Backtest evidence

Is buying RCI under -80 profitable, and does the three-line reading beat one line?

Is RCI Below -80 a Buy? 2,484 Backtests on USD/JPY

Finding

The textbook reading came last of the five. Buying under -80 and closing when the line returns to the level left 4 of 72 settings profitable in both 2024 and 2025 on the 15-minute chart, 3 of 72 on the hourly and 9 of 72 on the 4-hour. Reading the same threshold the other way, buying over +80, gives 39, 33 and 6 of 72. The three-line reading had 19 of 27 settings in profit in 2025 and 6 of 27 in 2024; 4 cleared both years and none of the nine 4-hour settings did. RCI sits at or past -80 on 6.99% to 17.69% of bars and at or past +80 on 9.56% to 31.31%, so the extreme is not rare. RCI's ranking beat the plain change over the same closes in 17 of 36 cells.

Key results

Fade settings profitable in both years
4 of 72 on 15-minute, 3 of 72 hourly, 9 of 72 on 4-hour
The same threshold read as momentum
39 of 72 on 15-minute, 33 of 72 hourly, 6 of 72 on 4-hour
Three lines aligned
4 of 27 profitable in both years; 0 of 9 on the 4-hour chart
How often RCI is past ±80
At or under -80 on 6.99% to 17.69% of bars; at or over +80 on 9.56% to 31.31%
Ranking against the plain change
RCI's set won 17 of 36 cells; the two sets overlap by 50.31% to 70.83%
Win rate of the textbook fade
52.83% to 86.96%, and most of those years still ended negative
Last year's winner, this year
Three of six landed 205th or 206th of 207

Scope

  • Instrument: USD/JPY
  • Period: 2024-01-01 to 2024-12-31 and 2025-01-01 to 2025-12-31, with 2025 also split in halves
  • Timeframes: 15-minute, 1-hour, 4-hour
  • Conditions: 5 readings x 6 periods x 3 to 5 levels = 207 settings

Method

RCI is Spearman's rank correlation between a bar's place in the window and its close's place among the sorted closes. Five readings were measured: the fade (past the level is a reversal), the follow (past the level is momentum), a break out of ±level, a crossing of the level, and short, mid and long lines all past the same side. Periods 9, 14, 21, 26, 36 and 52; fade and follow levels 70, 80 and 90 with give-back at 0 and 40; break levels 50, 70, 80 and 90; crossing levels -80, -40, 0, 40 and 80; triples 9/26/52, 9/14/21 and 14/26/52 at levels 50, 70 and 80. The control is the percentage change across the same closes RCI ranks (ROC with a lookback one bar shorter), taking the same number of bars from the low-change end. Trades run through Formiq's backtester at the close of the signal bar, a fixed 0.3 pip spread, 0.1 lots and no slippage. 207 settings x 3 timeframes x 4 windows = 2,484 runs. Every figure is checked in research/rci-verify.run.ts.

Limitations

  • One pair over two years.
  • Fills are at the close with a fixed spread and no slippage, so live results will differ.
  • The three-line reading takes 2 to 45 trades a year on the 4-hour chart, and figures from counts that small move a lot.
  • The control is one measure of magnitude, not an inventory of everything RCI discards.