Trend: moving averages, the EMA cloud, MACD and ADX

A note on where these belong before the arithmetic, because it changes how you read the rest of the page. Tools like these are most useful before you are in anything — on a higher timeframe, to answer «what is the general direction of this thing» so that you are not fighting it. They are context, not entries. The article on technical analysis and order flow is the longer version of that argument.

The simple moving average

Add the last twenty closes, divide by twenty, plot the result. Tomorrow, drop the oldest and add the newest. That is the whole thing, and every other tool on this page is a variation of it.

Two properties follow directly from that arithmetic and are worth holding on to:

• Every one of the twenty counts the same. The close from twenty bars ago has exactly as much say as this morning's.
• It moves when a value leaves as much as when one arrives. A line can turn up because an old low dropped out of the window, not because anything happened today. That surprises people the first time they notice it.

The exponential moving average

The EMA fixes the first property and keeps the second one quiet. Instead of an equal share each, weight falls off smoothly as values get older — recent bars matter most, older bars fade but never fully disappear.

In practice it is computed one value at a time: take a fraction of today's close, add the rest of yesterday's EMA. For a 20-period EMA that fraction is about 9.5%, and it comes from 2 ÷ (period + 1). Nothing else is stored, which is why the EMA was the version that survived the era of calculating by hand.

The EMA cloud

A cloud is two exponential moving averages with the space between them filled in. That is the entire construction. A fast one and a slow one — eight and twenty-one are common on intraday charts — and the shading is there so your eye can read the relationship without comparing two lines.

What the fill genuinely buys you is one glance instead of two: which average is on top, how far apart they are, and whether the gap is opening or closing. When you are watching a chart for hours, «one glance instead of two» is not a small thing. That is the whole case for it, and it is a real one.

What it does not buy you is any new information. It is worth being blunt about the failure mode, because the cloud looks authoritative:

In a market that is going sideways, two averages of similar length cross each other repeatedly. The cloud flips, thins to nothing, flips back. None of those flips mean anything — they are the arithmetic of two similar numbers taking turns being slightly larger.

A cloud is a display choice, not a signal generator, and it inherits every gram of lag from the two averages underneath it. Treated as «which side of the recent average are we on, and is the market separating or coiling», it earns its space. Treated as a permission slip to enter, it will hand out that permission at the worst possible moments — the middle of a range.

MACD

MACD is subtraction. Take a fast EMA and a slow EMA — classically 12 and 26 — and plot the difference between them. Then take a 9-period EMA of that difference and call it the signal line. The bars people stare at, the histogram, are the distance between those two.

So MACD is not measuring price. It is measuring the gap between two smoothings of price, which is a rough way of asking whether the recent trend is pulling away from the older one or being reeled back in. Above zero: the fast average is above the slow one. Rising: they are separating. Falling toward zero while still above it: still up, losing conviction.

That last state is the one MACD is actually good at, and it is why it survives in a higher-timeframe context routine. It is much better at this is tiring than at this is starting.

What it does not tell you: anything about scale. A histogram twice as tall as last month's is not twice as meaningful — MACD has no fixed range, so its values are only comparable with its own recent history, on that instrument, on that timeframe. Comparing a MACD reading across two different markets is comparing two different units.

ADX and the DI lines

Welles Wilder published this in 1978 alongside RSI and ATR, and of the three it is the one most often misread. It is built in two steps.

First, two directional indicators. +DI measures how much of the recent movement was upward extension — today's high beyond yesterday's — and −DI does the same downward, each normalised by the average true range so different instruments are comparable. Those two do carry direction.

Then ADX. Take the difference between +DI and −DI, divide by their sum, smooth it. Because that starts with a difference that is then made positive, the direction cancels out.

ADX does not know whether price is going up or down. A strong downtrend and a strong uptrend produce the same high reading. It answers one question only: is movement here organised, or is it milling about?

Read that way it complements everything else on this page rather than competing with it. A moving average tells you which way; ADX tells you whether «which way» is currently a meaningful question. Low and flat is the genuinely useful state, because it says the trend tools around it are about to produce noise — and it tends to say so before the noise starts costing you.

The conventional thresholds — 20, 25 — are conventions, in exactly the sense the first page of this section describes. They were chosen for daily bars. On a five-minute chart of an index future they are a starting point and nothing more.

What none of the four tell you

• Whether anyone is actually there. All four are computed from closed bars. A clean crossover on thin overnight volume and the same crossover at the open look identical to the arithmetic.
• Where the move is likely to stop. They have no concept of a level. That lives in the volume profile and in the previous session's extremes.
• Whether the last leg was bought or sold into. A candle that closes high after being absorbed by a passive seller and one that closes high on genuine aggression produce exactly the same average. That difference is delta and the footprint, and it is a different instrument entirely.