This is a good question. The way Coppock described his indicator, it only gives a signal if it turns up after having fallen into negative territory. It seems to me that these would be the best signals. It has also been my experience that shallow signals, which occur not very deep into negative values, are the most likely to be premature signals. Often, a better signal occurs after the Coppock has subsequently fallen deeper into negative values.
All of that said, we should not lose sight of what the Coppock indicator is. It is a very specific example of a long term momentum oscillator. It does not have fixed boundaries like the Relative Strength Index (RSI). This suggests that there is scope for some flexibility in interpretation. One example (noting that exceptions do not prove anything) is the
So, the rule for me is to keep thinking through what the indicator does and be flexible if necessary. However, there are two vital caveats. The first is that if there are few or no upward trending stocks at the time the Coppock turns up, it may be a premature signal. The second is to be far more risk averse in the case of acting on a turn up from above zero or only very shallow in negative values.