I will tell you; ”While in theory randomness is an intrinsic property, in practice, randomness is incomplete information.“ - Nassim Taleb
Only because signal(information) emerged in the data(China news) and signal makes prediction easier, still faulty because you never really know; depends on the importance of the signal. Check my post below.
Most trading days of the year are uneventful, put simply, they're randomly constructed. A few trading days of the year, lets say 14 days out of 252 trading days, carry the signal to move the market and produce returns, in our example below, positive returns. Hypothetically, lets say at year 2020 end of year market return is 20%. In this hypothetical, the 20% return will be a result of (2) weeks worth of trading days, so 14 days, out of 252 trading days of the year. Contrary to popular belief, market returns are not Gaussian but follow more of a Cauchy/Levy distribution. Keeping this in mind, it highlights that most trading days of the year, the chart, is constructed of randomness. You don't know all the forces behind the scenes, you have incomplete information on what is going on. And we know empirically that data does not allow you to use it to predict when prediction matters, such as the tails. This means that data is always going to noisy, in fact, history is noisy. They're especially noisy if you're just looking at a normal trading day, and 1 normal trading day alone. For a day trading charlatan to tell me that he can predict next 30 minutes, with prior 30 minutes...or something else similar to this, I want to slap this arrogant bullsh*tter fool across the face. Not saying he won't make money or will make money, but if that is the strategy, I can confidently say, it is a result of luck.
This is brilliant. Thank You I could not have posted better. Backtesting is a deadend...you cannot put the future into a can...then it would not be the unknown. ES
You just set minimums and maximums on the past with no regard to reality and what is going on at the time....stress testing is valid...but backtesting is old news...