Golden Cross And Death Cross

The cross is an averaging lag, expressed in months

A golden cross is a 50-period moving average crossing above a 200-period one, normally on a daily chart. A death cross is the same event in reverse. Both are widely reported as turning points, and both are, by construction, among the latest signals in common use.

The reason is arithmetic rather than opinion. The 50-day average only rises above the 200-day average when the last 50 closes have averaged higher than the last 200. If a market has spent six months falling, the 200-day average is still full of higher prices from the start of that period. Lifting the 50-day average above it requires a rally sustained long enough to change the shorter average by a substantial margin, which takes weeks, not days.

Counting how far price has already travelled

Work a concrete case. An index trades at 4,000 and falls to 3,400 over sixty sessions, a decline of 15 per cent. During that slide the 200-day average drifts down from roughly 4,050 to around 3,900 while the 50-day average falls faster, to about 3,550, and the death cross printed somewhere in the middle of the decline, not at the top.

Now the recovery. Price bottoms at 3,400 and rallies. To produce a golden cross, the 50-day average has to climb above roughly 3,850 where the 200-day now sits. Since the 50-day average is the mean of the last fifty closes, that requires those fifty closes to average above 3,850, which cannot happen until price has been trading well above that level for most of two months. By the time the cross prints, price is typically back near 4,000. That is a rise of 600 points off a low of 3,400, or roughly 18 per cent, and 4,000 is where the decline began. The whole of the fall has been retraced before the signal appears.

That figure is also the answer to a separate question people get wrong. A 15 per cent decline needs an 18 per cent rise to get back to level, because the rise is measured from the smaller number. Losses and recoveries are not symmetrical, which is the arithmetic behind every argument for keeping drawdowns shallow.

That is not a flaw in the signal. It is the definition of the signal. A 200-period average is a slow, deliberate measure, and asking it to be timely is asking it to be something else.

It is worth noting how much of that lag comes from the 200 and how much from the 50. The 200-day average moves a quarter as fast as the 50-day one for a given change in price, since each new close shifts a 200-period mean by a two-hundredth and a 50-period mean by a fiftieth, so almost all of the movement in the gap between them comes from the shorter line. The 200-day average is effectively a slowly drifting reference level, and the cross is the 50-day line catching up to it.