Mechanism and definitions: what “RSI strategy” assessment depends on
RSI (Relative Strength Index) is an oscillator that transforms price changes into a bounded value, commonly between 0 and 100. An “RSI strategy” is not only the RSI value; it also includes rules that connect RSI readings (and sometimes additional conditions) to actions and evaluation. To assess RSI strategies, you therefore need data that covers (1) the indicator inputs, and (2) the rule evaluation inputs.
A complete description should specify:
- The RSI lookback length used to compute RSI.
- The price data used inside RSI (for example, which price field is used to derive gains and losses).
- The timeframe of the data (for example, bar duration) because RSI depends on the sequence length in that sampling.
- Any additional conditions used alongside RSI (for example, filters based on trend, volatility, sessions, or risk limits).
Direct answer: data inputs, provenance, timeliness, and quality checks
To assess RSI strategies in a way you can independently verify, collect the following data categories.
1) Inputs for the RSI calculation
You need the minimum set of inputs that fully determines RSI values:
- Price series for the chosen instrument and timeframe.
- RSI parameters (lookback length and any calculation variants).
- Exact mapping from raw market data to the series used for calculations (for example, how bars are formed and whether corporate actions or symbol rollovers are handled, when relevant).
Assumptions for calculations are required. If someone uses different definitions (or a platform-specific RSI variant), RSI values can differ even with the same timeframe and lookback length. Without parameter transparency, assessment becomes non-reproducible.
2) Rule-evaluation data (what turns indicator readings into outcomes)
If the strategy includes trading rules, you also need data used to evaluate those rules:
- Decision timestamps: when the strategy checks RSI and when it acts.
- Execution price model: what price is assumed for entries/exits (for example, close of bar vs. next bar open).
- Transaction cost assumptions: spreads, commissions, and any modeled slippage.
- Constraints: whether positions are held for a fixed period, reversed, closed conditionally, or prevented from overlapping.
Even for informational analysis, you should be able to state the assumptions clearly enough to re-run the logic on the same input series.
3) Provenance: where the data comes from and how it was produced
You need provenance to judge whether the results could reflect artifacts:
- Source of the price data (vendor, broker feed, or exported dataset) and its update method.
- Symbol and contract continuity rules (particularly if the instrument changes identifiers over time).
- Data cleaning actions (missing bars, outliers, weekend gaps, or corrections) and how they were handled.
If the provenance and cleaning steps are not described, you cannot distinguish strategy behavior from data quirks.
4) Timeliness and alignment checks
RSI is computed from historical sequences, so timing mismatches can create misleading outcomes:
- Ensure the strategy does not “use the future.” For bar-based strategies, confirm whether the RSI value used at time t is computed strictly from information available up to time t.
- Confirm the time zone and bar alignment used in the dataset.
If timing is unclear, the assessment is not independently verifiable.
5) Quality checks: robustness and failure modes
You should look for quality issues that often break indicator-based strategies:
- Overfitting: many RSI parameters and thresholds tuned to one period.
- Regime change: RSI behavior can shift when volatility, trend strength, or market structure changes.
- Cost sensitivity: modest changes in spread or execution assumptions can materially change net outcomes.
- Sampling issues: changing timeframe or resampling bars can produce different RSI dynamics.
Include at least one limitation that could fail the strategy in practice. For example, RSI-based logic can react late during fast transitions, and transaction costs can erase apparent edge when trading frequency is high.
Evidence and example structure: what to be able to reproduce
A useful assessment typically provides enough information for you to replicate the decision path:
- List RSI parameters and the exact price field used.
- State how bars are created and which timestamps are used for checks and execution.
- Provide the cost and execution assumptions used in any evaluation.
- Show at least one short, explainable example window where the RSI logic triggers and explain why.