EMA Clouds: settings, formula and limits

EMA clouds: definition and what the cloud shades

The original source is the page of the 'Ripster EMA Clouds' indicator on TradingView, published on 9 March 2021 by the account ripster47, with its source code open. Its description presents the 'EMA Cloud System' as 'a Trading System Invented by Ripster where areas are shaded between two desired EMAs'. That is the whole definition: a cloud is the area between two averages.

The published code says what decides the colour: a single comparison. If the short average is greater than or equal to the long one, the cloud takes one colour; otherwise, the other. The change of colour is therefore exactly the crossover of the two averages, and the thickness of the cloud is the gap between them. By default, the two lines themselves are not plotted: only the area is visible.

Two points of vocabulary. This cloud is not the Ichimoku cloud, which is not built from exponential averages. And other scripts, by other authors, carry a similar name on TradingView and shade the gap between other pairs of EMAs: this lesson describes the one by ripster47. Page consulted in October 2026.

The author's published settings: five pairs, three displayed

What the code contains. Five clouds, each with its pair of lengths: 8 and 9, 5 and 12, 34 and 50, 72 and 89, 180 and 200. The first three are displayed by default, the last two are switched off. A menu lets you replace the exponential averages with simple ones. A release note from July 2023 states that these defaults are the ones the author uses.

What the description adds. It names the 5-12, 34-50, 8-9 and 20-21 combinations, adding that others are possible. The 20-21 pair is therefore not among the code's defaults, and the 72-89 and 180-200 pairs are not in the description. It also gives the intended timeframes: 10 minutes for day trading, 1 hour or daily for swings.

The reading its author suggests. The description attributes three uses to him: the 5-12 or 5-13 cloud 'acts as a fluid trendline for day trades'; 8-9 clouds 'can be used as pullback Levels', optionally; and price over or under the 34-50 clouds 'confirms either bullish or bearish bias'. It adds that the cloud area 'serves as support or resistance'. These are the author's claims: the page backs them with no measurement.

The starting price is not the close. In the code, the default source is the candle's median price, (high + low) ÷ 2, and not the close that a platform's moving averages normally use. A 12-period EMA added next to it, on closes, will therefore not follow the edge of the cloud exactly. Before comparing two screens, check the source, the lengths and the type of average.

How an EMA is calculated, in three steps

The script merely calls the platform's exponential average function. StockCharts' ChartSchool, on its moving averages page, describes the calculation in three steps. The first value is the simple average of the first N prices. The multiplier is 2 ÷ (N + 1). Then, on each candle: EMA = (price − previous EMA) × multiplier + previous EMA.

MetaTrader 5's help page on moving averages writes the same recurrence in another form: EMA = price × P + previous EMA × (1 − P), where P is the share given to the latest price. For the two averages of the 5-12 cloud, the multiplier is 2 ÷ 6, one third, and 2 ÷ 13, or 15.4%: the short average gives the latest price more than twice the weight, which is why it turns first.

Two practical consequences. An EMA depends on every value before it: StockCharts writes that the more data points you use, the more accurate it is, so two charts that do not load the same history do not show quite the same cloud on its first candles. And the cloud does not exist until the long average has received its N prices: twelve candles for the 5-12, fifty for the 34-50. The difference between a simple and an exponential average is covered in the article on moving averages.

A cloud worked by hand: the 5-12 over fourteen candles

The prices are invented. They are the median prices of the first fourteen candles of the diagram: 100.00, 100.60, 99.80, 100.40, 101.00, 100.20, 99.60, 100.00, 100.80, 100.40, 99.80, 99.40, then 100.60 and 102.00. Results are rounded to two decimals for display, but each step starts from the unrounded value.

The 5-period EMA. Its first value is the simple average of the first five prices: 501.80 ÷ 5 = 100.36, on candle 5. On candle 6: (100.20 − 100.36) ÷ 3 + 100.36 = 100.3067, that is 100.31. The same operation, repeated up to candle 12, gives 99.90 (99.9032).

The 12-period EMA. It only exists from candle 12: 1,202.00 ÷ 12 = 100.1667, that is 100.17. The cloud therefore starts there, and it starts red: the short average, at 99.90, is below the long one, at 100.17. The gap is −0.26.

Candle 13, median price 100.60. EMA 5 = (100.60 − 99.9032) ÷ 3 + 99.9032 = 100.14 (100.1355). EMA 12 = (100.60 − 100.1667) × 2 ÷ 13 + 100.1667 = 100.23 (100.2333). The short one is still below, by 0.10: the cloud stays red, but it thins.

Candle 14, median price 102.00. EMA 5 = (102.00 − 100.1355) ÷ 3 + 100.1355 = 100.76. EMA 12 = (102.00 − 100.2333) × 2 ÷ 13 + 100.2333 = 100.51. The short one moves above, by 0.25: the cloud turns green. The median price is then at 102.00, that is 2.60 above its low, 99.40, set two candles earlier. The signal comes after the rise has started, never before.

Reading the cloud over fifty-two candles

The rise. The cloud stays green from candle 14 to candle 31, and it thickens almost without a break during the rise: the gap between the two averages peaks at 3.15 on candle 26, the candle with the highest median price, 114.60. A thick cloud says that recent prices are far from the average of older prices. It does not say the trend will continue: it is thickest on the very candle where the trend stops.

The reversal. Price falls from candle 27. The cloud thins, the gap drops to 0.06 on candle 31, and the colour changes on candle 32, when the median price is at 106.00: six candles after the top, and 8.60 lower. Someone who bought when the cloud turned green and sold when it turned red, at the median price and with no costs, would have made 106.00 − 102.00 = 4.00, on a rise that travelled 15.20 from low to high.

The market with no trend. From candle 35 to candle 52 (shaded area), the median price swings between 103.40 and 109.00. The red cloud pinches almost to nothing: the gap is down to 0.14 on candle 38, and 0.01 on candle 45. It turns green on candle 46, with a median price of 108.80, 0.20 from the top of the area, then turns red again two candles later, at 105.20. The same round trip loses 3.60, in an area only 5.60 wide.

None of these readings is an order to buy or sell, and the example proves nothing but the mechanism. A rule built on a cloud is to be tested on your market, your timeframe and your costs, with the method in the article on how to backtest a trading strategy. Tradoshi does not record clouds at the time of your trades: they are read on your chart. The journal is for the next step: you tag every trade taken on this reading, for example '5-12 turned green', and the dashboard filters by tag to show what it produced for you.

What the research says about moving average crossovers

No published study, to our knowledge, measures EMA clouds under that name. The success rates that circulate about them could not be traced to a publication, and this lesson repeats none of them. What has been studied is the family the cloud belongs to: moving average crossover rules.

The reference study is Brock, Lakonishok and LeBaron, Simple Technical Trading Rules and the Stochastic Properties of Stock Returns (Journal of Finance, 1992). It tests two rules, moving average and trading range break, on the Dow Jones index from 1897 to 1986, and its abstract concludes that the results provide strong support for the technical strategies. The population is specific: a single US equity index, over ninety years.

Seven years later, Sullivan, Timmermann and White, in Data-Snooping, Technical Trading Rule Performance, and the Bootstrap (Journal of Finance, 1999), take the same data to correct for data-snooping bias, which consists in keeping, after the fact, only the rules that worked. They widen the universe from 26 rules to 7,846 and extend the index to 1996. From 1897 to 1986, certain rules did outperform the benchmark, even after the adjustment. From 1987 to 1996, the ten years the original study could not have seen, this is no longer the case: the authors find scant evidence there that these rules were of any economic value, and none on the S&P 500 futures contract from 1984 to 1996.

Three gaps separate these tests from a cloud on your screen. The averages in them are arithmetic, that is, simple: the second study says so in a footnote, and neither covers exponential averages. The data is daily, whereas the indicator's author aims first at the 10-minute chart. And the market is a US equity index: none of this carries over as it stands to a single stock, to forex or to cryptocurrencies.

The limits of EMA clouds

The cloud adds no information to its two averages. Its colour is the sign of the gap between them, its thickness is the size of that gap. It is the calculation of a MACD line, short average minus long average, set to other lengths and drawn on price instead of in a separate pane. The shading makes the crossover easier to see. It makes it neither earlier nor more reliable.

An average lags by construction. StockCharts writes it: moving averages are based on past data, so they lag behind price, and the longer the average, the more the lag. In the example, the colour changes six candles after the top, 8.60 lower. Shortening the lengths reduces that lag and multiplies the colour changes: no setting removes the trade-off.

Without a trend, false signals multiply. The same StockCharts page warns that moving average crossovers produce relatively late signals and, when there is no strong trend, many whipsaws. In the shaded area of the diagram, the only turn to green comes almost at the high and is cancelled two candles later. The 8-9 pair is the most exposed: two lengths that close give two averages that almost coincide, and cross at the slightest gap.

Three clouds are not three confirmations. The 8-9, 5-12 and 34-50 clouds are calculated on the same price series, with the same formula. When they are all the same colour, they describe one thing, a rise or a fall already in place, at three speeds. The page on technical analysis explains why two tools from the same family do not confirm each other.

The support or resistance role is not measured. The indicator's description asserts it, with no figure. In the example, the median price enters the cloud on candle 28 and leaves it through the bottom on candle 30, without stopping there. Nor does a cloud say anything about stop distance, which is reasoned in volatility with the ATR, or about the exit, which the lesson on the Parabolic SAR covers.

Frequently asked questions