FAQs Frequently Asked Questions

When you check the seemingly impressive returns on a timing site, here are very important questions you should ask: 

• How has the site performed on an annualized basis since Jan 2005? Most all sites fall short here. The past few years have been most challenging to timing sites. Some sites that show high returns overall are due to making abnormally huge returns in 2008 while making only mediocre returns in 2005-2007. 2011 was particularly challenging since it goes down as one of the most volatile, trendless years in market history. Such inconsistent returns are liable to cause an ulcer!
• How many switches are made? Some sites switch 75-100+ times/year. This drives up commissions costs. 
• Has the site switched its strategy midstream? Read the fine print. Some sites report high annualized returns then show that the strategy was optimized midstream. In other words, they were theoretical prior to the switch, but report returns as if the whole run were live.
• The total return is massive. Ignore total return. It is meaningless. Total returns are often massive and boggles the mind. For example, my model’s +33.1%/yr return since July 1974 would give a total return of return of 2,560,467%. Stated another way, $1 would have become $25,605. With enough time, the power of compounding is powerful indeed. My article ‘How I Made +18,241.2% in the Stock Market Over 7 Years’ was titled that way because that huge percentage is intriguing to the potential reader, but this massive % should be broken down into annualized returns which make more sense. In fact, the first thing I do in that article is break down that mind boggling percentage into an annualized percentage which averages out to +110.5%/yr over these 7 years. As an aside, my returns were actually higher than +110.5%/yr because I did not invest money earmarked to pay for taxes each year when I was trading the markets, but kept it in my trading account, thus, at a total tax rate of +50.5% (federal, state, etc) back in 1997-2000, my starting capital base each year was significantly less. However, there was no way using normal accounting standards that KPMG who did the verification could account for this. On the other hand, my return of over 70,000%+ Kevin Marder cites in his book Conversations With Top Traders does account for this.
• Does the website show theoretical signals going back many years but its live signals are less than a year old? Do your due diligence. Check to see if the creator of the model has any prior performance track record or something that demonstrates a high level of competence. Googling the model creator’s name can be a good way to find information on the person’s achievements. Google is an excellent way to find information fast then you can piece together all the links that appear on the person in question to get a clearer picture. 
  
  In addition, keep in mind some sites may boast high theoretical returns over a long period. It is essential to know if they possibly over-fit the data to create those high returns. Over-fitting occurs when excessive attention is paid to past data while failing to account for the system’s predictive value going forward. This is a common trap that affects many timing systems which is why so many fall short. The system may yield impressive results over a historical 20-year period because the parameters were fit to maximize profits over that period. But going forward, the returns will fall short because the system was over-fit. 
  
  In essence, a timing system must start by containing internal logic that makes sense, then the system can be built around this internal logic. This is where many years of market experience are necessary. This avoids the 'black box' situation of over-fitting data. Unfortunately, so many timing systems on the internet lack internal logic, but manage to boast high theoretical returns based on over-fitting their past data. I highly recommend Robert Koppel’s book "Bulls, Bears, and Millionaires" where he interviews Mike Dever who discusses the perils of over-fitting data.