A practical observation workflow for four-hour, hourly and 15-minute charts: data timing, invalidation, execution costs and why win rate alone does not establish profitability.
Multi-timeframe analysis organizes the same market at different time scales. It does not provide three independent forecasts. Four-hour, hourly and 15-minute charts contain overlapping price observations, so agreement between them cannot establish a promised improvement in win rate. The useful questions are whether your observation process is consistent, whether information was available at the decision time, and what would invalidate the interpretation.
This is an observation exercise, not a validated trading system. Choose the intervals before examining results, rather than switching until the chart supports your preferred conclusion.
| Chart | What to record | What it does not establish |
|---|---|---|
| Four-hour | Completed ranges, predefined swing points and context | An upward trend cannot retrace |
| Hourly | Whether a pullback remains within the recorded range; invalidation | A drawn support level will hold |
| 15-minute | Whether the observation condition occurred; quotes and costs | A pattern requires you to enter |
For example, defining an upward structure as two successively higher confirmed swing highs and lows is more reproducible than saying the chart looks strong. But confirmation itself needs a rule. If a swing low needs two subsequent completed candles, a historical test must not use that low as a confirmed signal on the candle where it first appeared.
Suppose a data provider builds a four-hour candle from 08:00 to 12:00 UTC. At 10:15, that candle is still forming. Its final high, low and close at noon cannot be treated as information known at 10:15.
| Decision time, UTC | Available four-hour information | Unavailable final information |
|---|---|---|
| 10:15 | Previous completed candle; observations actually recorded up to 10:15 | Final close of the 08:00–12:00 candle |
| After the 12:00 update | Newly completed 08:00–12:00 candle | Final values for 12:00–16:00 |
If the system uses closed-candle signals, match each decision to the latest higher-timeframe observation that had completed and become available. Do not fill its final value backward across the preceding four hours. An indicator updating during a live candle is not automatically defective; the mistake is substituting the final historical value for what was visible at the time.
Record the data provider, time zone, daylight-saving treatment and candle boundaries. Different four-hour charts do not by themselves prove manipulation. Align the interval boundaries and quote type before comparing providers.
These prices are invented and are not a live signal. A researcher records an hourly range of 1.0800–1.0850 in advance, finds that the predefined four-hour context condition is met, then watches the 15-minute chart.
These branches test consistency, not profitability. Record no-trade decisions, missing data and triggered but unfilled orders too. Otherwise, the sample may contain only attractive examples.
Write a single-timeframe baseline and then the extra higher-timeframe filter. Compare the same instrument, period, quote source, costs and capital constraints. Reserve out-of-sample data. Repeatedly choosing parameters after seeing test results is not independent validation.
Win rate alone does not establish profitability. In an invented example, 40% of trades win 2R and 60% lose 1R, producing a pre-cost expectation of 0.2R per trade. A different example wins 0.5R on 60% and loses 1R on 40%, producing -0.1R before costs despite the higher win rate. R is a predefined unit of planned risk, not guaranteed return. Actual assessment also requires costs and tail losses.
Report trade count, average net outcome, maximum drawdown, holding time, unfilled-order frequency and sensitivity to alternative costs. A filter may reduce the sample so much that estimates become less stable. Do not show a higher win rate while hiding fewer observations.
Not inherently. Overlapping observations are not independent evidence. Another chart should answer a specific question rather than offer another opportunity to confirm a bias.
No. They illustrate a division of tasks. A different combination needs its own rules and information timing; this guide does not establish an optimal combination.
No. Position size, contract value, execution, correlated exposures and provider risk are separate. Check assumptions with the risk-budget calculator and verify the provider through the official-register workflow. A chart cannot establish a broker's authorization.
CME's technical-analysis introduction explains chart information. The CFTC advisory on online trading systems discusses hypothetical results, execution and cost limitations. Neither endorses this timeframe combination. This is educational analysis, not a backtest report or personal trading advice.