Best indicator for intraday trading: what testing actually shows

Contents

1. Is there a single best indicator for intraday trading?
2. What should you check when you test a common intraday indicator?
3. Why does the same indicator win on one market and lose on another?
4. How do you test indicators on the same data with fees and slippage?
5. How do session times and the opening range affect intraday signals?
6. How do you know a winning indicator was not just lucky?
7. What can an indicator test tell you, and what can it not?
8. Frequently asked questions
9. References

Is there a single best indicator for intraday trading?

No. No single technical indicator works best for intraday trading in every market. Test results depend on the market, the bar size, the session you trade and, most of all, trading costs. The useful question is which indicator rule still holds after costs and on unseen data in your market. Only a fair test can answer that.

An intraday indicator is a calculation on bars shorter than one day, such as 1-minute or 5-minute bars, that turns recent prices or volume into a number you can trade on. Asking for the best indicator is like asking for the best tyre without saying whether the car drives on ice, sand or a race track. The tyre matters, but the road decides which one wins.

What should you check when you test a common intraday indicator?

Before you test a common intraday indicator, fix its settings and session, then check the trade count, the net average trade after costs and how its signals cluster around the open. Most intraday indicators measure one of three things: price against an average, the speed of a move, or the size of a move. For what each common indicator does, see our guide to the best indicators for day trading. The table below covers five widely used intraday indicators and lists only what to fix before a test and what to check in its results.

Indicator	Fix before the test	Check in the results
VWAP	The session open at which it resets	Whether every bar has volume data, and how many signals fall near the open
RSI	The lookback length and the entry levels	The trade count and the net average trade, because many small trades make costs weigh heavily
MACD	The fast, slow and signal lengths	How late the signals arrive on short bars compared with the move they try to catch
Moving averages	The lookback length, and simple or exponential	Whether the lookback spans the break between sessions
ATR	The lookback length and the multiple you use	ATR sizes risk but gives no direction, so test it as a stop, size or filter next to a directional rule

VWAP (volume-weighted average price) is the session's average traded price, weighted by how many contracts traded at each price. It usually restarts at each session open, so its value in a test depends directly on how you define the session. The other four work on any bar size, so the setting that matters most when you test them is the lookback length. The strategy generation in AlphaPrime (Valynox) takes that lookback length as a parameter range, so an indicator is tested over a span of values rather than one hand-picked number.

Why does the same indicator win on one market and lose on another?

The same indicator rule can win on one market and lose on another, because three things change underneath it: how the market moves, how large each bar is, and what each trade costs.

• The market. Some markets tend to keep moving after a break. Others tend to snap back. A reversal rule built on RSI and a trend rule built on a moving average need opposite behaviour, so they cannot both be "best" on the same market.
• The bar size. On 1-minute bars an indicator reacts to noise. On 30-minute bars the same indicator reacts to larger swings but gives fewer signals. A 14-bar setting covers 14 minutes on one chart and 7 hours on the other.
• The costs. Commission and slippage stay about the same per trade, whatever the bar size. Shorter bars capture smaller moves, so costs take a larger share of each trade.

The cost effect fits in one line: net average trade = gross average trade − round-trip cost. The gross average trade is the profit or loss per trade before any costs. The round-trip cost is commission, fees and slippage for both the entry and the exit. If the first number is not clearly larger than the second, the indicator has no tradable edge on that bar size, however good its chart signals look.

How do you test indicators on the same data with fees and slippage?

A fair comparison changes only the indicator and keeps everything else the same. If the market, data, exit, costs or execution differ between tests, you are comparing setups, not indicators.

1. Fix the data. Use one market, one bar size and one date range for every indicator.
2. Split the history. In-sample data is the part of history you use to choose and tune rules. Out-of-sample data is the part you keep aside and use only once, at the end. Decide the split before the first run.
3. Fix the session. Decide whether you trade the day session only or all hours, and when positions must be closed.
4. Fix the exit. Give every rule the same exit, for example a fixed stop, a fixed target and a flat exit at the session close. Then any difference comes from the entry.
5. Add costs from the first run. Slippage is the difference between the price the test assumes and the price you would actually get. Use commission plus at least one tick of slippage per side as a starting point.
6. Execute on the next bar. If a signal is known only when a bar closes, fill it on the next bar, not on the bar that created it.
7. Add a baseline. Test a simple rule with no indicator, such as entering at a fixed time each day. An indicator must beat the baseline to earn its place.

In AlphaPrime, backtesting applies the same data resolution, fees and slippage, next-bar execution and trading sessions to every candidate, so each indicator is judged under the same conditions.

How do session times and the opening range affect intraday signals?

Session handling changes intraday indicator values more than most traders expect. The open behaves differently from the rest of the day, and an indicator's lookback can quietly reach back into the previous session.

The opening range is the high and low of the first minutes of a session, for example the first 15 or 30 minutes. Volume and price range are usually largest near the open and the close and quieter in the middle of the day. An indicator tuned on calm midday bars can give very different signals in the first half hour. A breakout of the opening range is also a strategy idea in its own right, so a test can mix two effects without noticing.

Three session facts matter for common markets. US stock index futures on CME trade nearly around the clock from Sunday evening to Friday, with a daily one-hour break from 17:00 New York time. The NYSE cash session, which many traders use as the "day session", runs from 9:30 to 16:00 New York time. TAIEX futures on TAIFEX have a day session from 08:45 to 13:45 Taiwan time and a night session from 15:00 to 05:00 the next morning. Exchanges can change these hours, so confirm them on the exchange's contract page. The futures market hours tool shows each session in your own time zone.

Use these checks in every intraday indicator test:

1. Look at the lookback at the open. On day-session-only 5-minute bars, a 14-bar RSI at 9:40 uses two bars from today and twelve from the previous afternoon.
2. Check when VWAP resets. A VWAP that starts at the night session open and one that starts at the day session open give different values at the same moment.
3. Run the test twice. Run it once with all signals, and once with no new entries in the first 15 or 30 minutes. If most of the profit comes from the open, the edge belongs to the open, not to the indicator.
4. Close at the session end. If the strategy is meant to be intraday, force an exit before the close, so no position carries an overnight gap.

How do you know a winning indicator was not just lucky?

You know a winning indicator was not just lucky only when it also works on data that played no part in choosing it. If you try many indicators and settings, one of them will look good on any history by chance.

Think of a thousand people flipping coins. A few of them will very likely throw seven heads in a row. They are not skilled at flipping coins. Testing indicator variants works the same way. Bailey, Borwein, López de Prado and Zhu (2014) show that the more variants you try, the higher the best backtest result tends to be, even when no variant has any real edge. Sullivan, Timmermann and White (1999) tested thousands of simple technical trading rules on a long stock index history. They found that the rule that looked best on the older data did not keep its advantage on the later data they examined.

• Keep a held-back period untouched until your rules are final, and look at it once.
• Use walk-forward analysis. Walk-forward analysis repeatedly tunes on one window and tests on the next window, so every reported trade comes from settings chosen before that trade's data.
• Count your attempts. Write down how many indicators and settings you tried. A good result after five tries means more than the same result after five hundred. White (2000) describes a test built for exactly this question: is the best of many rules better than a benchmark once the search is taken into account?
• Prefer broad regions. If RSI 14 works but RSI 12 and RSI 16 lose money, you have found noise, not an edge.

AlphaPrime's strategy validation offers Basic, Standard and Extensive validation, walk-forward analysis and thresholds you set yourself.

What can an indicator test tell you, and what can it not?

An indicator test can tell you how a precise rule behaved on one market's history after costs. It cannot tell you that the rule will keep working.

A good test answers useful questions. Did this entry rule beat a simple baseline after realistic costs? Did it hold up on data it never saw? Does it depend on the open, on one year, or on one exact setting? It also gives you a fair ranking of ideas, because every idea faced the same conditions.

A backtest describes the past. It is not a forecast. Markets change their behaviour, costs change, and a few hundred trades leave many possible outcomes. Treat a rule that passes every check as a candidate worth watching with small size, not as proof that a strategy works.

What this means if you use AlphaPrime: you can describe several indicator ideas in strategy generation, with their parameter ranges and entry and exit conditions, then judge them all on the same data, costs and session before any of them earns a place in validation.

Frequently asked questions

What is the most accurate indicator for intraday trading?

No indicator is the most accurate in every market. Accuracy changes with the market, the bar size, the session and the trading costs. A rule that wins often can still lose money if its losses are larger than its wins or if costs eat its small gains. Test each indicator on your own market with costs and on out-of-sample data.

Is VWAP good for intraday trading?

VWAP is popular for intraday trading because it resets each session and shows where price is compared with the average traded price of the day. Whether a VWAP rule has an edge depends on the market and the costs, and it must be tested like any other rule. Check when your platform resets VWAP, because the reset time changes the signals.

Which bar size is best for intraday indicators?

No bar size is best for every strategy. Shorter bars give more signals but smaller moves per trade, so commission and slippage take a larger share. Test the same rule on two or three bar sizes with identical costs. Then compare the net average trade, not the number of signals.

How many indicators should I combine for intraday trading?

Fewer is usually easier to test honestly. Every extra indicator adds settings to tune, and more settings make it easier to fit noise in the past. Add an indicator only if it improves the out-of-sample result, not just the in-sample one.

References

• Bailey, D. H., Borwein, J., López de Prado, M., & Zhu, Q. J. (2014). Pseudo-Mathematics and Financial Charlatanism: The Effects of Backtest Overfitting on Out-of-Sample Performance. Notices of the American Mathematical Society, 61(5), 458–471.
• Sullivan, R., Timmermann, A., & White, H. (1999). Data-Snooping, Technical Trading Rule Performance, and the Bootstrap. The Journal of Finance, 54(5), 1647–1691.
• White, H. (2000). A Reality Check for Data Snooping. Econometrica, 68(5), 1097–1126.