Why indicators lag
Every indicator is calculated from past prices and can therefore only describe what has already happened. Expecting more of it means misunderstanding the tool.
Learning objective: After this lesson you can explain why no indicator can predict the future.
An indicator is a formula applied to past prices. So it can't contain anything that wasn't already in the price. That's not criticism, it simply follows from how it's built.
Describing, Not Predicting
That gives the most important limit. An indicator describes a state, it doesn't predict anything. If someone tells you a signal predicted a move, they've got the order mixed up. The signal came because the move had already started.
The Lag
The second limit is the delay. Every bit of smoothing costs time. An indicator that's reliable reacts late. One that reacts fast gives lots of false signals. You can't have both at once, you can only choose.
Too Many Indicators
The third limit shows when you look at many indicators at the same time. If you have enough of them, you'll find confirmation for any decision. Then it's no longer analysis. It's a machine that supplies reasons for what you wanted to do anyway.
Technically, most indicators are filters running over the price series. A filter that may only use past values must lag. You can spread the delay out and shape it differently, but you can't get rid of it as long as you don't know future prices.
The Smoothing Trade-Off
That leads to a trade off every indicator has. Smooth more and you get fewer false signals and more delay. Smooth less and it's the other way round. If you look for the best setting on old data, you almost always end up with overfitting, because the best setting for the past depends on exactly the period you looked at.
Combining Without Gain
Combining several indicators looks like more certainty but adds no new information. They all come from the same price series and therefore move very similarly. A combination only makes you more certain if each part contributes something of its own, for example figures from the annual report and a trend signal. Several indicators from the same price don't.
Summary
- An indicator describes, it doesn't predict.
- Reliable and fast at the same time isn't possible, you can only choose.
- Many indicators give apparent confirmation without new information.
Did you get it?
Why can't an indicator contain anything about the future?
Because it's calculated only from past prices.
What trade off does all smoothing have?
More smoothing means fewer false signals and more delay, less smoothing the opposite.
Why does combining many indicators help so little?
Because they all come from the same price series and move very similarly, so they contribute nothing independent.
Check your understanding
Your task
Picture a 20 day average and a 100 day average. Write down which one gives more false signals and which one lags more, and explain why in one sentence. Then name one piece of information from outside the price series that you would need so that a combination really adds something of its own.
Sources and further reading
- The paper on the probability of backtest overfitting shows why optimizing parameters such as smoothing degrees against historical data regularly produces overfitting. View source ↗
Related
- Moving averagesStage 4
- RSI, MACD, and Bollinger BandsStage 4
- Why prices moveStage 1
- Matching questions for this stageQuestions