Why Models Fails – Increased Use or Faulty Design?

Posted Mar 20, 2016 by Martin Armstrong

QUESTION: 

It seems successful trading/investing/forecasting systems work only when used by a few. When these same systems get adopted by more and more users their effectiveness diminishes to a point where they no longer work. Do you agree with this, and if yes do you think as your forecasts become more well known and applied they too would lose their effectiveness?

A

ANSWER: No. Systems fail because of a lack of historical data testing. For example,Black-Scholes failed creating the Long-Term Capital Management crisis in 1998 because the data used for developing the model was only back to 1971.

Models that only function within the “noise” level will work for a period and then totally fail. If you do not even test something on the Great Depression, how can you expect that model to even survive such an event? It is IMPOSSIBLE to change the long-term trend. Everything is connected. To really make gold suppressed, you have to do that to all commodities, create deflation when the central banks are trying desperately to create inflation without success, and it goes on and on.

These excuses are clearly by people who look for excuses to justify their “theory” behind some trend. The markets move collectively. There are no single trends for the market will arbitrage it into its proper place within the general scheme. So models that become less efficient were improperly designed from the outset. We are all behind the collective movement and that includes the average person who does not even trade or invest. The rest focus of unemployment, GDP, trade, whatever, and respond to what? To the trend of the whole.