RSI Optimization: Specify the Signal Before Tuning Levels
Define raw or smoothed RSI, threshold-cross direction and exit rules before comparing a reproducible TradingView strategy parameter grid.
View Chrome ExtensionBuilding a Strategy with RSI
RSI compares smoothed upward and downward price changes and expresses the result on a 0–100 scale. Its length controls that calculation. The usual 30 and 70 reference levels describe oscillator conditions; they do not guarantee a price reversal or specify an order by themselves.
Define the threshold event. Buying while RSI is below 30, buying as it falls below 30 and buying when it returns above 30 are three different rules. For the hypothetical sequence 29, 28, 31, the return-above-30 event occurs at 31, while the below-30 state was true on the previous two observations. Also specify whether an upper-threshold event closes a long or starts a short.
Separate RSI’s internal smoothing from an optional moving average plotted on the RSI series. A displayed signal line changes strategy results only if the order conditions use it. Decide whether entries compare raw RSI with thresholds, a smoothed series with thresholds, or RSI with its signal line before searching a smoothing length. Keep source price, calculation method, confirmed-bar timing, costs and exits fixed.
Optimizable Parameters
| Parameter | Description | Example Test Values |
|---|---|---|
| RSI length | Length used in the oscillator calculation | 10 / 14 / 18: 3 choices |
| Upper threshold | Upper reference for the chosen exit or entry event | 70 / 75 / 80: 3 choices |
| Lower threshold | Lower reference for the chosen entry or exit event | 20 / 25 / 30: 3 choices |
| Additional smoothing | Hold disabled for the baseline grid; test separately only if used by orders | Disabled; optional separate comparison |
Optimization Example and Test Design
Hypothetical baseline search: 3 lengths × 3 upper levels × 3 lower levels = 27 configurations, including 14/70/30. Fix the long/short policy and use raw RSI for this example. A separate experiment with four smoothing lengths would produce 108 configurations if all other inputs remained the same and the strategy actually used that smoothed series. It would not be evidence of improvement by itself. Inspect the entry event and order timing on several bars, save the unchanged baseline, and evaluate selected candidates on dates excluded from all tuning. No six-minute run, 62% win rate or best BTCUSD configuration is reported here.
Frequently Asked Questions
Related Strategy Optimizations
Start Optimizing Your RSI Strategy
Install the extension, sign in and check your plan access. Automated parameter optimization requires an Advanced plan or above.
View Chrome Extension