[Hayek]

Originally Posted by taotra

Most indicators were never designed to predict. Even if they were, development would be based on statistical probability. And as we all know, probability is not certainty.


That's just plain false. Some people always try to argue that things change in trading. What has worked in the past, doesn't seem to work anymore, etc. The fact is, nothing changes in trading and it doesn't matter which market it is. There are only 3 directions: up, down, sideways (technically there are only 2 because if you go down to the tick level there is practically always an up or down, but that's splitting hairs), and 2 levels of volatility: high and low. These 5 factors are present in every financial market today as they were 10 years ago, or a 100 years ago in the Dow. If people made a killing decades ago using MAs then that's because the market was in a trending phase. MAs are designed to measure trend and they do that reliably and exceptionally well. If the market phase changes to one of consolidation or high volatility, MAs do poorly, but then that's expected. In that case, you use something else, like oscillators, Fibonaccis, support/resistance. If your backtests bombed it's because your 3-year data must have had a significant number of consolidation phases in it, which your model did not take into consideration.

Take a look at the daily Dow-Jones from around 1900 to today. It's basically one smooth J-curve, rising continuously from 0 to 10,000. MAs would have captured the majority of not all of that trend.

Hindsight.

If you already knew the market was in up trend, or down trend, or sideway phase, even without any indicator's help you would have been a billionnaire.