makes it difficult to pick out market turning points and trends. However, these smaller time frames contain much more data than the daily charts, making them more useful if you know how to make sense of everything. One of the main problems people tend to ignore is signal aliasing.

Aliasing is caused when the sampling frequency is much lower than the frequency components in the signal. For example, if there is cyclic behavior in the 5 minute chart, the hourly or daily might show a frequency that does not exist. This is illustrated in the figure below:

As an example, the top chart could represent a 5 minute chart, and the bottom chart shows what would happen if you just sample slower (like in the hourly or daily chart). This creates a false cyclic behavior that can cause "false signals" in the slower time frames.

This is a common problem in DSP (digital signal processing), which is generally solved by anti-aliasing filters. Essentially, if you apply a filter to the data BEFORE down-sampling to a daily time-frame, the aliasing effect can be eliminated.

I have an example of an anti-aliasing filter below using the 5 minute BTCUSD chart. I am using a low-order infinite impulse response filter (most people only use moving averages for filtering, which is horrible DSP practice, but that is a topic for another time). If the light blue blue line is then sampled every hour, this will produce an hourly chart that is resistant to aliasing.

I understand the simplicity and readability of higher time frames is extremely important, but I think it's important to understand the technical limitations that come with those benefits.