Moving Averages, Plainly

A moving average is the least mysterious indicator in existence and it still generates an enormous amount of confused writing. The confusion is not in the maths, which takes one sentence to explain. It is in what people believe the line is telling them.

So: take the closing prices of the last twenty periods, add them up, divide by twenty. That number is the twenty-period simple moving average. When the next period closes, drop the oldest price, add the newest, recompute. Plot the result over time and you get a line that follows price around, smoother than price itself, and always a little behind it.

That last part is not a flaw to be engineered away. It is what the thing is.

What does a moving average actually do to price?

It removes short-term variation so that longer-term direction becomes visible.

Price on any timeframe is a mixture of movement at many scales. Some of it reflects sustained pressure over weeks; some is a single large order filling in one candle. A moving average blends adjacent periods together, so the one-candle events get diluted and the sustained pressure survives. The longer the period, the more aggressive the blending.

The cost is stated plainly by the arithmetic. Because the average includes prices from N periods ago, it can only respond to a change in direction after enough new prices have accumulated to shift the sum. A 200-day average that has been rising will keep rising for weeks after price turns down, because 199 of the 200 inputs still come from the old regime. This is not lag in the sense of a delay that better code could fix. It is lag in the sense that the number is defined as a summary of the past.

Everything else in this guide follows from that one fact.

Simple or exponential — does the type matter?

Less than the arguing suggests.

A simple moving average (SMA) gives every period in the window the same weight. The price from twenty days ago counts exactly as much as yesterday's, right up until it drops out of the window entirely, at which point its influence goes from full to zero in one step.

An exponential moving average (EMA) weights recent periods more heavily, with influence decaying smoothly the further back you go. Nothing ever fully drops out; old prices just fade toward irrelevance. The practical effect is that an EMA turns sooner after a sharp move and does not produce the small artefacts an SMA can when an unusual old price leaves the window.

Put a 50-day SMA and a 50-day EMA on the same chart. They will track each other closely, diverging most during fast moves and converging again during quiet ones. On a phone, at typical zoom, the two lines will frequently overlap into one thick line. Choosing between them is a real choice, but it is a small one.

The period is the large one. A 10-period average and a 100-period average on the same chart are not two versions of the same tool; they describe entirely different things. One tracks the last two weeks of a daily chart, the other the last five months. Whether you use SMA or EMA at either setting changes the line by a fraction of what changing the period does.

And the period is only meaningful in combination with the timeframe underneath it. "The 50" means fifty hours, fifty days or fifty weeks depending on the chart, and people say it as though it were a fixed quantity. Timeframes covers why this ambiguity causes more confusion than any other single thing in chart reading.

Why are the standard settings 20, 50 and 200?

Because they are round numbers that approximately match calendar periods on a daily chart, and because everybody else uses them.

Twenty trading days is about a month. Fifty is about a quarter. Two hundred is about a trading year in equities. These were sensible conveniences in markets that closed at night and on weekends. Crypto trades continuously, so a "200-day" line covers 200 actual days rather than 280 calendar ones, and the equity heritage of the number does not really transfer.

The honest position is that these settings have no special predictive property. What they do have is attention. A very large number of people watch the 200-day average on Bitcoin, financial media reports when price crosses it, and that shared focus can make the level matter through nothing more than coordination. That is a real effect and it is a different kind of claim from "the number works." It also means the effect is strongest on the most-watched assets and essentially absent on a token nobody charts.

What is a crossover, and how late is it?

A crossover is when a shorter-period average crosses a longer-period one. Shorter above longer is usually called bullish, the reverse bearish, and the 50-over-200 version on a daily chart has the marketing-friendly names golden cross and death cross.

The mechanics are straightforward. The short average responds faster to new prices. When recent prices run well above the older ones for long enough, the short line rises through the long one. That is all a crossover is: an arithmetic consequence of price having moved in one direction for a sustained stretch.

Which is exactly the honest note. A crossover is a report on movement that has already happened. By the time a 50-day line crosses a 200-day line, price has typically been travelling in that direction for weeks or months. The signal is not early, cannot be early, and is not designed to be early. Anyone presenting a golden cross as news is presenting the average of the last fifty days as news.

That does not make crossovers useless. As a way of formalising "the medium-term picture has changed relative to the long-term picture," a crossover is clear, objective and hard to argue with after the fact. It just cannot be both a lagging summary and a leading indicator, and most of the writing about it wants to have it both ways.

The specific failure mode worth knowing is choppy conditions. When price oscillates sideways, two moving averages will cross back and forth repeatedly, generating a string of contradictory readings in quick succession. This is sometimes called whipsaw and it is the reason crossovers look far better on historical charts of trending assets than they do in practice. Adding volume context is one of the few things that helps here, because crossovers that happen on almost no participation are the ones most likely to reverse immediately.

Does any of this work on a phone?

Better than most indicators, actually.

Moving averages are drawn on the price panel rather than in a separate pane below it, so they cost you no vertical space. Two lines over candles is legible on a 6-inch screen in a way that a stacked oscillator panel is not, and what survives at mobile scale is exactly this kind of large, smooth structure.

What does not work on a phone is fine judgement about whether a cross has actually occurred. Two lines converging and touching looks identical to two lines crossing when each line is a pixel wide and the vertical scale spans thirty percent. If the answer matters, zoom in or rotate to landscape.

The other mobile-specific point: watching for a crossover by opening the chart repeatedly is a poor use of a phone. A crossover is a slow event by construction, so checking it hourly tells you nothing that checking it weekly would not. If a price level near the average is what you actually care about, an alert on that level does the watching for you, and alert fatigue is worth reading before you set fifteen of them.

Where this goes next is worth flagging. Moving averages and RSI both compress price into a single line and both lose the same information doing it: how many people were involved. That is a separate reading, and it is the one most mobile users never turn on.