RSI and MACD in plain terms
RSI (Relative Strength Index) and MACD (Moving Average Convergence Divergence) are technical indicators that turn price movements into numerical readouts. They are often used to discuss momentum and trend-like behavior, but they do not directly measure future returns.
A useful way to interpret them is to separate two ideas: (1) what each indicator is mathematically summarizing, and (2) what additional conclusions people sometimes try to draw from those summaries. The first part is relatively stable; the second part is highly assumption-dependent.
How each indicator works
RSI is typically calculated from recent changes in price and expressed as a number on a fixed scale (commonly 0 to 100). In simple terms, RSI increases when upward price changes dominate and decreases when downward price changes dominate. Many platforms also visualize RSI with reference levels (commonly 30 and 70), which are used as informal thresholds.
MACD is built from moving averages. A common interpretation is that MACD reflects the difference between a faster and a slower moving average, and that a “signal line” is derived from MACD to help identify changes in momentum. When the MACD line moves relative to its signal line, people often describe this as an inflection in trend momentum.
Because both indicators are driven by price inputs, you should expect the same indicator to behave differently under different market volatility, timeframes, and parameter choices.
What you can infer (and what you cannot)
You can generally infer the following from RSI:
- RSI provides a bounded snapshot of whether recent price changes have been more dominated by gains or losses.
- A rise or fall in RSI suggests momentum is strengthening or weakening over the selected lookback window.
You can generally infer the following from MACD:
- MACD summarizes how fast and slow averages are separating or converging.
- Changes in the MACD line and its relationship to a signal line describe shifts in the moving-average-based momentum.
What you should not infer:
- RSI “overbought” or “oversold” does not uniquely imply an immediate reversal. These labels are descriptive conventions, not guarantees.
- MACD crossovers or histogram changes are not deterministic forecasts. The same event can occur many times in non-trending conditions without producing a clear outcome.
- Historical behavior of RSI and MACD does not establish future results. Even if a pattern worked in the past, changing volatility, regime, and execution conditions can break the relationship.
Limitations and common failure modes
A material limitation for both RSI and MACD is that their outputs depend on settings (such as lookback periods and moving-average types) and on the chosen timeframe. Changing those inputs can materially change what “signals” look like.
Another failure mode is treating indicators as standalone decision rules. Since RSI and MACD are summaries of price behavior, their readings can overlap across different scenarios—especially when markets move sideways and noise dominates.
You should also be cautious about what is not visible in an indicator value:
- Costs and execution can change realized outcomes even if price behavior matches your expectations.
- Jurisdiction-specific trading rules and practical constraints can affect what you can actually do with any observed setup.
Finally, different data sources can produce slightly different indicator values because of differences in how price series are constructed (for example, the exact timeframe and data handling).
Verification and next question to ask
To independently verify what RSI and MACD are doing, you can:
- Recalculate or reproduce the formulas in a spreadsheet or chart tool using the same parameters and timeframe.
- Test how interpretations change when you alter settings (for example, different RSI lookbacks or different moving-average parameters).
- Compare indicator behavior across different market regimes (trending vs. ranging) to see when momentum-based summaries are less informative.
If your goal is interpretation quality rather than prediction, a good next question is: “Which assumptions about timeframe and parameters must be true for my interpretation to be meaningful?”