The big picture
Ichimoku combines historical price ranges and displaced plots into a compact picture of trend and equilibrium. It can help organise a chart, but it does not discover future prices or establish a company’s value. This topic makes every calculation and time shift explicit, then turns a cloud-breakout idea into a testable research protocol for PSX data. No verified evidence is supplied that it reliably identifies the beginning of major PSX bull runs.
Core concepts
Learning objectives
- Calculate Tenkan-sen, Kijun-sen and both leading spans.
- Read current cloud boundaries without date-alignment errors.
- Specify an unambiguous Kumo-breakout experiment.
- Evaluate whipsaws, gaps, data quality and parameter uncertainty.
1. Build the conversion and base lines
1.1 Range-midpoint arithmetic
Under conventional settings, Tenkan-sen is (highest high over 9 periods + lowest low over 9 periods) ÷ 2. Kijun-sen uses the same operation over 26 periods. A period means the chart interval: nine completed daily observations differ from nine weeks. These are midpoints of extreme prices, not simple moving averages of nine or 26 closes. Their flat sections can arise because the rolling extremes have not changed.
1.2 Choosing a complete observation window
Use consistently adjusted and dated open-high-low-close data. A split, rights issue or unusual corporate action can create an artificial range if the series is inconsistent. Missing trading sessions and zero-volume stale quotes also deserve investigation. Do not calculate a supposedly standard line from a shortened window without clearly labelling that choice. Completed bars avoid treating an unfinished intraday high or low as a final observation.
Step-by-step method
- Specify interval, adjustment convention and window lengths.
- Find each window’s maximum high and minimum low.
- Calculate the midpoint and verify one observation by hand.
In an invented nine-day window with a PKR 120 high and PKR 100 low, Tenkan is PKR 110. If the 26-day extremes are 130 and 90, Kijun is also 110 despite the different histories.
2. Construct the cloud and the lagging line
2.1 Senkou A and Senkou B
Senkou Span A is (Tenkan + Kijun) ÷ 2. Senkou Span B is the midpoint of the highest high and lowest low over 52 periods. Conventional charts display both 26 periods ahead, shading the area between them as the Kumo. The greater value is the upper boundary, regardless of line colour. Colours are software conventions and should never substitute for checking values.
2.2 Displacements and information dates
Chikou Span is the current closing price displayed 26 periods behind. It is useful for visual comparison with previous price structure, but its location can cause serious backtest leakage. A point plotted at an old date may have been calculated using a later close. Store calculation date and display date separately. Likewise, the cloud drawn ahead was calculated from information available now; it is not a forecast observed in the future.
Step-by-step method
- Calculate both spans using only information available at the calculation date.
- Record forward and backward display offsets separately.
- For historical signals, verify when every required value became known.
If Tenkan is 110 and Kijun is 106, Span A is 108. A 52-period high of 140 and low of 80 gives Span B of 110. Both values are plotted ahead, but neither uses a future price.
3. Interpret equilibrium without promising support
3.1 Above, below and inside the Kumo
Price above the cloud is commonly classified as an upward trend state; below it as downward; inside it as uncertain or transitional. Tenkan crossing Kijun describes a change in relative recent range midpoints. Combining these observations can reduce the number of candidate signals, but correlated filters do not provide independent confirmations simply because several lines agree. All originate from the same price history.
3.2 Crossovers, thickness and conflicting evidence
Cloud thickness is the distance between its spans. It measures separation between the underlying calculations, not a proven probability that support will hold. A thick cloud can reflect a wide historical range; a thin cloud does not guarantee an easy breakout. Markets can gap through either boundary. Record conflicting states explicitly and distinguish a descriptive chart label from a forecast, valuation conclusion or executable order.
Step-by-step method
- Compare closing price with the correctly aligned current cloud.
- Record line slopes, crossover state and boundary distance.
- State what would contradict the trend interpretation.
An invented close is 112 above cloud boundaries 108 and 110, while Tenkan remains below Kijun. The learner records mixed evidence instead of converting the above-cloud label into an unconditional purchase instruction.
4. Specify a Kumo-breakout research rule
4.1 Signals, orders and invalidation
Replace “buy a breakout” with an exact rule. For example, a research signal could require a completed daily close above the upper cloud boundary after the previous close was not above it. Define whether a Tenkan–Kijun filter is required, when the signal is observed, and the earliest assumed order time. Daily and weekly rules are separate hypotheses; a weekly bar is not complete halfway through the week.
4.2 PSX daily and weekly implementation
An index chart can describe market conditions but does not imply that the index itself is directly tradable or that a constituent portfolio replicates its return without costs. On individual shares, liquidity, price bands, suspension and announcements can prevent assumed fills. Define position limits and an invalidation rule, while recognising that a stop or limit instruction cannot guarantee an exit price during a gap or unavailable liquidity.
Step-by-step method
- Write the exact signal, observation time and permissible order rule.
- Specify position sizing and an independently defined invalidation condition.
- Test gap, suspension and unfilled-order scenarios before evaluating returns.
A paper strategy signals at Monday’s final close but assumes buying at that same close. The learner corrects the experiment to a realistically available later execution assumption and separately models an opening gap.
5. Validate and document the experiment
5.1 Out-of-sample testing
Use a development sample to choose rules, a validation period to assess changes and a final untouched test period. Rolling or walk-forward evaluation can show how results vary through time. Include corporate actions, delistings and an investable universe known at each date. Comparing only today’s surviving companies or selecting the most attractive parameters after viewing all returns overstates the evidence.
5.2 Costs, regimes and failure analysis
Report turnover, charges, slippage, drawdown, time in market, failed orders and sensitivity to nearby settings. Compare with an appropriate passive benchmark under consistent cash and dividend assumptions. Inspect trending and sideways periods separately; repeated crossovers in a range can produce whipsaws. A result that disappears after modest cost or timing changes is fragile. The final research report should distinguish measured historical outcomes from any unsupported expectation about the future.
Step-by-step method
- Reserve an untouched period and document all parameter trials.
- Apply realistic costs and adverse execution scenarios.
- Report failure cases and benchmark comparisons alongside profitable periods.
A hypothetical backtest looks strong before costs but makes frequent small trades during a sideways year. After a realistic spread and fee model, its advantage disappears. The learner records the failed hypothesis instead of advertising a reliable PSX signal.
A cloud signal must pass an information-date test
- Clean dated OHLC data
- Calculate range midpoints
- Separate calculation and plot dates
- Specify observable breakout rule
- Test realistic execution and failures
Reference table
| Line | Calculation | Display offset |
|---|---|---|
| Tenkan | (9-period highest high + lowest low) / 2 | None |
| Kijun | (26-period highest high + lowest low) / 2 | None |
| Senkou A | (Tenkan + Kijun) / 2 | 26 periods ahead |
| Senkou B | (52-period highest high + lowest low) / 2 | 26 periods ahead |
| Chikou | Current close | 26 periods behind |
Glossary of key terms
- Chikou Span
- The current closing price plotted backward under the chosen displacement.
- Displacement
- A difference between the calculation date and chart display date.
- Kijun-sen
- The conventional 26-period high-low midpoint base line.
- Kumo
- The shaded area between Senkou Span A and Span B.
- Look-ahead bias
- Using information that was unavailable at the historical decision time.
- Senkou Span A
- The midpoint of Tenkan and Kijun, conventionally plotted forward.
- Senkou Span B
- The conventional 52-period high-low midpoint plotted forward.
- Tenkan-sen
- The conventional nine-period high-low midpoint conversion line.
- Whipsaw
- A signal reversal that repeatedly changes a position in a range.
Frequently asked questions
Does plotting ahead mean the cloud predicts future prices?
No. The values are calculated from already available historical data.
Is Tenkan a nine-close simple average?
No. It is the midpoint of the window’s highest high and lowest low.
Does a thick cloud guarantee support?
No. Thickness describes separation between two historical calculations.
Are daily or weekly PSX settings proven best?
No such evidence is established here. Treat each interval and setting as a hypothesis requiring independent testing.
Common pitfalls
| Common mistake | A better approach |
|---|---|
| Using displaced Chikou data at its display date | Track when the underlying close was actually known. |
| Treating several related lines as independent evidence | Recognise their common price inputs and test incremental value. |
| Assuming same-close execution after observing that close | Choose a feasible later execution convention. |
| Selecting only successful stocks and parameters | Include survivorship controls, recorded trials and untouched tests. |
Your learner task
Calculate all five lines on a clean dated sample, retain calculation and display dates, and hand-check three observations. Specify one paper-only breakout hypothesis, a realistic execution rule, costs, benchmark and untouched test period. Submit failure cases and explain why the forward plot is not knowledge of future prices.
Sources & verification
Use the linked documents to check the factual and regulatory context. Source dates and limitations are stated where relevant. Educational examples and analytical frameworks are not personalised recommendations.
- JISPL · Knowledge curriculum ↗
Course identity and subject coverage; factual corrections are explained in this expanded lesson.
- Fidelity: Ichimoku Cloud ↗
Component definitions and conventional chart interpretation; not evidence of PSX strategy performance.
- PSX Data Portal ↗
Official issuer data and notices for dated input verification.
Editorial review: 22 October 2026