TradingView Indicators

What an indicator is

An indicator takes prices, and sometimes volume, and applies a formula to them. The result is drawn over or under the price chart. A moving average is a line over the price. The family of trading indicators also includes oscillators drawn in a panel below the chart, bands around the price, and markers placed on single bars.

Every indicator is a summary of the past. It can be a useful way to see the past. It does not know the future, and a nicer picture does not change that.

Built-in, community and custom

• Built-in. These come with the charting platform. They are well known formulas such as moving averages, the relative strength index and bands. Their definitions are normally in the platform documentation, so you can read exactly what they calculate. Check the docs for the current list and the exact settings.
• Community. Other users write scripts and publish them for anyone to add to a chart. Quality varies. Some show their source code and some hide it. A published script has been written by a person with their own goals, and being popular or highly rated is not evidence that it works.
• Custom. You write the script yourself, or pay someone to write it. Many platforms use a small scripting language for this. See Pine Script for how that works. The advantage is that you can read every line.

How to read an indicator script critically

If the source is visible, read it before you trust the picture. If the source is hidden, you cannot check any of what follows, and that alone is a reason to treat the output as unverified. Look for four things.

1. Repainting

A script repaints when the value it shows for a past bar changes later. The common cause is that the script uses the live price of a bar that has not closed yet. While the bar forms, the price moves, so the indicator moves, and a marker can appear and then vanish. On the finished chart only the final state is left, which makes the history look cleaner than it ever was in real time.

A made-up example. The rule is: signal on when the live price is above the average of the two earlier closes and the live price. The earlier closes are 102 and 101. The same bar is shown at four price updates.

At update 1 the average is (102 + 101 + 103) / 3 = 102.00, and the price of 103 is above it, so the signal is on. At the close the price is 100, the average is 101.00, and the signal is off. A viewer looking at the finished chart sees no signal on this bar. A trader watching live saw one twice. A bot that acts on the live value would have acted on a signal the finished chart never shows.

2. Lookahead

Lookahead means the calculation uses information that did not exist yet at that moment. A typical case is a script that pulls data from a higher time frame, such as a daily value shown on an hourly chart. If it reads the daily high or close at the start of the day, it is using a number that was only known at the end of the day. The chart then looks as if it called the move perfectly. Platform docs describe settings that control this, so check them for the script's settings. The general trap is covered in look-ahead bias.

3. Hidden parameters

An indicator has inputs: the length, the multiplier, the source price, the smoothing method. Some are in the settings box and some are fixed in the code. Two scripts with the same name can use different defaults, and a different default can change every signal. Write down every input you use. A result that works only for one exact length is a sign the setting was fitted to the past, which is overfitting.

4. What the code does on each bar

Look for any use of future bars, any shift that moves a value backwards in time, and any value that is rewritten after the bar closes. If you cannot tell, test it live on a demo for a while and compare the markers you saw as they formed with the markers on the finished chart.

Why a pretty chart is not a tested rule

When you scroll back through history with an indicator on, the markers often sit near turning points. That is partly because you are looking at them with hindsight, and partly because of repainting and lookahead. You also notice the hits and forget the misses, and you may have scrolled through many settings and many instruments until one looked good. This is the same selection problem as in overfitting.

An indicator also has no exit, no position size and no cost. It does not say how much to risk, when to leave if wrong, or what the spread and fees take on each trade. Without those it has no result you can measure. A line crossing another line is an event on a chart, not a trade.

From indicator to alert

An alert watches a condition and sends a message when it becomes true, for example when the price crosses a line. That can notify you, or it can be sent to a program that places an order. Alerts raise their own questions. Does the alert fire the moment the condition is true on a live bar, or only once the bar has closed? An alert that fires on a live bar can fire and then be wrong at the close, exactly like the table above. Read TradingView alerts for the details, and check the platform docs for the trigger options.

From indicator to strategy

A strategy adds the missing pieces: an entry rule, an exit rule, a position size, and often a stop. It can then be run over past data to produce a list of trades. The page on strategies explains the parts, and the strategy tester page explains what its report does and does not show. A report from a tester is a backtest. It depends on the assumptions for fees, slippage and how orders fill inside a bar, so check those settings. A backtest is evidence about the past only.

A short checklist before trusting any rule built on an indicator

• Can you read the source, and did you read it for repainting and lookahead?
• Is every input written down, and does the result survive small changes to them?
• Does the rule have an exit, a size and a stop?
• Are costs included, and are they realistic for your broker or exchange? They vary, so check yours.
• Was it tested on data it was not tuned on, and then forward tested on a demo with no money at risk?

What this page does not tell you

It does not say which indicator is good, and it makes no claim about any named script. It does not say any indicator works. It shows how to read one without being fooled by the picture.