One of the toughest jobs an investor has is finding indicators that work during all kinds of markets, not just bull markets.
Too often analysts find something that has signaled key market turning points and they jump on it as if it were a holy grail to analysis. The problem is that most times these indicators work well…until they don’t, turning paper profits to realized losses.
Analysts may use momentum indicators like moving averages, point and figure, fundamentals, or even more complicated versions that combine multiple data sources to make investment decisions. They look at recent history, see that their indicator would have been profitable, and then apply it going forward. Things may start off well, but eventually the future proves it is not the past, as such a technique necessitates, and all gains are wiped out.
Data Fitting
Take the lesser known NVI (Negative Volume) indicator, for example.
Developed in the 30s by Paul Dysart, the indicator seeks to put more value on lower volume days and no value on higher volume days. When the indicator rises it means the S&P rose on a day that volume was less than the previous day’s volume. If volume is increased, the day’s price change is discarded.
The theory behind the indicator is that contrary to popular belief large institutions are actually the more active market participants during days of low volume, thus giving more weight to those days, and the less important retail investor is the driver of volume on higher volume days.
The chart below shows the NVI’s recent history and indeed how a strategy of buying the market when the NVI crosses its moving average has captured the majority of the bull move (^GSPC) since 2009. Following the indicator as a trading signal, investors wouldn’t have sold out of any of the market’s pullbacks and be up around 67% so far on paper. Not bad!
Have we found our holy grail?
Follow us on Twitter @ETFguide
Law of Large Numbers
In statistics, the law of large numbers states that the more data one has in analysis, the more likely the realistic mean is to be like the theoretical mean. We discussed this in more detail with our subscribers recently in our Technical Forecast concerning the “Sell in May and Go Away” tactic and how it hasn’t been the best strategy over the past five years.
In the above chart, the analysis only includes one data point, the Buy signal in late 2009.
In more detailed statistical analysis, terms such as p-value, significance levels, and normal distribution utilize the law of large numbers to ultimately prove that the more data points a data set has, the more the results can be trusted.