Once you've got moving averages, the next idea writes itself: what happens when a fast MA crosses a slow one? That crossover is one of the most-watched trend signals in all of markets โ famous enough to make headlines.
Paint the two classic lines on the chart above and follow along.
The two crosses
- Golden cross โ a faster MA crosses up through a slower one (the classic pair is the 50 over the 200). It signals that recent momentum has overtaken the long-term average: the trend is turning up. It's considered a bullish, longer-term signal.
- Death cross โ the faster MA crosses down through the slower one. Recent prices have fallen below the long-term trend: bearish.
On the chart above, find where the 55 (blue) climbs up through the 200 (white) after a bottom โ that's the golden cross, and it has historically marked the start of Bitcoin's big up-legs. Where the 55 slices below the 200, that's the death cross.
The catch every beginner learns the hard way
Moving averages are lagging โ they're averages of the past. A cross of two slow MAs happens well after the low or high is in. By the time a golden cross prints, price has often already run a long way up. So the cross is not a precise entry โ it's a regime confirmation. It tells you "the tide has turned" long after the exact turn.
The other trap: in a sideways range, the two MAs tangle together, crossing back and forth and firing false signal after false signal. Crosses only mean something when there's an actual trend to confirm โ which is why you read structure first (last lesson) and use the cross as backup, not gospel.
Next: Bollinger Bands โ wrapping a moving average in a volatility envelope.
In the lab
You've read about crosses โ now let the chart find every one of them for you.
The scanner paints both lines, marks every 55/200 cross in the coin's history โ gold arrows up, red arrows down โ and haloes the most recent one. Scroll back through them on the 1D timeframe and score the signal honestly: how many golden crosses led to real up-legs, and how many printed after most of the move was already done? That lag is the whole lesson. (On the monthly timeframe there aren't 200 bars of history yet, so the scanner honestly shows nothing โ averages need data before they exist.)