First published: 13 August 2019 · Reviewed and updated: 23 September 2026
This is a strategy page: exact rules for trading forex with exponential moving averages, which EMA to use for which style, and — unusually — what the most popular EMA signal actually did on real EUR/USD data. If you first want to know what an EMA is and how it is calculated, start with our guide to what the exponential moving average is.
Quick answer: which EMA strategy should I use?
Which EMA: 9 and 21 for intraday charts, 20 and 50 for 1-hour and 4-hour swing trading, 50 and 200 on the daily chart for trend direction.
The strategy that holds up: trade only in the direction of the daily 200 EMA, wait for a pullback into the 20–50 EMA zone on the 4-hour chart, enter on a 4-hour close back beyond the 20 EMA, stop beyond the swing and the 50 EMA, and exit on a close through the 50 EMA.
What to avoid: an unfiltered crossover. On EUR/USD monthly data from 2022 to 2026, a 6/12 EMA crossover lost on 5 of its 6 positions.
EMA strategy /ˌiː.emˈeɪ ˈstræt.ə.dʒi/ noun · forex trading
A set of trading rules that uses exponential moving averages to decide trend direction, entry, stop placement and exit. The EMA itself is only a line; the strategy is the rules around it.
Synonyms: EMA trading strategy, moving average strategy, trend-following system. Main types: crossover, pullback, price-cross, EMA stack. Related: crossover /ˈkrɒs.əʊ.və/, trend filter, whipsaw, Golden Cross.
Used in a sentence:
- “My EMA strategy only takes longs while the daily chart is above a rising 200 EMA.”
- “The 9/21 crossover whipsawed all morning because the hourly EMAs were flat.”
- “I skipped the setup: the ECB was due in twenty minutes, and the strategy has a news gate.”
1. Which EMA to use for your style
The single most searched question about EMA strategies is “what EMA should I use?”. The honest answer is that the period matters less than the timeframe it is applied to and the rules around it. These are the conventions most traders use — and conventions have one real advantage: when many traders watch the same line, price tends to react around it.
| Style | Chart | Trend filter | Entry EMAs | Typical hold |
|---|---|---|---|---|
| Scalping | 1–5 min | Hourly 50 EMA | 9 / 21 | Minutes |
| Intraday | 15 min – 1 hour | 4-hour 50 EMA | 9 / 21 or 12 / 26 | Hours |
| Swing | 4 hour | Daily 200 EMA | 20 / 50 | Days to weeks |
| Position | Daily | Weekly 50 EMA | 50 / 200 | Weeks to months |
There is no evidence that each currency pair needs its own “optimal” EMA periods, and pair-specific settings found by searching past data rarely survive live trading. Pick the style first, then use the standard periods for it.
2. What crossovers did on real EUR/USD data
Most EMA strategy guides show a hand-picked chart where a crossover catches a big trend. Here is the opposite exercise: every signal a classic crossover gave, on official data, with nothing removed.
The data is the monthly average EUR/USD rate published by the Federal Reserve Board (H.10 release) from January 2022 to August 2026. On it we calculated a 6-month and a 12-month EMA and recorded every cross.
Why it failed — and what that teaches
- Ranges destroy crossovers. From mid-2023 to late 2024 EUR/USD moved sideways between roughly 1.06 and 1.11. The EMAs flattened and crossed three times, each a loss.
- Trends are caught late. The bullish cross of April 2025 arrived after EUR/USD had already risen 8.5% from its January low. The one trend the rule did catch, it caught in its second half.
- Headlines point the wrong way. The Fed’s cut in September 2024 looked like a signal to sell dollars. The dollar rose instead, and the EMAs gave a bearish cross in November 2024 — just before EUR/USD bottomed and began a 13.4% rally.
A correction to the previous version of this page. It stated that after the Fed’s late-2024 cuts, EUR/USD held its daily 200 EMA and formed a Golden Cross confirming a bullish trend. The official data shows the opposite: EUR/USD fell 6.7% between September 2024 and January 2025. We have replaced that example with the data above.
Monthly averages are used because they are an official, reproducible series; they are not tradable prices, and a crossover on daily or hourly data would give different dates. The point is not the exact percentage but the pattern — lag and whipsaw — which appears at every timeframe. That pattern is why every rule in the strategy below exists.
3. The strategy: trend-filtered pullback
This strategy keeps what EMAs do well — defining the trend — and avoids what they do badly: signalling turns. It does not try to catch the start of a trend. It waits for a trend to exist, then buys its pullbacks.
Long rules (reverse everything for shorts)
- Market
- Major pairs with tight spreads: EUR/USD, GBP/USD, USD/JPY, AUD/USD.
- Filter
- Daily close above the 200 EMA and the 200 EMA higher than it was 10 days ago. If the 200 EMA is flat, the strategy is off.
- Setup
- On the 4-hour chart, price pulls back into the zone between the 20 EMA and the 50 EMA.
- Trigger
- A 4-hour candle closes back above the 20 EMA. Enter at the next open.
- Stop
- Below the pullback low and below the 50 EMA, whichever is further.
- Size
- So that the distance to the stop, plus the spread, costs a fixed share of the account — commonly 1% or less.
- Target
- Take half at twice the risk (2R), move the stop to entry.
- Exit
- Close the rest on a 4-hour close below the 50 EMA.
- News gate
- No new entries from 30 minutes before a Fed, ECB, BoE or BoJ decision until the press conference ends.
What this strategy gives up is clear: it will never catch the first leg of a trend, and it will sit out ranges entirely. What it gains is that the trades most exposed in Fig. 2 — those taken while the longer-term average was flat — are exactly the ones the filter is designed to skip.
4. The intraday version: 9/21
The 9/21 crossover is the most popular intraday EMA signal. It works under a narrow set of conditions, so the rules are mostly about restricting when it is used.
9/21 intraday rules
- Chart
- 5 or 15 minutes.
- Filter
- Hourly price above a rising 50 EMA for longs, below a falling one for shorts.
- Session
- London open to the end of the London–New York overlap only.
- Trigger
- 9 EMA closes above the 21 EMA, in the filter’s direction.
- Stop
- Beyond the last swing on the entry chart.
- Exit
- Opposite cross, or at the end of the session — no overnight holds.
- Cost check
- If the spread is more than about a fifth of the target, skip the trade.
5. Four EMA strategies compared
| Strategy | Signal | Works best | Main weakness | Essential filter |
|---|---|---|---|---|
| Crossover | Fast EMA crosses slow EMA | Strong, lasting trends | Late entries; repeated losses in ranges | Higher-timeframe trend direction |
| Pullback | Price returns to a sloping EMA and resumes | Established trends with orderly retracements | Misses the start of trends | Slope of the trend EMA |
| Price-cross | Price closes through one EMA | Fast breakouts after consolidation | Very frequent false breaks | Close-only trigger and volatility filter |
| EMA stack | Several EMAs fan out in order (e.g. 9 > 21 > 50) | Confirming trend strength | Confirms late by design | Use as a filter, not an entry |
6. Central bank decisions and the news gate
EMA strategies depend on trends, and in forex trends come mainly from the expected path of interest rates. That makes central bank decisions the most important events for any EMA trader — and the most dangerous moments to act on a short-timeframe signal.
September 2026: four decisions in nine days
| Date | Decision | Lesson for an EMA strategy |
|---|---|---|
| 10 Sep 2026 | ECB raises its deposit rate 25 bp to 2.50% | A second hike in a cycle that began in June — a trend already visible on the daily EMAs before the day |
| 16 Sep 2026 | Federal Reserve raises 25 bp to 3.75–4.00%, 12–0 | First US hike since 2023. Whether it changes the EUR/USD trend is a question for the daily 200 EMA over the coming weeks, not the release candle |
| 17 Sep 2026 | Bank of England holds at 3.75%, 6–3 | No rate change, three votes to hike: a “hold” can still move sterling. The news gate applies to holds too |
| 18 Sep 2026 | Bank of Japan raises to around 1.25%, 7–2 | A hike, yet the yen weakened because it was fully priced. A trader following the EMA rather than the headline would not have bought yen |
Two episodes, one lesson
In September 2024 the Fed cut and the dollar rose. In September 2026 the Bank of Japan hiked and the yen fell. In both cases the obvious reading of the headline was wrong, because the market had already priced the decision and was trading what came next. An EMA does not read headlines — it reports what price actually did — and that is exactly why it is useful after a decision, and useless in the minutes during one.
For the policy language behind these moves, see hawkish vs dovish monetary policy.
7. Risk and position sizing
An EMA strategy decides where the stop goes. Position sizing decides how much that stop costs. Keep them separate: the stop sits at a structural level — beyond the swing and the EMA — and the position size is whatever makes that distance cost your fixed risk.
position size (lots) = amount at risk ÷ (stop distance in pips × pip value per lot)
On a $10,000 account risking 1%, a 40-pip stop on EUR/USD with a pip value of $10 per standard lot gives $100 ÷ (40 × $10) = 0.25 lots. A wider stop simply means a smaller position; the money at risk is the same. To work out pip value on any pair, see what a pip is worth.
Why this matters for EMA strategies in particular: they have long losing streaks by design, because they sit through ranges and give back part of every trend. Six losses in a row is normal. Sizing that cannot survive that is not sized correctly.
8. How to test your own version
- Write the rules so precisely that two people would take the same trades. “Strong trend” is not a rule; “200 EMA higher than 10 days ago” is.
- Split the data. Design on one period, test on another you have not looked at.
- Include a range. Any trend strategy looks good on 2025 EUR/USD. Test it on 2023–2024 as well.
- Include costs — spread and commission on every trade.
- Record every trade, not a sample. Count the longest losing streak and the deepest drawdown, not just the total.
- Do not optimise the periods. If 23/57 beats 20/50 on past data, that is almost certainly noise. Standard periods are more robust than tuned ones.
9. What the research says
- Neely and Weller, in a Federal Reserve Bank of St. Louis working paper, found that technical trading rules can earn economically significant excess returns in foreign exchange, and attributed this to strong, persistent trends in exchange rates. They also found that adding central bank intervention data did not improve out-of-sample results.
- Brock, Lakonishok and LeBaron (1992), in the Journal of Finance, found that buy signals from simple moving-average rules on long-run stock data were followed by higher returns than sell signals — one of the studies that made moving-average rules a serious research subject.
Both findings come with the caveats in this page: the edge depends on trends persisting, it is measured before costs, and published edges tend to shrink once they are widely known. The EUR/USD example in section 2 is a reminder of what happens when trends do not persist.
10. Pre-trade checklist
- Filter passed: daily close on the correct side of a sloping 200 EMA.
- Not a range: EMAs separated and fanned out, not flat and tangled.
- Setup present: price in the 20–50 EMA zone on the 4-hour chart.
- Trigger confirmed: a closed candle back beyond the 20 EMA, not an intrabar touch.
- News gate clear: no Fed, ECB, BoE or BoJ decision or major data release inside the next 30 minutes.
- Stop placed beyond the pullback low and the 50 EMA.
- Size calculated so the stop plus spread costs your fixed risk.
- Exit written down before entry: 2R partial, 50 EMA close for the rest.
11. Common mistakes
| Mistake | What goes wrong | Fix |
|---|---|---|
| Trading every crossover | Lag and whipsaw, as in Fig. 2 | Only trade in the direction of a higher-timeframe filter |
| Trading in a range | Flat EMAs cross repeatedly, each time at a loss | Switch the strategy off when the filter EMA is flat |
| Trading the headline | A cut can strengthen a currency; a hike can weaken it | News gate, then read the higher-timeframe EMA |
| Stops in pips, not at structure | Stopped out inside normal pullbacks | Stop beyond the swing and the EMA; adjust size instead |
| Optimising periods | Settings that fit the past and fail live | Use standard periods; test out of sample |
| Too many EMAs | A reason can be found for any trade | Two or three EMAs, each with one job |
| Ignoring costs on low timeframes | A paper edge disappears in the spread | Skip trades where the spread is a large share of the target |
12. Frequently asked questions
What is the best EMA strategy for forex?
There is no single best strategy, but the most robust structure combines three things: a higher-timeframe trend filter, a pullback entry and a fixed exit rule. A common version uses the daily 200 EMA to decide whether to look for longs or shorts, waits for price to pull back into the 20 to 50 EMA zone on the 4-hour chart, and enters on a 4-hour close back beyond the 20 EMA. A plain crossover with no filter is the weakest version, because it lags every turn and whipsaws in ranges.
What EMA should I use for forex trading?
Match the EMA to your holding period. For intraday trading on 1 to 15 minute charts the 9 and 21 EMAs are standard. For swing trading on 1-hour and 4-hour charts the 20 and 50 EMAs are common. For trend bias on the daily chart the 50 and 200 EMAs are the reference. These periods are conventions rather than optimal values; their main advantage is that many traders watch the same lines.
Do EMA crossovers work in forex?
Only in trending markets, and even then they arrive late. On EUR/USD monthly averages from 2022 to 2026, a 6/12-month EMA crossover that was always in the market lost on five of its six positions: it was whipsawed in the 2023 to 2024 range and entered each trend after much of the move had happened. Crossovers are more useful as a confirmation of trend direction than as a stand-alone entry signal.
How do I avoid false EMA crossover signals?
Use a higher-timeframe filter and only take crossovers in its direction, for example longs only while the daily chart is above a rising 200 EMA. Ignore signals when the EMAs are flat and tangled, which marks a range. Require a closed candle rather than an intrabar cross, and do not act on short-timeframe crosses around central bank announcements.
What is the 9 and 21 EMA strategy?
It is an intraday crossover strategy: buy when the 9 EMA closes above the 21 EMA and sell when it closes below. It works best during the London and New York sessions, when spreads are tight and moves are directional, and only in the direction of a slower filter such as the hourly 50 EMA. Outside those conditions it produces frequent small losses.
What is the 50 and 200 EMA strategy?
It uses the relationship between the 50 and 200 EMAs on the daily chart to define the long-term trend. The 50 crossing above the 200 is called a Golden Cross and below it a Death Cross. Because both averages are slow, the signal arrives weeks or months after the turn, so it is better used as a filter that decides which direction to trade than as an entry trigger.
Where should I put my stop loss in an EMA strategy?
Place the stop where the trade idea is proven wrong: beyond the most recent swing point and beyond the EMA you are trading from, such as the 50 EMA in a pullback setup. Then size the position so that the distance to the stop costs no more than your fixed risk per trade. A stop placed at an arbitrary number of pips, rather than at a structural level, is the most common reason EMA trades are stopped out before the trend resumes.
How do central bank decisions affect EMA strategies?
They create both the sustained trends EMA strategies need and the two-way spikes that ruin short-term signals. The Federal Reserve cut rates by 50 basis points in September 2024 and the dollar still strengthened: EUR/USD fell 6.7 percent over the next four months and the monthly EMAs gave a bearish cross, not a bullish one. The practical rule is to avoid new entries around announcements and let the higher-timeframe EMA show whether the decision changed the trend.
Which timeframe is best for an EMA strategy?
Higher timeframes give fewer but more reliable signals, because spreads are a smaller share of each move and noise is lower. The 4-hour and daily charts suit most part-time traders. Lower timeframes can work, but costs, slippage and session timing matter far more, and a strategy that is profitable on paper at one-minute resolution often is not after costs.
Should I use EMA or SMA for a trading strategy?
Use the EMA when you want the average to react quickly to the latest prices, for example for entries and trailing exits. Use the SMA when you want a smoother, more stable line, for example as a long-term filter. Many strategies combine them. What matters more than the choice of average is having a filter, a defined exit and consistent position sizing.
Can I use an EMA strategy on gold or indices?
The mechanics are identical on any liquid market, including gold and stock indices, because the EMA only uses price. What changes is behaviour: gold reacts strongly to real interest rates and risk sentiment, and indices gap at the open. Test the same rules on each market separately rather than assuming settings that work on EUR/USD will transfer.
How do I backtest an EMA strategy?
Write the rules so precisely that two people would take the same trades, then test them on data the rules were not designed on. Include spreads and commission, record every trade rather than a sample, and check results across several years and at least one ranging period. Be wary of optimising the EMA periods: a combination that looks best in hindsight rarely stays best.
How many EMAs should I put on a chart?
Two or three is enough. A slow EMA for trend direction, a medium EMA for pullback entries and optionally a fast EMA for timing cover every role. Adding more lines rarely adds information and makes it easier to find a reason for any trade you already wanted to take.
Is an EMA strategy profitable?
It can be, but not automatically. Academic research on currency markets has found that trend-following rules have earned excess returns because exchange rates can trend persistently, yet the same rules lose money in ranges and after costs. Profitability depends on the filter, the exit rules, costs and discipline, not on the moving average itself.
13. Sources
Primary sources for the data and research on this page:
- Board of Governors of the Federal Reserve System — U.S. Dollars to Euro Spot Exchange Rate, monthly (H.10), series EXUSEU: fred.stlouisfed.org/series/EXUSEU
- Federal Reserve — FOMC statement of 16 September 2026: federalreserve.gov
- European Central Bank — key ECB interest rates: ecb.europa.eu
- Bank of England — Monetary Policy Summary and minutes, September 2026: bankofengland.co.uk
- Bank of Japan — monetary policy decisions: boj.or.jp
- Neely, C. J. and Weller, P. — Technical Analysis and Central Bank Intervention, Federal Reserve Bank of St. Louis Working Paper 1997-002: fraser.stlouisfed.org
- Brock, W., Lakonishok, J. and LeBaron, B. (1992) — Simple Technical Trading Rules and the Stochastic Properties of Stock Returns, Journal of Finance 47(5): doi.org/10.1111/j.1540-6261.1992.tb04681.x
The EMAs, crossover dates and segment results in section 2 are calculated directly from the Federal Reserve series above and can be reproduced in any spreadsheet. Figures current as of 23 September 2026.
Start with the indicator: the formula, the weights and a calculator in our guide to what the exponential moving average is.
Go further: filtering signals in moving average crossovers, and a full trend system in trend trading with moving averages.
Signal2forex.com - Best Forex robots and signals
