The parameters I am thinking of for an MACD indicator are the fast average, the slow average, and the smoothing average. I've run into some situations where I screwed up and made the fast average longer than the slow average. What I always see in examples is the smoothing average fitting somewhere between the fast and slow average intervals. But sometimes I feel silly and make it larger, and this has shown some promise. I'm trying to understand, in principle, what some of this can imply, other than that I'm a moron. We have that one covered when I accidentally flipped fast and slow. Generally I'm trying to use MACD with a mean reversion approach. All I can think of for the signal inverse being successful is that I'm basically 100% out of phase in my timings. The resulting indicator is often very speculative but promising. I haven't been able to make one work sustainably by itself; that's where other stuff comes into play etc. I don't know yet what to make of the smoothing average being longer. What it looks like it does generally is keep positions open for a little bit longer, while still reacting to any quick turnarounds that are happening. That's the best I could do. For what it's worth, I would be curious about stories of others goofing up their indicators since I think it would help me in understanding a little more how they behave and misbehave.