Moving Averages: SMA vs EMA and the Golden Cross Record

Simple moving average (SMA): the mean of the last N closes

A moving average takes the average price of a set number of recent bars, recalculates it on every bar, and joins the results into a line. The most basic kind, the simple moving average (SMA), adds up the last N closes and divides by N. An SMA20 is the average of the 20 most recent closes including the current bar, and each time a new bar forms, the oldest close drops out. That is why an SMA can move even when the new close has not changed, simply because a large old value has left the window. The value being averaged is usually the close, and the figures in this course and the tools on this site calculate from closes too. Whether tomorrow's SMA20 rises or falls depends on whether the new close is above or below the close 20 bars ago, so if there was a big spike 20 bars back, you can anticipate the line turning on the day that bar drops out. Many people read the slope of the line as the trend, but changes in that slope are mixed with this effect of old values leaving the window. Some variants average the midpoint of the high and low or weight by volume, so if your value differs from another chart, first check what was averaged.

Exponential moving average (EMA): more weight on recent values

The exponential moving average (EMA) is calculated by blending the new close with the previous EMA: EMA = α × current close + (1 − α) × previous EMA, where α = 2/(N+1). For an EMA20, α is about 0.095, so the current close accounts for 9.5% and the previous EMA for 90.5%. The weight of older closes shrinks with every bar but never reaches zero. TradingView uses the SMA of the first N closes as the first EMA value, so values can differ slightly from tool to tool depending on where the calculation started. This is also why several tools on this site load a generous number of bars before calculating. Because old values never drop out all at once as they do in an SMA, the EMA does not suddenly turn on the day a spike from 20 bars ago leaves the window. In exchange, a sharp move in a single recent bar shows up more strongly. If the same chart shows slightly different EMA values in two tools, the cause is usually a different starting point or number of bars loaded, and the gap narrows as more bars accumulate.

How much they lag: SMA vs EMA

An average is built from past values, so it always trails price. While price rises at a steady pace, the SMA sits at roughly the price level of (N−1)/2 bars earlier: 9.5 bars back for an SMA20 and 99.5 bars back for an SMA200. The EMA is often said to lag less, but with the same N it falls behind a steady trend by the same distance (mathematically, both lag by (N−1)/2 bars). The difference shows up when the direction changes. The EMA gives the most recent bar about twice the weight the SMA does (9.5% versus 5% for a period of 20), so it turns first, and for the same reason it jumps around more on a single sharp bar. What this difference means is not that one is better but that there is a trade-off. A line that turns quickly shows a reversal early but also gives more false turns; a slow line shows it only once the move is clear, but late. Shortening the period produces the same trade-off. So rather than searching for a moving average that lags less, it is more realistic to know roughly how many bars your line lags on the bar length you watch.

Common periods, and what changes in crypto

Internationally, 20, 50 and 200 are the customary periods, while in the Korean stock market 5, 20, 60 and 120 are. The Korean convention comes from stock-market trading days: 5 days is one week, 20 about a month, 60 about three months and 120 about half a year. Crypto trades every day with no weekends, so 20 daily bars are 20 days, not a month. The period is also a count of bars rather than days, so an SMA20 on a 1-hour chart covers 20 hours (see the article on timeframes). This site's Pro Trading Chart uses SMA 5, 20 and 60, and the Multi-Timeframe Trend Matrix uses EMA20 and EMA50. There is no evidence that any particular period is the right one; the fact that many people watch the same numbers is closer to the reason those numbers stay in use. Carrying stock-market conventions straight over to crypto also changes what a period means. A 60-day line on stocks covers about three months, but 60 daily bars on a coin cover two.

Golden crosses, death crosses and bullish or bearish alignment

When a shorter-period line crosses a longer-period line from below, it is called a golden cross; when it crosses from above, a death cross. Traditionally the terms refer to the 50-day and 200-day lines crossing, and in Korea crosses of the 20-day and 60-day lines are also widely watched. For a cross to happen, the average of recent prices has to overtake the average over a longer period, so a golden cross usually appears only after price has already climbed a good way off its low. In the figure below, too, the close on the crossing bar is already about 20% above the low. That is why many people looking at a cross also check how far the crossing bar's close is from the low and whether the long line is still sloping down. A cross where the short line breaks through while the long line is still falling only says the bounce was large, not that the longer move has changed. Bullish and bearish alignment likewise just summarize which way price has moved so far, and by the time a bullish alignment is complete, a good part of the rise has often already happened.

• Golden cross: the short line crosses the long line from below
• Death cross: the short line crosses the long line from above
• Bullish alignment: shorter lines above longer ones in order, like the 5-, 20-, 60- and 120-day lines from top to bottom
• Bearish alignment: the long lines on top and the short lines below

Measured: what actually happened after golden crosses

For this course, we took 29,946 daily bars of 10 coins on Binance (BTC, ETH, BNB, XRP, ADA, DOGE, LTC, LINK, TRX and SOL, each from its Binance listing date to September 2026), found every bar where the SMA50 crossed above or below the SMA200, and compared the prices that followed with a baseline measured over all bars. The share of cases where the close 20 bars later was higher was 48.3% after golden crosses, slightly below the baseline, and 56.8% after death crosses, which was actually higher. On BTC 4-hour bars, looking 30 bars (5 days) ahead reversed this: golden crosses were followed by a higher share than the baseline and death crosses by a lower one. Looking 6 bars (one day) ahead, though, the result went the same way as on the daily bars. Even with all 10 coins combined, each type of cross occurred only around 90 times, and because the coins move together, these are not independent cases either. The direction changed with what was measured and over what horizon, so 'price rises after a golden cross' was not confirmed in this data.

• Baseline (all bars, 20 bars later): up 50.4%, median +0.15%
• 87 golden crosses: up 48.3%, median −0.55%
• 88 death crosses: up 56.8%, median +3.13%
• 59 BTC 4-hour golden crosses: 47.5% after 6 bars (baseline 52.0%), 59.3% after 30 bars (baseline 52.8%)

Why moving averages look like support and resistance

In an uptrend, you often see price pull back to a moving average and then rise again. The average sits roughly in the middle of recent prices, so it is mathematically natural for a pullback to reach its neighborhood, and orders may also gather near a line that many people watch, such as the 200-day. But with the 5-, 20-, 60-, 120- and 200-day lines all on the chart, there is almost always some line near price. Bounces off a line stick in memory while clean breaks through it are easily forgotten, which makes the lines feel more reliable than they are (see the article on the limits of chart analysis). Treating the area as a zone with some width rather than a single line leads to fewer mistakes (see the article on support and resistance). A sensible order of checks is to see first whether the line slopes in the direction of the trend, whether price has reacted near the same line several times before, and whether volume rose when it did. Then separate a wick that pokes through the line and comes back from a close that clearly settles beyond it.

A common misconception: the cross moves the price

Saying price rises because a golden cross appeared gets the order backwards. A moving average is calculated from price, so a cross is a late confirmation of a price change that has already happened, and the cross itself carries no information that was not already in the price. In a sideways market, where price moves back and forth within a range, the two lines tangle and crosses repeat at short intervals; following them means getting in late and watching price reverse, again and again (see the article on sideways markets). You can check how a crossover rule did in the past with the historical performance in the Golden & Death Cross Scanner or with the Crypto Strategy Backtester, fees included, but those results do not guarantee the future either. It is true that many people watch for crosses, and around a well-known 50/200-day cross, news coverage and attention can pile up. But which way that attention moved price was not consistent, as the measurements above show. When looking at past results for crosses, separating sideways stretches, where crosses were frequent, from trending stretches leads to fewer misreadings.

What looks different in crypto and stocks

The calculation is the same, but when the market structure differs, the line means something different. The first difference is time. Korean stocks build bars only during weekday regular hours (9:00 a.m. to 3:30 p.m.), so a 20-day line covers about a month of trading days, and the same convention holds for US stocks on regular-session daily bars. Crypto builds bars through weekends and holidays, so the same 20 bars are 20 days. The second is gaps. When Samsung Electronics or SK hynix opens sharply higher on overnight chip-industry news from abroad, a whole day's change lands in one close, and the gap between price and the moving average suddenly widens. Crypto trades without a break, so the same news tends to be spread over several bars. The third is price limits. Korean stocks are capped at ±30% a day, which puts a ceiling on the size of one bar, while crypto and US stocks have no such daily limit. The fourth is volatility. Coins tend to move more in a day than US indexes or large tech stocks, so even the same SMA20 shows a different distance between price and line and a different frequency of crosses. It is best not to carry a feel for 'breaking the 20-day line' learned on stocks straight over to crypto.

A practical checklist

The most common trap when watching moving averages on a live chart is the bar still in progress. The last bar's close is the current price, which is not yet settled, so that bar's SMA and EMA are recalculated with every price change. A golden cross that appears mid-bar often disappears when the bar closes, and all the measurements above were taken on closed bars. When a daily bar closes also differs by market: Binance daily bars close at 00:00 UTC (9 a.m. Korea time), while a Korean stock's daily bar is settled after the market closes. The list below is not a rule for deciding signals but things to check before reading the lines.

• Bar length and period: an average of how many bars of how many minutes, hours or days
• Whether it is an SMA or an EMA, and whether it is calculated from closes
• Whether the last bar is closed or still in progress
• The slope of the long line, and how far price is from the line
• Whether recent crosses repeated in a sideways stretch or came once within a trend
• Whether past results were checked separately with a backtest that includes fees