Many bloggers are seeing a bubble right now, so I dusted off my analysis of the dot com bubble. Bubbles are different from normal tops. They rise sharply, and then pop. From roughly 5000 down to 3000, the NASDAQ composite lost 40% of its value in two months.
The banking crash of 2008 was a gentle top, by comparison. The two exciting months saw the SP500 drop from 1200 to 800, but the index had been trending down for a year.
O’Neil claims that this top was telegraphed by a series of distribution days. I generally use Chaikin’s money flow as a cumulative volume indicator. To use O’Neil’s method for market timing, you must squint at the individual volume bars, or you could just subscribe to the IBD. They do it for you.
Chaikin wasn’t much help here. Neither were momentum measures like RSI and MACD, breadth measures like high-low percentage, or volatility measures. One indicator that does look good is Wilder’s ADX. Trend following systems typically include ADX > 20. Such a system would have exited automatically on March 20, avoiding a lot of grief.
Since all indicators look backward, it’s good to verify that we can see what the indicator is seeing. January had two bad runs of 3-5 days each, but they were not too bad in the context of the ADX(20) lookback. February has a long stall, and the ADX line goes flat. In March, down days begin to exceed up days, in number and in magnitude.
By March 27, the index has failed to make a new high, and the ADX is signaling that the uptrend is over. If you were stubborn, you (unbelievably) got two more chances to bail out near 4500 before losing that level forever.