Direct answer: what data you need to assess Ichimoku strategies
To assess Ichimoku strategies in a self-contained and verifiable way, gather four categories of data: (1) the indicator inputs and calculation settings, (2) the provenance and timing of the market data used to compute those inputs, (3) the rules and assumptions that turn indicator outputs into an evaluation metric, and (4) quality checks that ensure the data is consistent and suitable. You also need at least one material limitation documented, because the same Ichimoku configuration can behave differently across market regimes.
Mechanism or definition: what “data” means for Ichimoku
Ichimoku is an indicator framework built from several time-based components derived from price series. When people say “Ichimoku strategy,” they usually mean a procedure that:
- Computes indicator components from an OHLC price history (or a provider’s equivalent inputs).
- Produces observations such as relative positioning or cross-conditions between components.
- Defines an evaluation approach (for example, measuring how often a condition occurs, or how outcomes behave after a condition under explicit assumptions).
Because the indicator components are time-window based, the data you need is not just “prices.” You need price data that supports the exact lookback periods and shifting conventions you plan to use.
Inputs to collect
- Price history at the intended timeframe: consistent OHLC (open, high, low, close) bars for the full period you plan to evaluate.
- Computation settings: the Ichimoku parameter values (for example, the lookback lengths) and the displacement/shift convention used by the approach you are assessing.
- Any additional derived fields used by your method: for example, if you compute midpoints or rolling highs/lows, document the exact formulas you use.
Assumptions to state
- Definition of the evaluation moment: when you observe the indicator condition relative to the bar close/open.
- Mapping to an outcome window: what “future” means in your assessment (how many bars ahead you measure).
- Trading and non-trading costs (if outcomes are evaluated): include assumptions for spreads/fees/slippage only if you explicitly model them; otherwise, restrict evaluation to indicator behavior rather than net results.
Evidence or example: an assessment checklist of what to verify
Use a control-oriented checklist to confirm you have the right data and that it is comparable across time.
A. Provenance and alignment
- Data source: identify where the OHLC bars come from (platform, feed, vendor) and whether they reflect continuous trading hours or only active sessions.
- Timestamp alignment: ensure bars represent the same cut-off times across your dataset, especially around rollover or session boundaries.
- Symbol or instrument specification: clarify the exact instrument (underlying pair) represented in the OHLC series, because different venues can have different liquidity and spreads.
B. Timeliness and stability of the dataset
- No mixing of datasets: avoid combining bars from different providers or different update policies without re-checking consistency.
- Stability over time: confirm the dataset covers the full range needed for indicator lookbacks plus any future outcome window.
C. Quality checks
- Missing bars and irregular gaps: identify gaps, duplicated timestamps, or sudden data jumps.
- Outliers: inspect extreme candles that may reflect bad ticks, roll artifacts, or corporate/event-driven changes (where applicable).
- Reproducibility: recompute the indicator components from the stored inputs using documented formulas; the resulting series should match the values produced by your chosen reference implementation (within expected rounding).
D. Evaluation rule transparency
- No standalone claims: if you assess performance, document the full rule set that translates indicator observations into decisions.
- Explicit metric definition: state whether you are evaluating frequency, conditional behavior, classification accuracy, or any outcome-based metric with assumptions.
Material limitation and failure mode (one example)
A common failure mode is regime shift: an Ichimoku configuration that aligns well with past volatility and trend persistence may produce different relationships when volatility compresses, trends weaken, or market microstructure changes. Even with the same inputs, the indicator-to-outcome relationship can vary, so historical relationships do not establish future results.
Limitations and risks: what can go wrong in the data
- Historical relationships can mislead: indicator conditions that were correlated with outcomes in the past may not remain correlated. - Execution and cost sensitivity: if you evaluate net outcomes, small differences in assumed costs or execution timing can materially change results.