Indicator methods
How to Confirm RSI Signals With MACD in Crypto Markets
Turn an RSI event and a MACD condition into a testable confirmation rule. Define deadlines, avoid hindsight, and count the setups MACD rejects.
Basis Research · 2026-10-03 · 8 min read
“RSI plus MACD” sounds like a complete trading method, but it leaves out the decisions that determine the result. Was the RSI event an oversold recovery, a centerline crossing or a divergence? Did MACD need to cross its signal, move above zero, or merely improve? How long was confirmation allowed to arrive?
A second indicator is useful only if its rule was known before the outcome. This field guide turns confirmation into a finite, auditable candidate process, using examples adapted from the Basis research archive. It does not claim that adding MACD improves returns by default.
01 / 07
Define the RSI event first
An RSI recovery above 30, a crossing above 50 and a price/RSI divergence are different events. Pick one. Record its length, threshold, bar-close policy, exchange, instrument and timeframe. A completed-bar crossing is not the same as a provisional intrabar crossing that disappears before close.
For a divergence based on swing pivots, note when the later pivot actually becomes knowable. Do not stamp the signal onto the earlier pivot candle and then pretend it was actionable there. Give each candidate an ID and a price level that independently invalidates the idea.
02 / 07
Choose exactly what MACD must confirm
MACD has a line, a signal line and a histogram. “Bullish MACD” could mean the MACD line is above its signal, the histogram crosses above zero, MACD itself crosses above zero, or the histogram simply becomes less negative. These tests disagree often. For example, MACD at −2 and signal at −3 yields a +1 histogram even though both lines remain below zero.
Write a binary condition, such as “the completed MACD histogram crosses from nonpositive to positive.” If you test several definitions, name separate method versions. Do not rescue a rejected RSI candidate by switching to whichever MACD definition happened to fit the later chart.
03 / 07
Put a clock on confirmation
Consider an RSI event on bar t. You might allow MACD to confirm at t, t+1 or t+2, and expire the candidate after that. If the price invalidation is reached at t+1, a MACD event on t+2 cannot retroactively revive the original candidate. A new RSI event must create a new candidate if your method allows one.
Decide whether RSI must stay above its threshold throughout the wait or whether the initial event is latched. Those are different rules with different counts. Record the deadline and the cancellation reason even when no order is placed.
- Create a candidate when the defined RSI event completes.
- Check only the preselected MACD condition until the deadline.
- Cancel immediately if price invalidation occurs.
- Expire unconfirmed candidates; never delete them from the sample.
04 / 07
Count the cases that do not confirm
In a fictional set of 60 RSI candidates, imagine 18 confirm, 12 invalidate while waiting and 30 expire. All 60 remain in the ledger. Eighteen divided by 60 is a 30% confirmation rate; it is not a win rate and says nothing by itself about expectancy.
Record what happened to the 42 excluded candidates. Did the filter avoid losses, miss winners, or simply delay entries until the reward-to-risk worsened? A gallery of the 18 confirmed examples cannot answer that question.
05 / 07
Respect timeframe completion
If RSI is calculated on five-minute candles and MACD on one-hour candles, a method requiring a completed hourly signal cannot use the current hour’s final MACD value before the hour ends. Each RSI candidate must reference the last completed higher-timeframe observation available at that moment.
A live-value method is possible, but it is a different method: archive the provisional MACD state at arrival time. A historical plot that later shows a clean crossing cannot justify an earlier decision made without that information.
06 / 07
Recheck price and execution after confirmation
Two panels agreeing does not make a trade attractive. When MACD finally confirms, compare the current entry with the original invalidation and objective. If the remaining price room no longer compensates for spread, fees and risk, log “confirmed but rejected.” That is a valid result, not a failed signal to hide.
Keep eligibility separate from fills. A limit order can remain unfilled; a marketable order can fill away from the chart price. Use actual entry timing in any comparison against the RSI-only baseline.
07 / 07
Test whether MACD adds value
Compare the frozen RSI-only rule with the frozen RSI-plus-MACD rule under consistent exits and costs. Report candidates, trades taken, avoided losses, missed winners, average delay, drawdown and net results—not just an improved win rate. Keep a later period out of rule selection for an honest check.
In Basis, open RSI on a liquid crypto chart, add MACD from the indicator panel and use the journal or a table to record the event, MACD component, deadline, cancellation and outcome. An accepted setup and an expired setup deserve equally complete records.