Nothing changes in the formula — RSI does not know what timeframe it is plotted on. What changes is what it summarises. A 14-period RSI on a 5-minute chart measures 70 minutes of price action, so it reaches 70 and 30 several times a session, and “overbought” stops discriminating between a strong move and an exhausted one.
That single sentence explains almost every complaint traders have about RSI intraday. The indicator is not less accurate at five minutes. It is answering a much smaller question, and most people are still reading the answer as though it were the big one.
If you want the mechanics of the indicator itself — how the ratio is built, what divergence is, why the line behaves as it does — start with what is RSI and how to use it. This page is only about what the timeframe does to it.
What 14 periods actually measures at each speed
J. Welles Wilder Jr. introduced the relative strength index with a default of 14 periods in New Concepts in Technical Trading Systems (1978), working with daily charts. That default has been carried across to every timeframe since, usually without anyone doing the multiplication.
| Chart timeframe | What 14 periods spans | Readings per US equity session |
|---|---|---|
| 5-minute | 70 minutes | 78 |
| 15-minute | 3.5 hours | 26 |
| 1-hour | 14 hours (about two sessions) | 6 or 7 |
| Daily | 14 sessions — just under three trading weeks | 1 |
The bar counts come straight from the session length: the New York Stock Exchange regular session runs 9:30 a.m. to 4:00 p.m. Eastern, which is 6.5 hours, or 78 five-minute bars. Your 5-minute RSI therefore prints a fresh reading 78 times a day, against the daily chart’s one.
Read the table again and the practical problem is obvious. A daily RSI at 72 is telling you something about three weeks of accumulation. A 5-minute RSI at 72 is telling you the last hour has been mostly green candles. Those are not the same claim, and only one of them is rare.
Why “overbought” stops meaning anything intraday
RSI is built from the ratio of average gains to average losses over the lookback. During a clean intraday trend almost every candle closes in the same direction, so there are barely any losses in the denominator, and the reading pins near its extreme and stays there.
On a trend day that behaviour is not a malfunction. It is the indicator correctly reporting an absence of pullbacks. But it means an overbought reading has no ability to tell you whether the move is about to end — the strongest moves of the session produce exactly the same reading as the weakest exhausted ones.
This is the mechanism behind the most expensive habit in intraday trading: shorting strength because a line went above 70. It is a specific case of the general problem in why indicators lag, sharpened by speed.
What to use instead of 70 and 30
Three adjustments, in order of how much they help.
- Move the thresholds to match the day’s character. In an uptrending session, treat roughly 40 as the floor RSI keeps bouncing from and stop treating 70 as a ceiling at all. In a downtrending session, treat roughly 60 as the cap. The useful signal is where RSI fails to reach, not where it arrives.
- Use RSI to filter, not to trigger. Let price structure — a level breaking and holding, a range edge rejecting — decide the trade, and let RSI decide whether you take it. A long at a support retest with RSI holding above 40 is a different proposition from the same retest with RSI at 22 and falling.
- Lengthen the period rather than shortening it. Moving from 14 to 21 on a 5-minute chart smooths the line and cuts the number of false extremes. It is later, which is the honest trade-off. Shortening to 7 or 9 does the opposite and is almost always a mistake at this speed.
Divergence at five minutes
Divergence is the most-taught RSI signal and the one that degrades fastest as the timeframe shortens.
With 78 bars a session, a 5-minute chart manufactures apparent divergences constantly, and the overwhelming majority of them resolve as the trend simply carrying on. The pattern is real; the base rate is against you. The sane use is defensive rather than directional: treat 5-minute divergence as a reason to tighten a stop or take partial profit on a position you already hold, and never as a reason to open a counter-trend one. How to take partial profits covers the mechanics of doing that without capping the trade.
The one multi-timeframe use that holds up
If you keep one RSI technique for intraday trading, keep this one.
Put RSI on both the 5-minute and the hourly chart. Use the hourly to decide which direction you are allowed to trade, and the 5-minute only to time the entry in that direction. A 5-minute reading of 30 means something quite different when the hourly RSI is at 60 and rising — that is a pullback inside strength — than when the hourly is at 35 and falling, where it is a continuation of weakness dressed up as a bargain.
This is the same discipline as any other alignment check, and the general framework is in multi-timeframe analysis. The indicator is doing nothing clever here. It is the two timeframes together that carry the information, which is true of nearly everything on a technical chart.
A workable 5-minute rule set
Not a strategy — a set of constraints that stops RSI from hurting you at speed.
- Never enter on an RSI level alone. Price structure triggers; RSI filters.
- Do not fade a reading above 70 or below 30 in a trending session. Fade only in a session you have already identified as ranging.
- Use 40/60 as your working bands on 5-minute charts, and let 70/30 mark strength rather than exhaustion.
- Require the hourly RSI to agree with the direction before you act on the 5-minute one.
- Treat 5-minute divergence as a management signal for an open trade, never an entry signal.
- Set the period once and leave it. Re-optimising it after a losing day is curve fitting, and a hand backtest will tell you more than any amount of dragging the number around.
Frequently Asked Questions
What is the best RSI setting for a 5-minute chart?
There is no setting that makes RSI reliable on its own at five minutes. A 14-period RSI is the standard and is a reasonable default; lengthening it to 21 smooths the line and reduces false extremes at the cost of lateness. The larger gain comes from changing the thresholds rather than the period, and from only taking signals that agree with the higher timeframe trend.
Why does RSI stay overbought on a 5-minute chart?
Because on a trending intraday move, almost every candle closes up, and RSI is built from the ratio of average gains to average losses. With few down closes in the lookback, the reading pins near the top and stays there for as long as the trend runs. A sustained reading above 70 on a 5-minute chart is evidence of strength, not evidence of exhaustion.
Is RSI divergence reliable on a 5-minute chart?
Much less so than on higher timeframes. A 6.5-hour US equity session contains 78 five-minute bars, which produces many apparent divergences per day, most of which resolve as the trend simply continuing. Divergence at five minutes is worth treating as a reason to tighten a stop or take partial profit on a position you already hold, not as a reason to enter a counter-trend trade.
How much price does a 14-period RSI cover on a 5-minute chart?
Seventy minutes. Fourteen periods multiplied by five minutes is 70 minutes of trading, so the indicator is summarising a little over one hour of price action. The same 14-period setting on a daily chart summarises 14 sessions, or just under three trading weeks. The formula is identical; what it is measuring is not remotely comparable.
Bottom line
RSI is not broken on a 5-minute chart. It is simply answering a 70-minute question 78 times a day, and the thresholds most traders use were designed for a three-week one. Move your working bands to 40 and 60, stop reading 70 as a ceiling in a trending session, demand agreement from the hourly before you act, and use divergence to manage trades rather than to start them. Do that and the indicator becomes a useful filter. Treat it as a trigger and it will hand you the wrong side of every good move of the day.