The bars in the middle of a MACD aren't a separate signal. They're just the gap between the two lines, drawn as a column. Once you see that, the whole thing demystifies.
MACD has three parts. The MACD line is one moving average minus a slower one — a measure of how stretched short-term price is from longer-term price. The signal line is a moving average of that. The histogram is the MACD line minus the signal line. That's the entire recipe. No hidden intelligence, no prediction — it's subtraction, plotted.
So what does the gap tell you? The distance between the two lines is how fast the first line is pulling away from its own average. In plain terms, the histogram measures momentum's rate of change — the acceleration of a trend, not the trend itself.
That reframe fixes two mistakes. First, a histogram shrinking while price still rises doesn't mean "sell" — it means the trend is still up but decelerating, losing acceleration, not direction. Second, the "MACD cross" traders wait for is nothing more than the moment the histogram passes through zero. It's the same event described two ways; you're not getting a second confirmation, you're getting the same fact in a different costume.
None of this makes MACD worthless. It's a clean, honest instrument for one job: measuring whether momentum is building or fading. The failure is expecting it to predict. It's a speedometer with a lag, not a windshield.
The first-principles takeaway
An indicator is a transformation of price and time — nothing more enters it. MACD takes price, subtracts two averages, and shows you the acceleration. If you can't say in one sentence what an indicator computes, you're trusting a black box. When you can, you stop asking it for the future and start using it for what it is: a measurement.