Moving averages: what they can and cannot tell you

A moving average is the average price over the last N periods, recalculated each period. That is all it is. It smooths out noise so you can see the direction underneath, and because it is an average of the past, it always lags the present. Both facts follow from the arithmetic, and together they define exactly what a moving average is good for: seeing context. They also define what it is bad for: telling you when to act.

Simple, exponential, and why it barely matters

A simple moving average weights every period equally. An exponential one weights recent prices more heavily, so it turns faster.

People spend a surprising amount of energy on this choice. In practice, on a daily chart, the difference between a 50-period simple and a 50-period exponential average is small and will not decide whether your strategy works. Pick one, use it consistently, and put the energy somewhere it matters, like your exits.

The period length matters more than the type. A 20-period average tracks price closely and changes direction often. A 200-period average barely moves and describes the long-term picture. Neither is correct. They answer different questions.

What are moving averages genuinely useful for?

Seeing the trend without arguing with yourself. Price above a rising average is an uptrend. Price below a falling one is a downtrend. This sounds trivially simple and its value is precisely that it is not open to interpretation. On a chart you have been staring at for twenty minutes, having one objective definition of trend prevents you from finding the answer you want.

A directional filter. Only taking long trades when price is above the 200-period average is a crude rule that removes a lot of low-quality trades. It will also keep you out of some good ones. That is the trade.

Dynamic context in a strong trend. In an orderly trend, price often pulls back toward a rising average before continuing. That gives you an area to watch. Note the word area, not signal.

A trailing reference. Some traders exit when price closes decisively through an average. It is a mechanical rule that lets winners run, at the cost of giving back some of the move.

What can a moving average not do?

They cannot predict. An average of the last fifty days contains no information about tomorrow beyond what the last fifty days already told you. Anyone describing a moving average as a forecasting tool is confusing description with prediction.

They cannot work in a range. This is the important limitation. In a sideways market, price crosses the average constantly, every crossing looks like a signal, and every one of them fails. A method built on crossings will bleed steadily in exactly the conditions that occur most of the time.

They do not create support. The idea that the 50-period average is holding price up gets it backwards. Price is not respecting a calculated line. What is often happening is that a genuine level sits nearby, or that a trend is orderly enough for the average to trail behind it. When people talk about a stock bouncing off its 50-day, the causation is usually a lot weaker than the phrasing suggests. Real levels come from where people actually transacted, which is the subject of support and resistance.

They lag, always. By construction. A moving average tells you a trend changed after it changed. That is acceptable when you are using it for context and expensive when you are using it as a trigger.

The golden cross problem

Two averages crossing, a shorter one moving above a longer one, gets a lot of attention. It is worth understanding why it disappoints in practice.

The signal is the product of two lagging indicators. By the time a 50-period average crosses a 200-period one, a substantial part of the move has already happened. The signal is real, in that it does describe a change in trend, but it describes it late, and it produces frequent false starts when the market is not trending.

You can trade crossovers profitably if the risk to reward is set up to absorb a low win rate, in the same shape as breakout trading described in breakout or pullback. What you cannot do is treat the crossover as a reliable prediction, which is how it is usually presented.

How many, and which periods?

One or two. Not five.

Five moving averages on a chart is not analysis, it is decoration, and it guarantees that at any moment some of them support any conclusion you might want to reach. That is the general problem covered in how many indicators do you actually need.

The common periods, 20, 50, 200, are common mostly because they are common: enough people watch them that they attract attention. That is a real if circular reason to prefer them over a number you optimised in a backtest, because a period that works only at exactly 63 is almost certainly curve fitting.

I use one longer average as a trend filter and nothing else. Its job is to answer a single yes or no question before I look at anything: is this instrument in an uptrend or not. Everything after that comes from price and levels.

Using one properly

The honest use is as a filter and a context tool, never as the trade itself:

1. Check the average first, before you look for a setup, so it filters rather than confirms. Looking at it afterwards means you will interpret it to agree with what you already want to do.
2. Let it veto, not initiate. If the average says downtrend, you skip your long setup. If it says uptrend, that permits the setup, it does not create one.
3. Take the actual entry from price, at a level, with a stop derived from structure.
4. Do not put your stop at the average. It moves every day, which means your risk changes daily for reasons unrelated to your thesis.

That ordering keeps the moving average doing the one job it is good at, and keeps the parts it is bad at in the hands of price.

The full framework, including where the trend filter sits in my weekly routine, is in The 9-to-5 Trader.

Educational only, not financial advice. Trading carries a real risk of loss.