In order to be used effectively, you cannot take every buy and sell signal that results from crosses of the signal lines. If you do, you will find that you will have a lot of losing trades and whipsaws. In order to minimize churning and whipsaws, filters must be added to eliminate some of these trades.
The filters that I am using for the MACD is that we will only take long trades with a flat to rising 50dema or from a rising 200dema with price at or above the 200dema and when there is a positive divergence. Reverse for shorts. We exit long trades on either the second sell signal or if the MACD falls below the zero line. Reverse for shorts.
I have labelled the relevant signals for the Qs above. The interpretations have been a little tricky, and I have used some discrection, but I hope that my explanations will convey how I have applied the rules. At A there was a negative divergence short signal with no ambiguity. At 1, on June 2, there was a valid buy signal as price bounced off of the rising 200dema, but I chose to ignore that signal as price had risen to within 2% of the falling 50dema when the signal occured. At B there was another buy signal off the rising 200dema with a little more room below the 50dema accompanied by a rising bottoms pattern in the MACD. This was also the second buy signal since the April short signal was initiated, and the short position was reversed to a long position. At this point we are long, and we ignore the first sell signal on June 28. As long as the MACD itself does not fall below its low of May 26, we will stay in the position until the second sell signal occurs even if price falls below the May low. At C we have the second sell signal and exit the long position. There is no signal to go short at C since the 50dema is rising and there is no negative divergence.
The signal at 1 was ignored by discretion. Signals 2 and 3 were not actionable according to the rules as set forth. It is possible that the signal at 3 could be considered an actionable long signal, but since the price low was well below the 200dema, the signal was ignored. In most cases, one would have expected a pullback to generate a second actionable buy signal. That did not happen here and a major rally was missed. However, this is not a reason to change the rules. Perhaps better discernment would have led you to conclude that was a valid long signal, and that would be ok, but having not taken it, I will not change the results retroactively. Even without the lastest rally, this strategy as described is up 16% for the year, which is excellent.
Currently, the MACD gave its second sell signal since the rally began, which is a strong indication that a significant pullback or correction is underway. A buy signal with a rising 50dema would be taken should it occur. This goes against my expectation that the Qs will fall below the August high, but that is not something that would fall under the umbrella of discretion.
One more advanced MACD behavior that I would take on discretion (but not include in the Strategy Tracker since it is not a part of the rules) is a hook sell signal. If the market rallies and the MACD attempts to cross above its signal line and fails by turning down without crossing, this would be a valid short signal.