A strong 15-minute rise can sit inside an hour that finishes lower. The charts need not disagree about the data: they may be summarising different parts of the same price sequence.
This Read the Reaction lesson builds an hourly candle from four smaller candles. Our news-versus-technical comparison separates a catalyst from the market response. Here, the practical question is how changing the interval changes the observation—and the decision clock.
One hypothetical hour, four candles
Imagine a fictional token trading in USD on one fictional spot venue. An announcement occurs at 14:05 UTC on an invented trading day. The following ordinary OHLC candles cover 14:00 up to, but not including, 15:00 UTC. Each row is labelled by its start and lasts fifteen minutes; a trade exactly at the next boundary belongs to the next candle.
Every price, volume and event here is hypothetical. These figures illustrate aggregation, not an observed news reaction, a backtest or expected returns. There are no missing or overlapping intervals. Volume counts units of the fictional token, not dollars.
In the table, O means open, H high, L low, C close and V volume. O/H/L/C are USD prices. Coinbase’s candle definitions describe the first and last traded prices, interval extremes and volume; those definitions underpin this exercise.
| Start | O | H | L | C | V |
|---|---|---|---|---|---|
| 14:00 | 100 | 106 | 99 | 104 | 120 |
| 14:15 | 104 | 105 | 101 | 102 | 90 |
| 14:30 | 102 | 103 | 99 | 100 | 70 |
| 14:45 | 100 | 101 | 98 | 99 | 60 |
The first candle includes five minutes before the announcement. Its OHLC values alone cannot isolate the post-announcement change or locate the 106 high relative to 14:05. A headline timestamp and a candle label are not interchangeable.
Build the hourly candle from its parts
To combine these four adjacent candles, keep the outer interval’s endpoints and all the extremes:
- Open = 100: the first candle’s open.
- High = 106: the largest of 106, 105, 103 and 101.
- Low = 98: the smallest of 99, 101, 99 and 98.
- Close = 99: the last candle’s close.
- Volume = 340 tokens: 120 + 90 + 70 + 60.
So the completed hour is O 100 / H 106 / L 98 / C 99 / V 340. Do not average the four opens or closes. The mean of the closes is 101.25, which is neither the hourly close nor a price every trader could have received.
This works because the smaller intervals exactly partition the larger one and use the same instrument, venue, price source and volume unit. With real data, verify coverage first. Coinbase notes that its historical candles may be incomplete and that intervals without trades are not published. A missing row is a data question to resolve, not permission to invent a price.
Why both readings can be correct
The first fifteen minutes open at 100 and close at 104: (104 − 100) ÷ 100 = +4%. The complete hour opens at 100 and closes at 99: (99 − 100) ÷ 100 = −1%. These are changes over different intervals, not conflicting measurements of the same return.
With a colour scheme that marks closes above opens as green, the first candle is green while the hourly candle is red. The hour’s upper wick still records the excursion to 106. The larger candle compresses the rally and retreat; it does not erase them.
The four smaller candles reveal more of that sequence, but not every trade within each quarter-hour. Even the first candle does not disclose whether its high or low came first. For candle anatomy, see the candlesticks and timeframes lesson.
Changing interval also changes indicator inputs. Four fifteen-minute closes do not become four hourly closes. You cannot obtain an hourly RSI by averaging four RSI readings. Recalculate on the appropriate close series and history; our worked RSI and moving-average example explains why initialization and smoothing matter.
Ask what was knowable at the time
Just after 14:15 UTC, the first candle’s close of 104 is known, subject to data delivery. The 14:00–15:00 candle is still forming. Its eventual close of 99 is not yet available, and neither are its final low or full-hour volume.
Suppose a research rule written before 14:00 requires a completed close strictly above 103. On fifteen-minute candles, the first close qualifies. The corresponding hourly rule cannot be assessed until that hour ends; its eventual close fails the condition. These are two different timing rules, not a fair comparison if the hourly outcome is supplied early.
Using the finished hourly candle to justify a decision made at 14:15 introduces future information. Equally, the hourly failure does not mean the earlier fifteen-minute condition never occurred. Review each decision using only what was available then.
Check labels as well as interval length. Binance’s candlestick documentation identifies candles by opening time and distinguishes closing time. A candle labelled 14:00 is not necessarily a signal confirmed at 14:00.
Give each timeframe a job
One possible process uses the last completed hourly candle for context and a completed fifteen-minute candle for timing. Another waits for hourly confirmation. Write the choice before examining the outcome; do not switch to whichever chart makes the current position look better.
In our example, waiting forty-five more minutes avoids acting on that particular early close, but it could miss a move that continued instead. Earlier action gets information sooner and also accepts more subsequent uncertainty. Neither outcome establishes a profitable method.
Specify a failure condition, loss budget and execution assumptions separately. A close at 104 does not guarantee an entry there after confirmation. Spreads, fees and slippage still matter, whichever interval you choose.
A repeatable comparison checklist
Before calling two charts contradictory, record:
- Same market: instrument, venue, price source and volume unit.
- Aligned windows: start/end boundaries, timezone and missing intervals.
- Correct aggregation: first open, maximum high, minimum low, last close and summed volume.
- Decision time: which candles had actually finished and arrived?
- Fixed role: context or trigger, with an explicit failure and risk plan.
Keep the news question separate: what was announced, how it differed from expectations, and what evidence connects it to trading activity. Two timeframes can describe the response in useful ways. Neither chart reads the headline, proves causation or guarantees that all information is reflected in price.
Educational content, not financial advice. All prices, volumes and events in the example are hypothetical. No timeframe or indicator guarantees a profitable trade.
