RSI divergence compares selected price extremes with RSI values under a fixed pairing rule. A lower price low with a higher RSI value is regular bullish divergence; the opposite pattern at highs is regular bearish divergence. The chart location of a pivot and the moment it becomes knowable are different. This article follows both clocks before calculating a possible trade.

One decimal place can hide the direction of a difference

When comparing two pivots, read the values used by the rule. Equal labels do not necessarily mean equal source numbers.

Two RSI pivots and label precisionLow falls from 100 to 99; RSI from 33.04 to 33.03. One-decimal labels both show 33.0, but the second RSI trough is not higher.LABEL 33.0 DOES NOT REPLACE THE VALUELow:10099RSI:33.0433.03One-decimal display: 33.0 → 33.0.Data: 33.03 < 33.04, not higher.The condition checks data, not just labels.

Fictional, confirmed and correctly paired pivots: Low 100 and 99, RSI 33.04 and 33.03. Our rule requires Low₂ < Low₁ and RSI₂ > RSI₁. Round labels to one decimal; retain source numbers. We do not reconstruct RSI or a particular platform’s default precision. The document MQL5 Reference — DoubleToString confirms that number-to-text formatting allows selection of decimal places; the displayed label separately represents the value.

Vertical distances are schematic: the diagram shows difference direction, not a common price/RSI scale.

The first condition passes; the second fails: RSI changes by −0.01. Displaying 33.0 at both points does not justify recording a higher trough. A rule allowing equality would be a different definition. Record data precision and comparison operator with indicator parameters. This does not make a tiny difference a useful signal; it explains why the label is incomplete evidence of the condition.

What RSI divergence actually is

Divergence describes different directions in price and the indicator. Here we compare two consecutive, confirmed price lows and read RSI from those same bars. We do not independently search for RSI troughs. Requiring separate RSI pivots would be another rule, with its own confirmation delay. Regular bearish divergence uses higher price highs and lower paired RSI values.

RSI is computed from selected price changes and smoothing; in our model the input is Close and the period is 14. It is not an order-flow measurement and divergence does not identify a reversal date. The piece on how to read the RSI explains the indicator itself. Record input, period, initialization and comparison rules with each test.

Regular versus hidden divergence

Hidden bullish divergence conventionally describes a higher price low paired with lower RSI; hidden bearish divergence describes a lower price high paired with higher RSI. These patterns are often interpreted as continuation candidates within a separately defined trend. They do not ensure that the trend resumes. Keep their pairing and trend rules separate from the regular-divergence test.

Regular and hidden divergence have different geometric conditions. A strategy must separately specify which direction to trade and how to define the preceding trend. The article on hidden and regular divergence compares both. Mixing definitions during a test changes the strategy rather than providing comparable observations.

Why divergence on its own is a weak signal

Consider these hypothetical H1 Bid Lows: t1 1.0730; t2 1.0728; t3 1.0720; t4 1.0725; t5 1.0728; t6 1.0718; t7 1.0708; t8 1.0698; t9 1.0690; t10 1.0695; t11 1.0700. The strict two-bars-each-side rule finds t3 and t9. They become available after t5 and t11 respectively. Given RSI values of 26 at t3 and 33 at t9, the price Low falls 30 pips while RSI rises 7 points. These RSI values are assumed inputs; Lows alone cannot reconstruct a Close-based RSI.

The second Low does not determine its RSI value. For a separate one-step calculation, suppose prior smoothed gain G = 0.0003, loss L = 0.0006 and prior Close = 1.0700. With n = 14, update G′ = (13G + max(ΔClose,0))/14 and L′ = (13L + max(−ΔClose,0))/14; RSI = 100G′/(G′+L′). A bar with Low 1.0690, Open 1.0700 and High 1.0705 can close at 1.0698, 1.0700 or 1.0702. The resulting RSI values are approximately 32.7731, 33.3333 and 34.4538. Against an earlier assumed RSI of 33, the first fails a strict higher-RSI condition and the latter two pass. This is one update from given state, not a reconstruction of all prior bars.

How to add confirmation and run the trade

Step 1 — spot the divergence first, then wait

In the example the pair becomes known after t11. Freeze a confirmation level of 1.0750, the assumed maximum High of t4–t8, already available then. Watch only t12–t14 for the first closed H1 bar with Close strictly above that level. Expire the setup if no such bar appears. Before entry, any Bid at or below 1.0690 invalidates it, even if that bar later rebounds.

Step 2 — wait for confirmation from market structure

Suppose t12 Bid OHLC is 1.0710 / 1.0755 / 1.0702 / 1.0752. Its Low avoids pre-entry invalidation and its Close exceeds the frozen level by 2 pips. A chart can mark t9 earlier, but this procedure cannot act at t9. A trendline or engulfing condition would be an alternative specification, not evidence that all confirmations work equally well.

Step 3 — set the entry, stop and target

After the closed t12 data arrive, assume a quote of Bid 1.0752 / Ask 1.0754 and a long filled at Ask 1.0754. In this Forex model, set SL at Bid 1.0685 and TP at Bid 1.0820; close any remaining position after t18 data arrive. Use one position without additions or re-entry. The timed-exit price is not supplied. Ordinary stops can fill worse than their level. Compare the stop with support or resistance, and see the stop loss and take profit for the order distinction.

Hypothetical example — bullish divergence on EUR/USD (illustrative values)
Contextprice is falling, drops toward 1.0720
First lowprice 1.0720, RSI at 26
Second lowprice lower, 1.0690, but RSI higher, 33 — that is the divergence
Confirmationt12 closes at Bid 1.0752 above the frozen 1.0750 level; assume subsequent entry Ask 1.0754
Stop lossBid 1.0685; with 10,000 EUR and USD 2 total commission, exact stop gives USD −71
First targetassumed TP Bid 1.0820: USD +64 net; a gap fill at 1.0680: USD −76 net

Let us be honest about false signals

For 10,000 EUR on a USD account, one pip of 0.0001 is USD 1. Assume USD 2 total commission and zero financing or currency-conversion charges. Entry 1.0754 to target Bid 1.0820 gives 66 pips, USD 64 net; exact stop 1.0685 gives −69 pips, USD −71 net. A stop fill at 1.0680 instead gives USD −76 net. The target-to-exact-stop net ratio is about 0.90, not a ratio measured from the old 1.0690 Low. No win probability follows from that arithmetic. Include divergence in a specified technical-analysis procedure; compare RSI and MACD divergence without assuming their filters remove losses.

What to do tomorrow to master RSI divergence

  1. Replay closed bars and record each pivot’s extreme time and later confirmation time. Keep the price-pivot and RSI-pairing rules fixed throughout the exercise.
  2. Use the hypothetical sequence above to identify t3 and t9 without looking ahead. Record t5 and t11 as their availability bars, then test the separate entry rule.
  3. Change the assumed entry Ask to 1.0756 while keeping target, stop and total commission fixed. Recalculate target USD +62 and exact-stop USD −73; the chart pattern is unchanged but its trade arithmetic differs.
  4. Record source RSI values and displayed precision separately. A higher-RSI rule checks the underlying numbers, as the diagram illustrates.
  5. Write the full test procedure before collecting observations: data source, pivot confirmation, validity window, entry, exits, spread, commission and position size. A chosen demo sample can test implementation; no fixed number of trades certifies profitability or readiness for live risk.