I got some data once for the probability that if a bull candle printed, the next candle would break the high of the previous candle, and vice versa for a bear. Over 10 samples the outcome could be 100 percent success ratio in some data, over 20 samples it could be 75% success, over 50 samples it could be 25% success ratio. The point i am getting at is that the probability of something happening has a random distribution over small sets of data. Statistically speaking, you need over 1000 samples to fully remove any random distribution and arrive at an average. The data i got for my scenario was easily quantifiable and thus easy to program a testing framework to collect the data. The probability that the trend would continue was something like 50.1% over 10,000 samples when using a reward of 1 to break the high of the previous candle, and a risk of equal distance (1:1 r:r). So there is an edge in trend continuation, but its really really small unless you start to combine other well researched edges.

To know that what you are dealing with will go up more than it will go down, you need over a thousand samples, mathematically speaking, otherwise you are still dealing with randomness. However if you are just investing, then you do not need a chart, you need fundamentals.