Understanding Technical Indicators

What an indicator measures — and what it doesn't

A technical indicator is a number calculated from price, and sometimes volume, that compresses recent history into something you can compare across companies. That is the whole of it. It is arithmetic applied to a price series.

The word "indicator" does a lot of unearned work here, because it suggests something is being indicated — that the number knows where the price is going. It does not. A moving average tells you what the average closing price has been over some window. Whether that means anything about tomorrow is a separate question, and one no indicator answers.

So why use them? Comparability. You cannot look at five hundred charts and form a consistent opinion. You can, however, compute the same number for five hundred companies and put them in order. An indicator turns "this one looks strong" — a judgement that changes with your mood and the last thing you read — into a measurement that does not.

That is the honest case for technical indicators, and it is narrower than the usual one. They do not see the future. They make a large number of companies comparable on a defined basis, which is what a systematic strategy needs in order to be systematic at all.

One consequence worth holding on to: an indicator is an input to a rule, never an instruction. "The 50-day average is above the 200-day" is a fact. What to do about it is a decision you make when you design the strategy — and the indicator has no opinion on it.

The families NiveshBuddy actually offers

The builder exposes three kinds of price-based condition, and two controls that decide how whatever qualifies is ranked.

Moving averages smooth. A 50-day average responds to a change in price direction within weeks; a 200-day average takes months. Neither is better — a shorter window notices real changes sooner and also reacts to noise that turns out to be nothing. Note these are simple moving averages. NiveshBuddy does not offer exponential or other weighted variants, so do not go looking for them.

Super Trend is a trend filter whose option names carry the trade-off openly. Fast (More Reactive) confirms a trend sooner and will confirm some that do not last. Slow (Stronger Confirmation) waits for more evidence and enters later, by which point some of the move has happened. There is no setting that is early and certain.

It works on the way out as well as the way in, but only on rebalance day. It is a filter, not a stop loss: it is not checked daily between rebalances. At each scheduled rebalance, a held stock whose trend has turned against it fails the filter and is sold along with the rest of that rebalance's trades. The trade log says so as "Supertrend (10, 2) stop-loss hit", the numbers being the filter's settings (Fast is 10, 2; Balanced 10, 3; Slow 10, 4). Only the Fixed and Trailing Stop Loss are checked on every trading day's close. So a faster setting means more holdings failing the filter at a rebalance, not exits between rebalances.

Rate of change is momentum stated plainly: the percentage move over a chosen lookback. A 3-month ROC and a 12-month ROC can rank the same companies in almost opposite orders, because they are answering different questions — "what has moved recently" versus "what has been moving for a year". Choosing the lookback is choosing what you mean by momentum.

Despite "Filters" in its name, ROC does not remove anyone from the list — it orders it. Tick one window and companies are ranked on that window alone. Tick several and the universe is ranked once per window, then each company's ranks are averaged, and the lowest average comes first (ties go to the longest window). So ticking 3- and 12-month asks for companies that look strong on both horizons at once, not for companies that pass two separate tests.

The 52-week high filter is a breakout idea in eligibility form. Rather than detecting the moment of a breakout, it asks whether a stock is currently trading near its one-year high — within 20%, say — which is a way of restricting a strategy to companies already near the top of their own recent range.

Ranking Method decides what "best" means once the filters have done their work. Absolute Returns orders by raw return. Volatility Adjusted Returns orders by return relative to how much movement produced it, which will favour steadier climbers over erratic ones. The same universe, ranked both ways, gives you two different portfolios — so this is a genuine design decision, not a formatting preference.

Where volume fits, and where it doesn't

If you have used other tools you will expect volume indicators — volume oscillators, volume-confirmed breakouts, and so on. NiveshBuddy has none of these. There is no volume indicator in the builder, and no volume-confirmation option attached to the other conditions.

Volume reaches a strategy in exactly one way: inside the Liquidity filter, which is computed as 20-day average volume × price. That gives an average daily traded value in rupees, and the strategy uses it to decide whether a company is tradable enough to consider at all.

The averaging window is the part worth understanding. Because it looks at twenty trading days rather than one, a single unusually busy session does not make a thinly traded company eligible — and a company whose activity has quietly dried up stops qualifying without anyone needing to write a separate rule for it.

The distinction to keep clear: Liquidity is an eligibility question, not a signal. It does not tell you a stock is worth holding. It tells you that if the strategy does want to hold it, you can probably trade it near the price you see.

What happens when you stack conditions

Every condition you add sounds like an improvement. Each one is also a filter, and filters multiply.

The following is illustrative — the numbers are chosen to show the shape of the problem, not measured from any particular universe.

ROC is missing from that table on purpose: ticking more ROC windows changes the order of what qualifies, not how many qualify. The narrowing comes from the eligibility conditions.

The last row is the problem. With eight qualifying companies, the strategy is no longer expressing a general idea about markets — it is describing the handful of stocks that happened to satisfy an increasingly specific set of tests over the period you tested. Add another condition and the historical result will very likely improve again, which is precisely what makes this so easy to do.

Two things are worth separating here. A condition that expresses your idea is doing work: if your thesis is about companies in a confirmed uptrend, a trend filter belongs. A condition added because the backtest looked better with it is doing something else entirely — and you usually cannot tell the difference from the result alone.

That is the beginning of overfitting, and it is a large enough topic to deserve its own treatment. Designing Robust Investment Strategies takes it properly: what curve fitting is, why a strong historical CAGR is not evidence on its own, and how to test whether a strategy survives conditions it was not built on.

Common misconceptions

• "An indicator predicts price." It measures what has already happened. Any predictive value comes from your assumption that a pattern persists — an assumption the indicator neither makes nor supports.
• "More indicators mean a better strategy." Each additional condition narrows the eligible set and improves the fit to history. Past a point you are describing the past rather than expressing an idea.
• "A moving average tells you when to buy." It tells you what an average has been. The buying rule is something you wrote; the indicator has no view on it.
• "Fast settings are better than slow ones." Fast (More Reactive) and Slow (Stronger Confirmation) trade earliness against reliability in opposite directions. Which suits depends on your holding period, not on which sounds more capable.
• "I can screen on fundamentals too." You cannot. NiveshBuddy's strategy rules are price-based. There is no P/E, ROE, ROCE, earnings, revenue or debt filter, and no article, video or strategy should suggest otherwise.

Key takeaways