A golden-cross alert arrives for Harbor Coffee, our fictional coffee business. The stock has already risen for 45 trading sessions, from $54.06 to $64.80. The label sounds like an early warning. Why is it so late?
Those are invented future prices, but the delay comes from real arithmetic. The two lines still carry months of older prices. To judge what the alert adds to your research, you need to see which prices are inside them.
An average with a moving window
A simple moving average, or SMA, adds the latest N prices and divides by N, giving each equal weight. Here we use daily closing prices. A 50-day or 200-day average counts trading sessions, not calendar days.
Return to the opening chart's five closes: $62, $63, $65, $64 and $66. They total $320, so the five-session SMA is $320 ÷ 5 = $64.
Each close gets one-fifth of the weight. The line smooths the daily bumps so you can follow the trend, the direction prices have been moving. A smoother line buys clarity by giving up immediacy.
Work through the smoothing
The next close is $68. The $62 drops out, leaving $63, $65, $64, $66 and $68. The total becomes $320 − $62 + $68 = $326. Divide by five again: $65.20.
The window moves; it does not grow. There are still five prices in the average.
The outgoing price matters as much as the incoming one. Even another $66 close would lift the SMA to $64.80, because it would replace the older $62. The average can rise on a day the stock goes nowhere.
An exponential moving average, or EMA, gives newer prices more weight. Each old price fades instead of dropping out on a fixed day. A five-period EMA can still carry prices from more than five sessions ago.
It needs a starting value, called a seed; here that is the $64 SMA at session 5. The standard five-period weight is 2 ÷ (5 + 1), or one-third. After the $68 close, the update is $64 + one-third × ($68 − $64) = about $65.33.
| Session | Close | SMA5 | EMA5 |
|---|---|---|---|
| 5 | $66 | $64.00 | $64.00 |
| 6 | $68 | $65.20 | $65.33 |
The EMA follows this jump more closely: it puts one-third of the weight on the new close, while the SMA gives it one-fifth.
Longer windows usually smooth more and respond later. That delay is smoothing lag. The EMA responds faster to new prices, but both averages follow prices that have already happened.
Read the level and the crossing
There is no universal high or low SMA. Harbor's $65.20 average means something next to its $68 close: the latest price is above its five-session average. You can also read whether the line is rising or falling. Compare lines using the same interval and the same split and dividend adjustments.
A moving-average crossover happens when one average moves from below another to above it, or back again. The common golden cross is the 50-day average crossing above the 200-day. A death cross is the reverse. We use SMAs here; some chart providers use EMAs.
For the alert's separate 420-session Harbor path, price falls, recovers and falls again. Each SMA uses its latest 50 or 200 closes, including that session's close. The plot starts at session 200, once both windows are available.
The low is $54.06 at session 200. The golden cross arrives at session 245, when the stock is already $64.80. The 50-day SMA is $58.99, just above the 200-day's $58.87. The shorter average has caught up after 45 sessions of rising prices.
The delay works in reverse, too. Price peaks at $78 in session 300. The death cross arrives at session 392, with price down to $64.20, after 92 declining sessions. We can see both turning points in hindsight; neither was known to be a lasting turn that day.
A crossing is an event. If the 50-day average stays above the 200-day tomorrow, that is the same relationship continuing, not another golden cross.
Ask what the evidence tested
The name sounds worse than the record. In its April 21, 2025 review, LPL Research reported on 36 past S&P 500 death crosses in its history starting in 1950. The rule was the daily 50-day average crossing below the 200-day.
Over the next 12 months, the index gained an average of 6.3%, with 72% of those periods positive. Losses occurred, but a death cross usually preceded a positive year in this sample.
These are US-dollar price-index returns, without reinvested dividends or trading costs; the early history includes the predecessor S&P 90. They describe what followed a signal, not the profits of a trading strategy.
A trading claim needs more: when to buy, when to sell, and what trading costs. Compare that result with holding the same index over matching dates, using the same return measure. A positive return alone does not show that the signal helped.
The optional chart-patterns lesson examines broader tests of moving-average rules, including why a historical winner can fail on fresh data. Momentum investing covers the evidence for strategies based on past returns.
What the alert adds
Quick reversals can produce whipsaws: the averages cross and cross back, sending alternating signals without a lasting trend. The optional MACD lesson shows how these reversals affect an indicator built from two EMAs.
Harbor's alert tells you something precise: its latest 50 closes now average more than its latest 200, after weeks of recovery.
The alert earns a place in your research as a description of that change. It supplies no probability for the next move. Volume adds a different observation: how many shares traded during the move.
In short
- An SMA keeps a fixed window: each new close replaces the oldest.
- An EMA lets old prices fade and gives new ones more weight; it still looks backward.
- A golden or death cross can arrive long after the price has turned.
- Positive returns after a signal do not prove it beats holding the index after costs.