Moving Averages, Drawn Out
Moving averages are where almost every other indicator starts. The line down the middle of a Bollinger setup, MACD, Keltner channels, Supertrend — all of them sit on top of some kind of moving average. Understanding this one properly makes the rest much easier.
Every chart here is drawn over the same real data as the Bollinger lesson: 200 four-hour BTC/USDT candles from Binance, 3 August to 5 September 2026, computed with the same math the 5PM Worker Simulator runs in a backtest.
1. How the average gets built
A moving average adds up the last N closes and divides by N. It *moves* because that stretch shifts forward by one every time a new bar arrives — the newest close joins, the oldest drops out.
The shaded region is the 20 bars being averaged right now.
- Price (close)
- SMA 20
BTC/USDT · 4h · Binance · 2026-08-03 → 2026-09-05
The shaded region is the 20 bars being averaged right now. Drag the slider to stop anywhere.
Notice the line doesn't start at the left edge. Averaging 20 bars requires 20 bars, so for the first 19 there's simply nothing to plot. Every indicator has this warm-up.
2. SMA, EMA, WMA, HMA — what actually differs
There are several kinds of moving average, and they differ in exactly one way: how much weight each close in the window gets.
An SMA treats all 20 closes identically. An EMA weights recent closes more heavily, decaying exponentially into the past. A WMA applies the same idea linearly. An HMA combines WMAs in an attempt to cut the lag further.
Overlay them yourself below, same period and same data. Two are on by default — four near-identical averages stacked at once is harder to read, not easier.
All four at 20 bars, over identical data — the only difference is how each weights the closes in its window.
- Price (close)
- SMA
- EMA
BTC/USDT · 4h · Binance · 2026-08-03 → 2026-09-05
The label at the end of each line identifies it. Watch EMA and WMA hug price slightly more closely than SMA.
The EMA here seeds its first value from the simple average of the first N closes. TradingView's Pine Script seeds it slightly differently, so the first few bars can differ marginally — a known, deliberate divergence that converges after a handful of bars.
3. Longer means slower
The one real trade-off a moving average has: a short period tracks price closely but gets shaken by every wobble; a long period is smooth but turns well after price already has. Neither is the right answer.
One average, from fast and twitchy to slow and smooth.
- Price (close)
- SMA 20
At 19 bars of warm-up, 181 of 200 bars have a value to plot.
BTC/USDT · 4h · Binance · 2026-08-03 → 2026-09-05
Drag from 5 to 60. The longer it gets, the further the line sits from price and the later it turns.
4. Golden crosses and death crosses
A single moving average doesn't make much of a rule on its own. So the common approach uses two — one fast, one slow — and treats the moments they cross as the signal. Fast crossing up through slow is called a golden cross; the other direction, a death cross.
Below is a 10 against a 30. Press play and a marker lands at each crossing as the sweep reaches it.
A 10-bar average against a 30-bar one, with a marker at each crossing.
- Price (close)
- SMA 10
- SMA 30
- Golden cross
- Death cross
4 of 4 crossings reached.
BTC/USDT · 4h · Binance · 2026-08-03 → 2026-09-05
Across this whole window there are only four crossings. The golden cross on 17 August landed just before the big move.
If you read the Bollinger lesson first, that date should look familiar. The squeeze — the point the bands were at their tightest — was 16 August, and this golden cross is the very next bar. One market event, flagged two different ways by two different indicators.
5. But crossings reverse, a lot
The real problem with crossover signals is stretches without a trend. When the two lines are tangled together, crossings fire in quick succession and flip direction each time. That's called whipsaw.
Toggle between the two settings below — same data, same idea, completely different numbers.
A faster pair reacts sooner and reverses more often. Both are shown on identical data.
- Price (close)
- SMA 9
- SMA 21
SMA 9/21: 10 crossings across the window, the tightest run being 4 within 12 bars around 08/27.
BTC/USDT · 4h · Binance · 2026-08-03 → 2026-09-05
The highlighted stretch is wherever crossings bunch up tightest. On the 9/21 setting, four of them land on 27 August alone.
The lesson isn't that 9/21 is bad. A faster pair reacts sooner and reverses more often; a slower pair does the opposite. They're two ends of one dial, not a better and a worse option — and which end suits your asset on your timeframe is something you have to backtest rather than reason out.
6. What the Hull moving average changes
The Hull moving average is a well-known attempt at cutting lag: take a WMA of half the period, double it, subtract the full-period WMA, then smooth the result by the square root of the period. It looks convoluted, but the goal is simple — react faster at the same nominal period.
Same period, far less lag — and the overshoot that comes with it.
- Price (close)
- HMA
- SMA
BTC/USDT · 4h · Binance · 2026-08-03 → 2026-09-05
An SMA (dashed) and an HMA (solid) at the same period. Watch how much sooner the HMA turns — and how it overshoots price at those turns.
Nothing about reduced lag is free. The HMA catches turns earlier, but it's jumpier, and at a reversal it tends to shoot past price before coming back. The Simulator's Hull MA Trend template is built around that behavior.
7. Common misreadings
- A moving average isn't support or resistance. It's computed from past price and drawn behind it — it follows, it doesn't push.
- A crossing confirms, it doesn't predict. Fast crossing above slow means the recent average has risen above the older one — a statement about what already happened.
- Don't read anything into the word "golden". It's a conventional name, not a claim about outcomes.
- Changing the period changes your sensitivity, not the market. Shorter automatically produces more signals; longer automatically produces fewer.
- In a rangebound stretch, every setting produces repeated crossings. 27 August in the chart above is exactly that.
8. Seeing it run in the Simulator
Two of the Simulator's templates are built on moving averages. Both use the same formulas shown here, and both can be backtested against real candle data for whichever exchange, symbol and timeframe you care about.
If you want to compare fast/slow combinations rather than guess at them, the Sweep tool runs a real backtest for every combination across the ranges you pick and ranks them — so the difference between 10/30 and 9/21 above becomes a result rather than an argument.
- MA Crossover — the basic version: it buys when the fast average crosses above the slow one and sells when it crosses back below.
- Hull MA Trend — uses the direction of a lag-reduced Hull moving average to decide whether a trend is in place.