RSI indicator: how to read the levels and test signals on intrade.bar

RSI climbed above 70, you opened a downward trade, and the price kept rising. This familiar mistake starts with an appealing idea: if the market is “overbought,” it’s time for it to stop. But the RSI indicator doesn’t know what an asset should be worth, and it doesn’t receive a schedule of future reversals. It compares recent price gains and losses. Understanding that difference clears up half the questions about the colored line. Let’s go through the calculation, the settings and the levels, and then test a specific hypothesis on the intrade.bar demo: RSI returning from an extreme zone within a range defined in advance.

What RSI shows and what its number doesn’t contain

RSI stands for Relative Strength Index. It compares the positive and negative changes of a single price series. It is not necessarily a comparison of a currency with an index or other assets. If the indicator is built on the closes of the selected pair, it describes precisely the sequence of that pair’s closes. The 0-to-100 scale makes the result easy to read, but it doesn’t turn it into a probability forecast.

When smoothed gains outweigh losses, the line sits closer to the top of the scale. When losses outweigh gains, it sits closer to the bottom. A value of 70 does not mean a 70% probability of a rise. An RSI of 30 doesn’t promise that the price is on sale at a 70 percent discount. What you’re looking at is a ratio of changes mapped onto a particular scale; the formula doesn’t estimate fair value or the outcome at a future expiry.

The size of the moves counts too. Four gains of 0.1 and one drop of 1 are nothing like four gains of 1 and a drop of 0.1. The number of rising candles is the same, but their contribution to the calculation is different. That’s why RSI is best read alongside the price: one number helps condense the history, but it doesn’t show whether that history was a steady move or a few sharp jumps.

The RSI formula with a simple example

First, the previous close is subtracted from the current close. If the price rose by 2, the gains series gets 2 and the losses series gets 0. If the price fell by 1, you get 0 and 1. Then each series is smoothed. Let RS be the ratio of the smoothed average gain to the average loss. The indicator’s formula is RSI = 100 − 100 / (1 + RS). Both source series are non-negative: the minus sign of a drop is not carried into the denominator.

Say the average gain is 2 and the average loss is 1. Then RS = 2, and RSI = 100 − 100/3 ≈ 66.7. If both averages equal 1, RSI is 50. When losses are zero and gains are positive, the limiting value is 100; with zero gains and positive losses, it’s 0. A completely flat series, where both values are zero, needs a separate rule in the implementation.

In Wilder’s classic calculation, recursive smoothing is applied after the initial averaging. For a period of 14, the new average equals 13 parts of the previous average plus the new observation, divided by 14. This differs from a simple average of only the last 14 changes. So a manual spreadsheet with a rolling window may not match the terminal, even though the source prices are the same. You need to check both the formula and the starting history.

How a new candle changes the reading

Take the same averages of 2 and 1 and add a gain of 1. The new average gain becomes (2 × 13 + 1) / 14 ≈ 1.929, and the average loss (1 × 13 + 0) / 14 ≈ 0.929. Their ratio rises, and RSI ends up at about 67.5. The price rose by less than the previous average gain, the average itself fell, yet the indicator went up. There’s no contradiction: what matters is the proportion between the two values.

A strong drop in the past keeps influencing the calculation through smoothing. It doesn’t disappear entirely as soon as it falls outside the last 14 candles. This helps explain why three similar recent candles can come with different RSI values. Each sequence had its own backstory. If you compare readings across different windows, give the calculation enough data and make sure the source is the same.

On an incomplete candle, the latest close is effectively replaced by the current price. RSI can dip below 30, come back above it and finish the period at 35. On the final chart, the brief crossing is no longer visible. That’s why, for a first test, it’s convenient to use only closed candles. Intra-candle reactions can be studied too, but they need detailed data, not a collection of ready-made screenshots.

RSI settings: period, source and timeframe

The common period of 14 means 14 calculation periods. On M5, five-minute closes are used; on an hourly chart, hourly ones. Switching the timeframe changes the input data even if the number in the settings stays the same. A shorter period usually reacts faster to new changes and visits the extreme zones more often. A longer one smooths out swings but holds on to the influence of earlier price action for longer.

There’s no perfect setting for every asset and expiry. Start with one variant, for example RSI 14 on closes, and don’t change it after every unpleasant result. Otherwise you’ll be researching your ability to fit parameters rather than the behavior of one rule. Also write down any extra smoothing you use: a separate moving average on top of RSI changes the events you’re looking at.

intrade.bar has a TradingView chart. If the RSI variant you need is available in your set of tools, use it with the parameters written down. You can also do your analysis on a separate TradingView chart, but then you need to match the asset, the quote source, the time and the timeframe. The same pair name doesn’t guarantee identical extreme values on every candle. The trade is still settled using intrade.bar quotes.

Levels 70 and 30: the market’s state matters more than the label

“Overbought” and “oversold” are familiar names for the zones, but the market doesn’t have to obey them. In a sustained rise, positive changes can dominate for a long time, keeping RSI above 70. Trying to sell every such candle turns into an argument with an ongoing move. The indicator tells you the rise has already been strong relative to the declines; it doesn’t set the moment when buyers have to go to lunch.

In a range, the same zone can appear near the upper boundary, after which the price sometimes returns to the middle. That’s a different context. To test it, you first need stable areas of the range marked before the signal. If the price is already breaking out of them and making new highs or lows, the original idea of a return needs rethinking. You can’t keep the “sideways” label just because a downward trade is already open.

Distinguish entering a zone from leaving it. A cross above 70 is one event, a return below 70 is another. Dozens of candles can pass between them. A threshold of 80 instead of 70 reduces the number of events, but it doesn’t prove their quality. The settings, the context and the type of crossing are fixed in advance, not chosen after you’ve seen a nice reversal.

A practice test on intrade.bar: RSI leaving the zone

On the demo, choose one available currency pair and M5. Find a stretch with several returns to comparable upper and lower areas; write down the boundaries. If there’s no clear range, the exercise for now consists of observing. The goal is to test a specific situation, so the absence of a suitable market shouldn’t be made up for with an invented signal. For the practice series, we use RSI 14 on closes.

Consider the upward version: the price approaches a lower area marked in advance, RSI on a closed candle is below 30, and on the next or one of the following closed candles it returns above 30. Meanwhile, the price doesn’t close below the area according to your chosen cancellation rule. That will be the event to observe. Don’t open several trades just because the line stays above 30 for several candles in a row: record the first return after a separate visit to the zone.

Before a practice entry on intrade.bar, check the direction, the notional amount (the same for the whole series), the payout and the expiry. For example, decide in advance to test 10 minutes in Sprint, if that expiry is available. This is an arbitrary, testable horizon, not the “right” time that follows from RSI 14. Write down the actual entry price. intrade.bar has no early closing, so you can’t count on exiting if the indicator turns back.

Skip the signal if the range has broken, the return candle is still forming, the entry is noticeably late or the price has already covered most of the way to the upper area. Define “too far” measurably before the series — for example, the entry must stay in the lower third of the original range. This is a practice filter that also needs testing. For the downward direction, the rules are mirrored: the upper area, RSI above 70, then a completed return below 70.

What to record after expiry

Your record should link three layers: the state of the price, the RSI event and the option’s terms. Save the range boundaries, the signal time, the indicator value, the opening price, the expiry, the amount and the payout shown. After settlement, add the outcome and a note on whether you followed the rules. If the opening and closing prices are equal, intrade.bar refunds the amount; keep such episodes separate from wins and losses.

In intrade.bar statistics, you can view demo trades, assets and expiries separately, and export closed trades to CSV. Add the reason for entry and the RSI values to an external spreadsheet or notes. That way you won’t confuse a lucky profit with a confirmed hypothesis. A handful of winning trades can be inspiring, but for a conclusion it’s important to account for the whole series and test the rule on the next, previously unseen stretch.

With a notional 85% payout, a win on 100 units brings 85 in profit, and a loss takes 100. The break-even point with equal amounts, no refunds and no extra costs is 1 / 1.85 ≈ 54.05%. So in this example, a 53% win rate isn’t enough. In a real review, use the actual payout of each trade: the same RSI reading can come with different financial terms.

When RSI helps and when it adds false confidence

RSI is useful when you can finish the sentence: “In my rule, I need it to…” For example, to mark the first return from the zone after a pullback. If the condition is already fully described by price, check whether the result changes when you add the indicator. Three more similar oscillators won’t necessarily give three independent confirmations: they often process the same closes.

Divergence — a mismatch between price highs or lows and RSI — can be studied separately. But it’s not worth adding it to the current test in the middle of a series. First define which peaks you’re comparing and when they’re confirmed. Otherwise exceptions will start multiplying faster than observations. A good working setup is usually easier to explain than to defend with a dozen caveats after every loss.

Start on the intrade.bar demo with one scenario and one set of settings. Watch how the line behaves in a trend and in a range, then test the described event with a fixed expiry. The task is to learn to see the link between the RSI number and an actual decision. Once that link is clear, a high reading stops being scary and a low one stops promising an easy reversal.

Questions and answers

#індикатори