Charts & market structureIndicators9 min readBeginner friendly

Moving Averages: One Dial, Not a Menu of Settings

Key takeaway

The video's closing line is the one to keep: moving averages "don't predict future performance — they only confirm established trends." Everything else is a detail about how late that confirmation arrives.
Smoothness and lateness are one dial, not two settings. 20 against 200 days, simple against exponential — every one of those choices buys fewer false signals with later ones, or the reverse. There is no setting that gives you both.
The sell rule as stated is missing its condition. "Price rises to the moving average and bounces" is only a sell signal when the average is sloping down — applied in an uptrend it sells every normal pullback.

Learning pathBuild a trend-following system: ride winners, cut everything elseStep 3 of 9

Read before this:Dow Theory: The Six Rules Every Chart Method Is Built On

Based on a clip by Charles Schwab (@CharlesSchwab) — YouTube

Watch the original

A moving average is a summary of the recent past, plotted

Add up the last N closes, divide by N, plot the result. Tomorrow, drop the oldest one and add the newest. That is the entire construction.

The video's description is exact and worth restating because most people carry a vaguer idea around: "we'll take each day's price and add them together. Then we'll divide that number by our time frame number." Then, crucially, "each day we'll drop the last day in the time frame and add today's."

Two consecutive days of a ten-day average. Only two of the twelve closes changed membership — one left, one joined — and the whole line moved on the strength of that swap. This is why a long average barely reacts to a single day: one new number out of two hundred cannot move a mean very far.

That mechanism explains the property the whole tool rests on. The line "smooths out" price, in the video's phrase, because averaging is exactly what removes variation. But it is worth being precise about what has been removed: not noise specifically, just variation, of which noise is one kind and a genuine turning point is another. The average cannot tell them apart, and neither can you at the moment it happens.

The three signals it names, and the condition the third one drops

Two buy signals and one sell signal are offered. The buys are stated with their conditions attached. The sell is not.

The three states the video says a moving average helps you identify: up, down, and sideways. It is worth noticing that the third one is where most of the trouble lives — the signals below all assume the average has a slope, and a flat average is where they misfire.

Break above an upward-sloping average. "When a price breaks above an upwardly sloping moving average, this could mean it's a good time to buy a stock." Note the slope condition is stated explicitly.
A support bounce. "This is when the security's moving average acts as a support level for the price. When the price comes down to the moving average and then rallies up again, this bounce could be used as a buy signal."
The sell mirror. "If the stock's price rises to the moving average and bounces, this might be a sell signal." Introduced with the words "on the flip side" — and the slope condition, present in the first signal, is not repeated.

That omission matters more than it looks. A price rallying into its moving average and turning down is a meaningful event in a downtrend, where the average sits above price and acts as resistance. In an uptrend the price touches its 20-day average constantly on the way up — that is what the second signal, the support bounce, is describing. The two descriptions are almost the same words for opposite conclusions, and the only thing separating them is which way the line is sloping.

Reading it without the slope

Price touched the average, so sell
Every pullback becomes an exit
You sell the strongest trends first

Reading it with the slope

Average sloping down: rally into it is resistance
Average sloping up: pullback to it is support
Flat average: the signal means nothing

So before applying any of the three, answer one question: which way is the line pointing? If the answer is "sideways", none of these signals is telling you anything, and the next section explains why that case produces the most signals of all.

Timeframe is a dial between false signals and late ones

The video presents 20, 50 and 200 days as three options. They are three positions on a single trade-off, and you cannot escape it by picking a different number.

The same price with all three drawn over it. The 20-day tracks every bend and will cross price often; the 200-day is nearly straight and will cross it rarely. Neither is more accurate — they are answering questions about different lengths of time.

The mechanism for the false signals has a name in the video: "A whipsaw is when the stock crosses over the moving average, giving one signal, and then reverses quickly, giving the opposite signal." And the observation that follows is correct — "a short-term time frame, like 20 days, usually shows more whipsaws".

A flat average and a price oscillating around it. Six crossings, six signals, and the price finishes where it began — so every one of those signals cost a spread and a commission and returned nothing. This is not a rare failure case; it is what a sideways market does to any crossing rule.

Where the video stops short is in describing the other side of the ledger. It says the longer averages "show a smoother average and have fewer buy and sell signals", and that the investor "may stay in the trade longer" — both true, and both stated as advantages. The cost is not named: a 200-day average will not confirm a change of trend until the change is months old.

The video does add one genuinely useful idea here: "the short-term average is confirmed by the intermediate- and long-term averages." Using a long average as a context filter — only take short-term signals that point the same way as the 200-day — is a different and better use than treating any single average as a trigger. It costs you nothing and halves the number of decisions you have to make.

Lag, and the one line worth sharpening

Every point in the window counts the same, so a violent day is diluted by however many quiet ones sit beside it.

The video's explanation is the right one: "Because each period is given equal weight, day 50 counts as much as day one. As you can see, a large gain or drop hardly factors into a moving average for some time. This is known as lag."

A sharp fall against a 50-day average. The average keeps rising for a while after the peak, because the days dropping out of the back of the window are lower than the ones that came before them. By the time the line turns down, the fall it is reporting has already happened.

Then comes the sentence to sharpen: "To avoid lag, consider using another analysis, such as a weighted moving average or exponential moving average." The description of how they work is accurate — "these types of moving averages consider recent data the most relevant and give it more weight". The word doing the damage is avoid.

How each type weights the days in its window. The exponential average is not looking at anything the simple one cannot see; it is looking at the same days and caring more about the recent ones. That makes it turn sooner — and it makes it turn sooner on days that turn out not to matter.

An exponential average does not remove lag. It reduces lag by increasing sensitivity, which is the same dial from the previous section under a different name. A 20-day EMA behaves roughly like a shorter simple average: earlier signals, more whipsaws. If lag were removable by reweighting, the limit case — weight only today — would be the price itself, which has no lag and no signal either.

What it is actually good for

Use it to describe the state you are in, not to decide when to act. The video says as much, and then places that sentence last.

"It's also important to note that moving averages don't predict future performance — they only confirm established trends." That is the most useful sentence in four minutes, and it arrives after every signal has already been demonstrated. Read in the other order, the whole thing reframes: these are not entry triggers with a known success rate, they are descriptions of what has already happened, arriving with a delay you chose when you picked the window.

Question	Can it answer it?
Which way has price been going lately?	Yes — this is what it is for
Is today's move large relative to the recent range?	Indirectly, via the distance from the line
Has the trend changed?	Only after the fact, by the length of the window
Will the trend continue?	No — nothing in the calculation addresses this
Where should my stop go?	It can host one, but it does not size it

What a moving average can and cannot do for you

That last row is the practical bridge. A moving average is a reasonable place to hang a trailing exit — it moves with price and it is unambiguous — but it says nothing about how much you should have on the line in the first place. That decision comes from where to set a stop loss and when to move it, and the position size it implies comes from cutting losses and position sizing.

And if you want to know whether any of these signals work on the things you actually trade, the video cannot tell you — no clip can. Pick one rule, one window, one instrument, and record 30 occurrences before you form an opinion. Writing the rule down first is the part that makes the record mean anything, which is what a written trading system is for.

Try this week

Open one chart and add the 20, 50 and 200-day averages. Before looking at price, write down what each one says about direction — they will not always agree.
Find a sideways stretch on that chart and count how many times price crossed the 20-day average. That count is your whipsaw rate for that instrument.
Find the last sharp reversal and measure how far price moved before the 50-day average changed direction. That distance is what lag costs you.
Write the slope condition into your own words for both the buy and the sell version of the bounce signal, so you never apply one in the other's market.
Pick one rule and one window, then log the next 30 signals — including the ones you did not take — before deciding whether it has an edge.

Common questions

What is a simple moving average and how is it calculated?

It is the mean of the last N closing prices, plotted as a line and updated each period by dropping the oldest price and adding the newest. A 20-day simple moving average adds the last twenty closes and divides by twenty. Because every price in the window counts equally, one new day can only move the line by a small fraction of the difference between the price entering and the price leaving.

Should I use a 20, 50 or 200-day moving average?

It depends on how long you intend to hold, not on which one is more accurate. A 20-day average reacts quickly and produces many signals, a large share of which reverse straight away. A 200-day average produces few signals and each arrives months after the move began. Many people use a long average as a direction filter and a short one for timing within it, which is a better use of the two than picking a single winner.

What is a whipsaw in trading?

A whipsaw is when price crosses the moving average, giving one signal, then immediately reverses and gives the opposite signal. It is the normal behaviour of any crossing rule in a sideways market, and it costs you the spread and commission on every round trip. Shorter averages produce more of them, which is the price you pay for their earlier signals.

Does an exponential moving average remove lag?

No. It reduces lag by weighting recent prices more heavily, and that same weighting makes it react to moves that turn out not to matter. Reducing lag and reducing false signals are opposite ends of one trade-off — no weighting scheme gives you both. Advice to switch to an EMA to avoid lag is accurate about the benefit and silent about the cost.

Can moving averages predict where a stock is going?

No, and the better explainers say so directly: they confirm trends that are already established rather than forecast new ones. A moving average is a summary of prices you have already seen, so it contains no information that was not already on the chart. Any predictive value you assign to it is really a belief that recent direction tends to persist, and that belief is what needs testing on your own instruments.