All posts
RSIindicatorsbacktestingUSD/JPYmean reversiontrend following

Best RSI Settings, Tested 330 Ways on USD/JPY

RSI run as a 30/70 fade and as a 50-line cross: 330 settings on USD/JPY, 2024 and 2025. The fade won 65% of its trades and lost money; the cross won 24% and made it.

RSI compares how much price rose against how much it fell over the last N bars and turns that into a number between 0 and 100. Almost every description of it says the same thing: below 30 is oversold, above 70 is overbought. A second, quieter reading treats 50 as the dividing line — above it the buyers are ahead, below it the sellers are.

The first is a mean-reversion tool. The second is a trend-following one. The same line produces opposite rules, and the usual advice does not say which reading the settings belong to.

So both were measured side by side, on the same pair, the same years and the same costs. On USD/JPY in 2025, 150 fade settings run long and short separately (6 periods × 5 entry levels × 5 exit levels) plus 30 midline-cross settings (6 periods × 5 levels) — 330 in all, on the fifteen-minute, hourly and four-hour charts, then repeated over 2024.

The short version: on the hourly chart the fade won 65.3% of its trades on average and survived both years in 31 of 150 settings. The midline cross won 23.7% and survived in 17 of 30. Not one cross setting won more than three trades in ten.

20%30%40%50%60%70%80%90%−2k−1k0+1k+2kwin rate →net pips for the year
buy the oversold line (98)sell the overbought line (112)cross the midline (30)one-hour chart, 2025, 10+ trades
One line, two ways to read it. The blue cluster never wins more than three trades in ten, and 25 of its 30 settings still finished the year in profit. The red one wins six in ten and nearly half of it finished below zero.

What RSI actually measures

Take the last N bars. Average the size of the up moves, average the size of the down moves, and express the up average as a percentage of the total. Period is the only knob.

  • Nothing but up bars gives 100
  • Nothing but down bars gives 0
  • Up and down in balance gives 50

So 50 is the point where gains and losses cancel, and distance from it measures how lopsided the recent bars have been. "Overbought" and "oversold" are an interpretation laid on top of that; the arithmetic only reports the imbalance.

A longer period does not reach 30

This is where the practical problem starts. Averaging over more bars mixes the ups and downs together and pulls the number toward 50. Counting every hourly bar of 2025 on which RSI was defined:

PeriodBars below 30Bars above 70Lowest value all yearHighest
713.97%14.13%1.097.0
910.81%11.20%2.494.5
145.73%6.54%8.089.5
211.80%4.17%16.484.7
300.39%2.03%24.980.8
500.00%0.32%31.274.7
03050701007914213050RSI period
everything the line touched all yearwhere eight bars in ten satvertical lines at 30 and 70; 6,226 hourly bars, 2025
Period 50 never printed a value below 30 all year. An oversold rule on a slow RSI is a rule that never fires.

A period-50 RSI never once closed below 30 in all of 2025. Its lowest reading was 31.2. "Use a longer period for cleaner oversold signals" describes something that does not happen — the signals do not get cleaner, they stop existing. On the four-hour chart both period 30 and period 50 stayed above 30 for the entire year.

The sweep shows the consequence directly: on the four-hour chart, 50 of the 150 long-fade settings never took a single trade, and 20 did nothing on the hourly chart either.

Putting RSI on a chart

PlatformHow
MT4 / MT5Relative Strength Index in the Navigator's indicator list; set the period
TradingViewSearch the indicator list for RSI and set the period
Browser (Formiq)On the chart by default; the backtester switches between the level reading and the level cross with a trigger

Drawing it is the same everywhere. The decision that matters is whether the rule fires when the line reaches a level or when it crosses one — this article treats those as two separate systems.

How this was measured

Enter the settings below and the same numbers come out.

ItemValue
PairUSD/JPY
Window2025-01-01 to 2025-12-31, with 2024 run identically for comparison
TimeframesM15 / H1 / H4
Bars tested24,903 on M15, 6,226 on H1, 1,610 on H4 in 2025
Fade, longBuy the close of the bar where RSI drops below the entry level; sell the close of the bar where it rises above the exit level
Fade, shortThe long reflected around 50: sell above 100 − entry, cover below 100 − exit
Midline crossBuy when RSI closes through the level from below, sell when it closes through from above, reversing on the same bar
Stops and targetsNone (measured in their own section)
Spread0.3 pips, no slippage, 0.1 lots, charged on entry and exit
Fade grid6 periods (7, 9, 14, 21, 30, 50) × 5 entry levels (20, 25, 30, 35, 40) × 5 exit levels (50, 55, 60, 65, 70) = 150, run long and short separately
Cross grid6 periods × 5 levels (40, 45, 50, 55, 60) = 30
MeasurementRun through Formiq's backtester; pips recomputed from each trade's own fill prices

The long and short fades are separate systems because a level rule points one way. "Buy below 30" and "sell above 70" produce entirely different results.

The win rate points the wrong way

SystemTimeframeProfitable 20252025 medianProfitable 20242024 medianBoth yearsMean win rate
Fade longM1577 / 150+5 pips67 / 150−45 pips29 / 15064.6%
Fade longH178 / 150+11 pips58 / 150−65 pips31 / 15065.3%
Fade longH481 / 150+95 pips36 / 150−211 pips13 / 15067.6%
Fade shortM1542 / 150−264 pips33 / 150−367 pips5 / 15057.7%
Fade shortH148 / 150−137 pips23 / 150−727 pips9 / 15061.2%
Fade shortH484 / 150+272 pips17 / 150−1,075 pips1 / 15061.0%
Midline crossM158 / 30−347 pips12 / 30−408 pips4 / 3022.3%
Midline crossH125 / 30+830 pips22 / 30+846 pips17 / 3023.7%
Midline crossH47 / 30−885 pips28 / 30+1,388 pips6 / 3023.0%

(Mean win rate covers the 2025 settings that took at least one trade. The median is over every setting on that timeframe.)

The figure at the top plots the hourly rows one point each, leaving out settings that traded fewer than ten times — win rate across, net pips up. The red and purple fade settings spread across a 40–90% band; the blue cross settings sit in a narrow 17–30% band. The medians still come out at +830 pips for the cross, +11 for the long fade and −137 for the short.

Selling the overbought line is the weakest of the three, surviving both years in 9 of 150 hourly settings. USD/JPY rose 1,632 pips through 2024, and a rule that sells strength gets run over in a year like that.

The same knob points opposite ways

The clearest result in the sweep is what the period does.

0%25%50%75%100%7914213050RSI period →
the oversold fade (25 settings each)the midline cross (5 settings each)share of settings profitable in both 2024 and 2025, hourly chart
Both curves peak, at opposite ends of the same knob: the fade at period 30, the cross at period 9.

Settings profitable in both 2024 and 2025, hourly chart:

PeriodFade longMidline cross
72 / 254 / 5
93 / 255 / 5
147 / 253 / 5
217 / 252 / 5
308 / 252 / 5
504 / 251 / 5

Both curves peak, and they peak at opposite ends of the knob. The fade is at its best at period 30 (8 of 25); the cross at period 9 (5 of 5), falling to 1 of 5 by period 50.

The mechanism follows from the reach table. A fade waits for a rare extreme, so a longer period raises the quality of each signal — right up to the point where the extreme stops arriving. Period 50 averaged three hourly trades a year. That is why the fade peaks at period 30 and falls away on either side: two forces meet there.

The cross is the opposite. Every crossing is a sample. Period 9 crossed 775 times a year on the hourly chart, period 50 only 231, and the smaller count leaves the year's result at the mercy of a handful of trades.

Levels behave more simply:

Fade entry levelBoth years
206 / 30
2513 / 30
304 / 30
352 / 30
406 / 30
Cross levelBoth years
403 / 6
455 / 6
504 / 6
554 / 6
601 / 6

The fade's exit level survived best at 70 (9 of 30, against 6 at level 50 and 5 at 65). The back half of "buy at 30, sell at 70" holds up in this test even where the front half does not.

What the standard 14 / 30 / 70 did

The most-quoted settings, run both ways. The cross column uses the same period 14 with the level at 50.

ReadingTimeframe20252024
Fade long (30 → 70)M15105 trades, 63.81% win, +332.1 pips93 trades, 75.27% win, +9.0 pips
Fade long (30 → 70)H125 trades, 60.00% win, −250.7 pips22 trades, 63.64% win, +26.1 pips
Fade long (30 → 70)H49 trades, 66.67% win, +1,148.4 pips7 trades, 85.71% win, +209.2 pips
Fade short (70 → 30)M15106 trades, 62.26% win, +435.5 pips94 trades, 50.00% win, −1,545.0 pips
Fade short (70 → 30)H125 trades, 64.00% win, −256.1 pips22 trades, 40.91% win, −1,429.7 pips
Fade short (70 → 30)H49 trades, 66.67% win, +1,059.2 pips8 trades, 50.00% win, −1,150.7 pips
Cross the 50 lineM152,840 trades, 22.64% win, −265.7 pips2,820 trades, 21.99% win, +94.9 pips
Cross the 50 lineH1638 trades, 24.14% win, +1,272.0 pips681 trades, 22.17% win, +666.0 pips
Cross the 50 lineH4164 trades, 27.44% win, −790.1 pips141 trades, 23.40% win, +1,798.5 pips

Three of those nine rows came through both years: the long fade on M15 and H4, and the cross on H1.

The 2024 fifteen-minute row is the one worth staring at. 93 trades, 75.27% of them winners, +9.0 pips for the year. Three winners in every four, and what is left over after a year is nine pips.

On the hourly chart the standard fade fires 25 times a year. There is not much to say about a sample that size. A 60% win rate there means fifteen winners and ten losers.

The hourly cross took 638 trades for +1,272.0 pips, then 681 for +666.0 pips. The count is two orders of magnitude larger, and the year-to-year swing correspondingly smaller.

Why the higher win rate loses

The two sides of the trade are different sizes. Averaged over every 2025 setting:

SystemTimeframeMean winMean lossMean bars held
Fade longH1+88.6 pips−120.8 pips188.9
Fade shortH1+80.7 pips−129.6 pips273.4
Midline crossH1+71.5 pips−20.5 pips21.6

The cross wins 3.5 times what it loses. At a 23.7% win rate, 0.237 × 71.5 − 0.763 × 20.5 still comes out positive. It rides a move as long as the line stays on one side of the level and reverses on the same bar when it does not, so a wrong call is cut early.

The fade does the reverse: its win is 0.73 of its loss. It collects small profits when price returns to the exit level and takes a large one when price does not come back. The holding times differ just as sharply — 188.9 hourly bars for the fade, about eight days, against 21.6 for the cross.

The same relationship holds on the other timeframes. The long fade ran +49.6 / −73.1 on M15 and +126.2 / −196.2 on H4; the cross ran +36.8 / −10.4 and +120.1 / −42.8.

Carrying last year's best setting forward

SystemBest of 2024Its 2024 resultApplied to 20252025 rank2025 median
Fade long, H1period 14, 40 → 70+1,139.8 pips (44 trades)−798.8 pips147 / 150+11.2 pips
Midline cross, H1period 9, level 45+3,601.3 pips (725 trades)+1,254.9 pips7 / 30+830.3 pips

The fade's 2024 winner finished 147th of 150 the following year. A 77.27% win rate over 44 trades and +1,139.8 pips became −798.8.

The cross's 2024 winner held 7th of 30 in 2025. Read the other way round, 2025's winner (period 9, level 40, +1,815.4 pips over 672 trades) made +1,003.6 pips in 2024 for 12th of 30. What carries over is the direction of the system, not the ranking of a setting.

Rank correlations sit at or below zero throughout:

System2024 ↔ 2025H1 2025 ↔ H2 2025
Fade long−0.002−0.202
Fade short−0.032−0.299
Midline cross−0.075−0.523

Filters, stops and targets

Two representative hourly settings with the usual additions bolted on.

ConditionFade long 14 / 30→70, 20252024Cross 14 @ 50, 20252024
Baseline25 trades, −250.7 pips22, +26.1 pips638 trades, +1,272.0 pips681, +666.0 pips
ADX ≥ 2025, −225.0 pips21, −99.2 pips290, +389.4 pips334, −827.0 pips
ADX ≥ 2525, −58.2 pips20, −164.8 pips163, +710.9 pips165, +94.6 pips
ADX ≥ 3022, −200.8 pips16, −403.2 pips79, +420.8 pips89, −152.8 pips
London + NY22, −561.0 pips20, +100.2 pips370, +1,669.9 pips400, −117.9 pips
Tokyo21, −113.6 pips18, +53.1 pips247, +284.3 pips236, +1,254.5 pips

The ADX filter left the cross worse off in both years. It does raise efficiency per trade — profit factor goes from 1.146 to 1.322 at ADX ≥ 25 — but the total falls from +1,272.0 to +710.9 in 2025 and from +666.0 to +94.6 in 2024. Filtering a trend system by trend strength did not help it.

The session filters reversed between years. London and New York took 2025 to +1,669.9 and 2024 to −117.9; Tokyo did the opposite. Either choice looks correct on one year's data.

Exit ruleFade long 2025Fade long 2024Cross 2025Cross 2024
None−250.7 pips+26.1 pips+1,272.0 pips+666.0 pips
SL 30 / TP 60−379.4 pips+72.0 pips+1,392.7 pips−343.9 pips
SL 50 / TP 100−256.9 pips+175.5 pips+1,947.4 pips−317.7 pips
SL 100 / TP 200−519.3 pips−737.3 pips+2,118.0 pips+755.1 pips
Time exit, 24 bars−536.7 pips−878.1 pips+2,960.5 pips−84.7 pips

Only the widest pair beat the baseline in both years for the cross (+2,118.0 and +755.1). Tight stops helped in 2025 and turned 2024 negative. The 24-bar time exit produced the best single number in the table and the worst kind of evidence for it: 2024 came out below baseline.

Nothing improved the fade in both years.

Cost lands differently on the two systems

SpreadFade long 14 / 30→70 (25 trades)Cross 14 @ 50 (638 trades)
0.0 pips−243.2+1,463.4
0.3 pips−250.7+1,272.0
0.6 pips−258.2+1,080.6
1.0 pips−268.2+825.4
1.5 pips−280.7+506.4
2.0 pips−293.2+187.4

At zero spread the fade still loses 243.2 pips. Going from 0 to 2.0 pips costs it 50.0 pips — 25 trades times 2.0. The finding that runs through this series, that a losing system is usually losing its trade count times the spread, does not apply here. The fade is not losing to costs; it is losing on the trades.

The cross is the case the identity does describe. It pays 191.4 pips at 0.3 (638 × 0.3), 13% of its +1,463.4 gross, and 1,276.0 pips at 2.0, which leaves +187.4. The same arithmetic held exactly in the Bollinger Band and moving average cross runs.

What this test supports

Only the findings that pointed the same way in both years:

  1. One line, two systems, opposite conclusions. On the hourly chart both years were profitable for 31 of 150 long-fade settings, 9 of 150 short-fade settings and 17 of 30 cross settings
  2. Sorting by win rate sorts the wrong way. Mean win rates were 65.3% for the fade and 23.7% for the cross; the size of each side cancels the difference out
  3. There is no best RSI period on its own. The fade wanted a long one (period 30, 8 of 25), the cross a short one (period 9, 5 of 5)
  4. A long period removes "oversold" rather than cleaning it up. Period 50 never printed below 30 on the hourly chart in 2025; its floor was 31.2
  5. The fade's losses are not a cost problem. At zero spread the standard hourly setting still finished at −243.2 pips

Choosing RSI mean reversion because it wins often is not supported by this test. The same inversion appears in other indicators, and the Bollinger Band article measures both sides within one indicator the way this one does. QQE smooths RSI and Stochastic RSI runs a stochastic over it; both were measured the same way. If you want to check this on your own pair and your own window, the conditions can be assembled without writing code.

Limits of this test

  • One pair, two years. Nothing here promises the same behaviour on another pair or another year
  • The two years were very different markets. 2024 rose 1,632 pips over a 2,237-pip range; 2025 finished 56 pips below where it started over a 1,900-pip range. The short fade's collapse in 2024 owes a lot to that trend
  • The fade holds until the exit level arrives, with no stop. A wrong call shows up at full size. Stops were measured separately
  • Many fade settings trade too rarely to judge. The hourly median is 29 trades a year, and 50 of the 150 four-hour settings never traded at all. Win rates and profit factors on counts like those are wide numbers
  • The cross grid holds only 30 settings, which affects every rank statistic computed from it
  • Entries and exits are taken at bar closes. Real fills differ
  • Spread is a flat 0.3 pips throughout. Real spreads move with the session and around releases
  • Volatility varies by hour, which the session-filter rows inherit

Questions people ask

What are the best RSI settings?
It depends entirely on which of the two RSI systems you are running, and the same knob points opposite ways in each. On the hourly USD/JPY chart, the oversold fade survived both 2024 and 2025 in 8 of 25 settings at period 30 but only 2 of 25 at period 7. The midline cross did the reverse: 5 of 5 at period 9, 1 of 5 at period 50. Choosing a period before choosing the system is choosing at random.
Is the standard RSI 14 with 30 and 70 any good?
On the hourly chart it fired 25 times in a year. It made −250.7 pips in 2025 from 25 trades at a 60.00% win rate, and +26.1 pips in 2024 from 22 trades at 63.64%. On the fifteen-minute chart in 2024 it won 75.27% of 93 trades and finished the year at +9.0 pips. A high win rate on that few trades says very little.
Does RSI mean reversion or trend following work better?
On the hourly chart, buying the oversold line survived both years in 31 of 150 settings and selling the overbought line in 9 of 150, while crossing the midline survived in 17 of 30. Mean win rates were 65.3% for the fade and 23.7% for the cross. The lower win rate finished ahead because its average win was 3.5 times its average loss.
Should I use a longer RSI period to get cleaner oversold signals?
A longer period does not produce cleaner oversold readings — it stops producing them. Across 6,226 hourly bars in 2025, RSI closed below 30 on 13.97% of bars at period 7, 5.73% at period 14, 0.39% at period 30 and 0.00% at period 50. The lowest value period 50 printed all year was 31.2.
How do I add RSI to a chart?
In MT4 and MT5, pick Relative Strength Index from the indicator list in the Navigator and set the period. TradingView has it under RSI in the indicator search. Formiq has it on the chart by default, and its backtester switches between the level reading and the level cross with a trigger.

Formiq is a free browser-based FX terminal with replay practice and no-code backtesting. Open the chart or see what the free plan includes.