NEWCard payments are live · from $18/month, USDT also acceptedSee plans
Tutorial · 10

How to Use Moving Averages in Crypto: SMA, EMA, Golden Crosses and Retests

A moving average smooths closes into a trend line. Real BTC daily charts from Jul–Oct 2026 walk through the EMA20/50/200 stack, the Aug 20 golden cross, September EMA50 retests, and the late-October break under the average.

Coinpred editorial teamUpdated

So far in this series we covered how to read candles, how to draw support and resistance, and how to use RSI. The next tool shows up on almost every serious chart: the moving average (MA).

One-line definition: an MA averages the last N closes by some rule and connects those averages into a smoother line. It answers “where has the center of price been lately?” It does not forecast the future, but it turns jagged noise into a readable trend and often behaves like dynamic support or resistance.

Every price below comes from Binance’s public BTCUSDT daily candles (UTC). Averages are computed from closes. You can download the same data and check us.

1. SMA vs EMA: two ways to smooth

The two workhorses:

  • SMA (simple moving average): arithmetic mean of the last N closes — equal weight on every bar.
  • EMA (exponential moving average): more weight on recent closes, so it hugs price and turns faster.

Schematic comparison of SMA and EMA

Neither wins forever. For daily crypto reviews we lean on EMA, because the market is fast and SMA can lag a beat; but if your system was written around SMA, do not casually swap in EMA — consistency beats “which line is smarter.”

The rest of this article uses EMA20, EMA50 and EMA200 — a very common crypto trio.

2. Big picture: three EMAs on BTC daily

July 1 to October 31, 2026:

BTC daily with EMA20 / EMA50 / EMA200

We mainly watch three things:

  1. Stack: shorter MA above longer MA is a bullish stack; the reverse is bearish; tangled lines mean chop.
  2. Price vs the average: closes above or below the key MA.
  3. Slope: rising, flat, or rolling over.

Approximate values near the 2026-10-31 UTC daily bar (from closes):

Metric Approx. (USDT) vs price (~77,336)
EMA20 80,693 price ~4.2% below
EMA50 80,460 price ~3.9% below
EMA200 73,178 price ~5.7% above
SMA20 81,099 a bit above EMA20

In plain English: at the time of writing, price had slipped under the short and medium averages but still sat above the long EMA200 — cool on the medium-term tape, long average not yet lost. Description, not a forecast.

3. Golden cross: EMA20 over EMA50 on 2026-08-20

A “golden cross” is a short MA crossing above a long MA; a “death cross” is the opposite. Catchy names, terrible standalone triggers — ranges will whip you with repeated crosses.

Scanning recent months for EMA20/EMA50 crosses, a clear golden cross printed on 2026-08-20:

Item Value
Cross date (UTC) 2026-08-20
Close that day 73,025.15
EMA20 that day ~65,276
EMA50 that day ~64,977
Close ~10 sessions later 77,682.00
Close ~30 sessions later 81,249.99

Golden cross on August 20 as EMA20 crosses above EMA50

What we actually care about:

  • The cross arrived after the late-August breakout from the Jul–Aug range (see the S/R tutorial), not out of thin air.
  • On the cross day, price was already well above both averages — more of a trend confirmation than a bottom-fishing signal.
  • About 30 sessions later the close was near 81,250. That is a hindsight path with pullbacks inside it, not a promise that “golden crosses rally for 30 days.”

4. Retests in an uptrend: price tagging EMA50

The second common use is treating an MA as dynamic support/resistance. September had a clean-ish example:

September pullback that hugged EMA50 then bounced

On 2026-09-15 the daily low was about 74,968 with a close near 75,644, while EMA50 was still grinding higher. Price tagged near that orange line, then lifted over the following sessions. Why that retest was readable:

  1. The broader stack was still bullish (short MA above long MA).
  2. The pullback contracted; there was no decisive close deep under the average that never recovered.
  3. Ideally you also stack location: does the MA retest overlap a support zone you already drew?

If price instead closes hard under the average and you still call it a “retest,” that is wishful labeling — next section.

5. Losing the average: late-October closes under EMA50

Trends end. Late October:

Date (UTC) Close EMA20 EMA50 Close vs EMA50
Oct 25 80,154.09 82,130 80,832 below (after whipsaws)
Oct 27 80,909.77 81,843 80,809 back above
Oct 29 77,991.89 81,433 80,719 clearly below
Oct 30 77,368.77 81,046 80,588 still below
Oct 31 77,336.01 80,693 80,460 still below

Late-October daily closes breaking under EMA50

October 25 already closed under EMA50 and October 27 reclaimed it; the damaging bar was October 29, closing near 77,992 — roughly three thousand under EMA50. Our blunt rules:

  • Wick through an MA: watch, do not declare failure yet.
  • Close back above: treat it as a probe.
  • Close well below, then fail to reclaim for several days: flip the MA’s role from “potential support” to “overhead resistance.”

That lines up with the RSI piece: after October 29, momentum cooled hard. Two tools pointing at the same fact beat either alone.

6. Bullish stack: short above long

When EMA20 > EMA50 > EMA200 and the lines slope up, you get a clean bullish ribbon. Late August into mid-September looked like that:

Bullish moving-average ribbon

One side statistic: in the ~90 trading days before writing, closes sat above that day’s EMA50 on about 74 / 90 days. That describes a bullish backdrop after the August breakout — it does not guarantee day 91. By late October price was already under the average.

A bearish stack is the mirror. Tangled, flat averages usually mean a range; crossover signals are dirtiest there, so we trade less and lean on range highs/lows instead.

7. Same length: SMA20 slower, EMA20 closer

Overlay SMA20 and EMA20 on the recent daily stretch:

SMA20 versus EMA20 on the same BTC window

At writing time SMA20 was about 81,099 and EMA20 about 80,693 — roughly a 0.5% gap. In a decline the EMA ducks sooner; the SMA drags a longer tail. If stops or sizing rules are tied to an MA, changing the formula quietly changes your risk, so know what you are swapping.

8. What a fake crossover looks like

Golden/death crosses earn their bad reputation in chop. Schematic:

Noise crossover versus trend crossover

Check Looks like noise Looks like trend
Distance after the cross Lines re-entangle quickly Averages keep pulling apart
Price structure Still inside the old range Range break or clear volume
Slope Both lines nearly flat Short MA steepens
Timeframe Only on a low timeframe Confirmed on the daily

Our filter: demand a daily cross + price on the right side of a key zone + no instant reverse cross over the next 3–5 bars. Slower, fewer whipsaws.

9. A five-step moving-average habit

Five-step moving-average routine

  1. Pick the timeframe: direction on daily (or weekly); execution on 4h/1h.
  2. Read the stack: bullish, bearish, or tangled?
  3. Watch retests: when price tags EMA20/50, does it hold or close through?
  4. Treat crosses carefully: always stack structure and volume; a naked cross is low weight.
  5. Write the plan: after a role flip, is old support now resistance? Where is invalidation?

If you make short-horizon direction calls on Coinpred, the MA stack is a useful filter for “is the backdrop even worth an active lean?” Contract mechanics are in BTC event contracts and how to use the platform.

10. Beginner pitfalls

  • Too many averages — six lines is chart fog.
  • Full size on every golden cross — ignoring structure and location.
  • Treating an MA as unbreakable support — October 29 already demonstrated failure.
  • Letting a low-timeframe death cross boss a healthy daily stack — mismatched horizons.
  • SMA today, EMA tomorrow — backtest and live rules no longer match.

FAQ

Which moving averages should I use?

No universal answer. Crypto dailies often use EMA20/50/200; equities love 50/200. Lock one set for three months before you shop for new numbers.

Is a golden cross a buy signal?

At best it is one candidate confirmation of trend. The August 20 cross mattered because it lined up with a range break, volume, and a strong RSI regime. Alone, in a range, it is weak.

If price breaks an MA, should I flatten immediately?

Depends how you define “break.” We weight the close and whether price stays away or reclaims. Wick pierces are often noise.

What does losing EMA200 mean?

A sustained break of a long average usually means the center of price has shifted down, so many playbooks cut long exposure or treat rallies as fades. “Sustained” needs time: one spike and reclaim is not the same as many closes underneath.

If an MA and a support zone disagree, which wins?

Static zones (prior highs/lows, round numbers) and dynamic averages can stack: when they overlap, give the area more weight. When they conflict, favor the cleaner, more-tested side and shrink size.

Can I build a quant strategy on MAs?

Yes — with trend filters, costs and invalidation baked in. Plain dual-MA crosses can work in trends and bleed for years in ranges. Everything here is educational — not investment advice.

How do 4-hour and daily MAs work together?

Daily sets direction and the key average; 4-hour times the retest or break. Do not let one 4-hour death cross veto a still-intact daily bullish stack.

Data and references

Charts drawn by Coinpred from public data. Markets are risky — decide and size for yourself.

Risk notice: this article is educational and is not investment advice. Event contracts and other derivatives are high-risk and you can lose your entire stake. The service is not available to users in mainland China.