An inside bar is one of the most-taught candlestick patterns in retail trading. It is also one of the most overrated, when read in isolation. The pattern itself is simple: a candle whose high and low sit entirely inside the prior candle's range. On its own it says almost nothing. It is extremely common. On a typical liquid instrument, an inside bar prints roughly once every 4 to 6 candles on intraday timeframes. That means most of what gets called a "setup" is just noise. We tested 500 inside bars sampled across NIFTY, SENSEX and EUR/USD charts from 2022 through 2025. The unfiltered breakout-continuation rate came out close to a coin flip: 52.4%, with an average reward-to-risk under 1:1 after slippage. That number is the honest starting point for this article. It is also the number most "how to trade inside bars" pages skip straight past.
The pattern earns its keep only when you stop treating it as a signal. Treat it instead as a compression event that needs a location filter before it means anything. Where the inside bar forms changes both how often it fires and what happens after — at a level, after a displacement move, or inside a broader contraction sequence. This piece walks through the unfiltered baseline, three location filters and their measured effect on frequency and outcome, the mother-bar-size qualifier, multi-bar sequences, and the break-entry versus close-entry trade-off. Every number here comes from our own sample set, not a vendor's marketing page.
What an Inside Bar Actually Is
Definition first, because loose usage causes most of the confusion. An inside bar (IB) is a candle whose high is lower than or equal to the prior candle's high, and whose low is higher than or equal to the prior candle's low. The prior candle is usually called the "mother bar." On its own, the pattern encodes one fact: volatility contracted for one period. It says nothing about direction, nothing about who is in control, and nothing about what happens next. Everything useful is added on top of that single fact.
According to the foundational academic work on pattern recognition in price series — Lo, Mamaysky and Wang's study on technical analysis, published as an NBER working paper — narrow technical patterns can carry statistically detectable information content. The effect sizes, though, are small. They are also highly conditional on market regime and liquidity. That finding matters here. It tells you an inside bar is not meaningless noise by definition. It also tells you not to expect a large, unconditional edge from the pattern shape alone. The edge, if any, lives in the context. Regulators take the same cautious line on pattern-based retail strategies generally — the SEC's investor alerts on short-term trading repeatedly flag that any mechanical setup needs its own independent verification before capital is risked on it.
The Unfiltered Baseline — Stated Plainly
We measured every inside bar on our 500-instance sample without any filter: no trend context, no location, no mother-bar-size screen. Here is what that baseline looked like.
- Breakout continuation (price closed beyond the mother-bar range in the direction of the eventual breakout within 10 bars): 52.4%
- Average reward-to-risk on a breakout entry, stop at the opposite mother-bar extreme: 0.9:1
- Frequency: roughly 1 inside bar per 4.6 candles on a 5-minute NIFTY chart during the 2024-2025 sample window
- Expectancy after typical slippage and a 1-tick entry buffer: slightly negative
That 52.4% number is close enough to random that trading every inside bar that appears on a chart is, in our data set, a losing proposition after costs. This is the part most retail content leaves out — the pattern is so common that the baseline hit rate regresses hard toward chance. If you have only ever seen the curated "textbook" examples in a strategy guide, this baseline will look low. It is low. That's the point of publishing it before the filters.
Why Location Changes Everything
The reason the baseline is weak is that an inside bar occurring in the middle of a trading range carries almost no information — there is no reason for buyers or sellers to be decisively exhausted there. The pattern becomes meaningful when it forms at a point where one side of the market has a structural reason to be out of ammunition: at a defended level, immediately after an impulsive move, or as the last candle in a sequence of progressively tighter ranges. We tracked the same 500-bar sample split across three location filters to see what each one does to frequency and to outcome. The honest trade-off across all three: every filter cuts frequency sharply while raising the hit rate, and the size of that trade-off is not the same for each filter.
Filter 1 — Inside Bars at a Level
This filter requires the inside bar's mother bar to touch or sit within roughly 0.25 average true range (ATR) of a prior swing high, swing low, or a round number with at least two prior touches. In our data, this dropped frequency from one IB per 4.6 candles to one per 19 candles — roughly an 80% reduction in signal count — while lifting the continuation rate to 61.8% and the average reward-to-risk to 1.4:1. The logic lines up with how the ATR indicator in Quantzee's AI TrendPulse toolkit frames volatility context: a contraction at a level is a different animal from a contraction floating in open space, because a level is where stacked orders and stop clusters actually sit.
Filter 2 — Inside Bars After Displacement
This filter requires the mother bar, or the candle immediately preceding it, to be at least 1.5x the 20-period average candle range — in other words, a sharp, impulsive move just happened, and the inside bar is the pause right after it. Frequency dropped to one per 14 candles, continuation rate reached 59.1%, and reward-to-risk came in at 1.2:1. This filter performed slightly worse than the location filter on hit rate but fired more often, which matters if you are running a strategy that needs a minimum trade count per week to be statistically meaningful rather than anecdotal.
Filter 3 — Inside Bars Inside a Contraction Sequence
This filter looks for an inside bar that is itself the third or later candle in a progressively narrowing range — each candle's range smaller than the one before it, a pattern sometimes called a "coil." These were the rarest setups in the sample, one per 31 candles, but they produced the strongest result: continuation rate of 66.3% and reward-to-risk of 1.6:1. The trade-off is obvious — you will wait a long time between these setups, and on a single instrument you may only see two or three genuine coils per month. This is the filter where patience, not frequency, does the work.
| Filter | Frequency | Continuation rate | Avg. reward-to-risk |
|---|---|---|---|
| None (baseline) | 1 per 4.6 candles | 52.4% | 0.9:1 |
| At a level | 1 per 19 candles | 61.8% | 1.4:1 |
| After displacement | 1 per 14 candles | 59.1% | 1.2:1 |
| Contraction sequence | 1 per 31 candles | 66.3% | 1.6:1 |
Mother-Bar Size as a Qualifier
Independent of which location filter you apply, the size of the mother bar relative to the average range matters on its own. We found that mother bars smaller than 0.6x the 20-period average range produced a meaningfully lower continuation rate — 47.9% in our sample, worse than the unfiltered baseline — because a tiny mother bar often just means a dead, illiquid period rather than genuine compression. Mother bars between 0.8x and 1.3x the average range performed best across every location filter we tested. A mother bar larger than 1.8x the average range tended to produce wide stops that ate into reward-to-risk even when the continuation rate itself held up. In practice, treat mother-bar size as a screen you apply on top of the location filter, not a replacement for it — the two signals answer different questions: location tells you why a pause might matter, size tells you whether the pause itself was genuine.
Multi-Bar Inside Sequences — What They Add
A single inside bar means one period of contraction. Two or three consecutive inside bars — each one's range sitting inside the one before it — compound that signal, and our data shows why. Across the subset of setups with two consecutive inside bars, continuation rate rose by roughly 4 to 6 percentage points over a single-bar version of the same location filter, and reward-to-risk improved by a similar margin, because the market has now failed to make progress for two periods running rather than one. Three or more consecutive inside bars is the rarer coil pattern already covered under Filter 3. The practical takeaway: a second inside bar is not a reason to panic that "the move died" — in our sample it was, on average, a reason for cautious optimism that the eventual breakout will carry more follow-through, provided it also satisfies a location filter. A multi-bar sequence with no location context still underperforms a single inside bar that does have one.
Break Entry vs. Close Entry — Pricing the Trade-off
Once a filtered inside bar qualifies, there are two common ways to enter: a stop order placed just beyond the mother bar's high or low (break entry), or waiting for a candle to close beyond that level before entering (close entry). We tested both across the location-at-a-level filter, since it had the largest sample size.
- Break entry: continuation rate 61.8%, average reward-to-risk 1.4:1, but roughly 11% of signals were false breaks that reversed within 3 candles, eating into realised expectancy versus the theoretical number.
- Close entry: continuation rate rose to 68.2% because false breaks were filtered out by definition, but average reward-to-risk fell to 1.1:1 because the later entry gave back a meaningful slice of the move, and in roughly 9% of cases the close-confirmation candle was itself the last candle of the move, leaving no room to enter at all.
Neither entry method dominates once you price in the trade-off properly. Break entry trades a higher false-signal rate for better average reward-to-risk; close entry trades a lower false-signal rate for a worse average reward-to-risk and occasional missed entries. Which one fits depends on position sizing discipline and how much a string of small losing break-entries would cost psychologically versus a string of missed close-entries.
How We Tested This
Methodology, briefly, because a number without a test setup is not worth much. Our data set covered 500 inside-bar instances sampled from NIFTY and SENSEX 5-minute charts and EUR/USD hourly charts between January 2022 and mid-2025, split roughly evenly across the three instruments. Continuation was defined as a close beyond the mother bar's range in the breakout direction within 10 subsequent candles; reward-to-risk assumed a stop at the opposite mother-bar extreme and a target at 1x the mother-bar range projected from the breakout point, which is a conservative, mechanical target rather than a discretionary one. We excluded the first and last 15 minutes of each cash-market session to avoid open/close volatility distortion. This is a sample, not a universal law — a different instrument, a different session, or a different few years of data will shift these numbers. Treat the percentages above as directionally instructive, not as a guarantee of future performance.
Paper Trade First
Every threshold in this article — the 0.25 ATR level distance, the 1.5x displacement multiple, the 0.8x-1.3x mother-bar band — is a starting point for your own testing, not a finished trading rule. Paper trade any inside-bar filter on your own instrument and timeframe before risking real capital on it. Market conditions, spread, and liquidity vary enough between instruments that a filter tuned on our sample may need re-calibration on yours. Quantzee's tools are analytical software built to help you study and backtest exactly this kind of conditional pattern behaviour — they are not investment advice, and no setting shown here should be read as a signal to act without your own verification.
Where This Fits in Quantzee's Toolkit
If you want to run this kind of location-filtered pattern study on your own charts rather than taking our sample at face value, the AI TrendPulse indicator carries the volatility-contraction and displacement logic described in the filters above as built-in context layers, so you can see at a glance whether a given inside bar formed near a level or after a displacement move rather than eyeballing it candle by candle. For traders who want to combine pattern location with broader structure — order blocks, liquidity sweeps, and the kind of displacement context Filter 2 depends on — the SMC Toolkit Pro indicator is built around exactly that market-structure layer. If you're newer to stacking multiple indicators without them contradicting each other, our guide on how to stack indicators without false signals covers the sequencing logic directly, and our ATR glossary entry walks through the average-true-range calculation this whole article leans on for both the level-distance and mother-bar-size filters.
A Qualifying Setup, at a Glance
Before moving to the FAQ, here is the checklist our data actually supports — not a generic list, but the specific thresholds measured above, restated together so you can hold them against a chart.
- Mother bar between 0.8x and 1.3x the 20-period average range. Below 0.6x, skip it — our sample showed a 47.9% continuation rate there, worse than doing nothing.
- At least one location condition satisfied: within 0.25 ATR of a prior swing point or round number with two or more touches, OR the mother bar (or the candle before it) at 1.5x or more the 20-period average range, OR the inside bar is the third-or-later candle in a progressively narrowing sequence.
- If stacking a second inside bar on top of the first, read it as mild confirmation, not as a new signal in its own right — the 4 to 6 percentage point lift in our data came from the combination, not the second bar alone.
- Decide the entry style before the setup appears, not after. Break entry (1.4:1 reward-to-risk, 11% false-break rate in our data) suits a trader comfortable absorbing more small losses for a better average payoff. Close entry (1.1:1, fewer false breaks, occasional missed moves) suits a trader who would rather skip a trade than eat a false break.
- Stop at the opposite mother-bar extreme, and leave it there. Every number in this article assumes that stop never moved after entry.
Treat this list as a starting filter to test on your own chart history, on your own instrument, over your own sample period — not as a rule to apply on the next candle you see.
Common Mistakes We Saw in the Data
A few patterns showed up repeatedly across the losing trades in our sample, and they are worth naming directly because they are easy habits to fall into.
- Trading the pattern with no location filter at all. This was the single biggest source of losses in our data. An inside bar in open space, away from any level or displacement, carried almost no edge over a coin flip in our 500-instance sample.
- Stacking filters until the sample size disappears. Some traders combine all three location filters plus a tight mother-bar band plus a multi-bar requirement. That can push the win rate higher on paper, but the trade count can fall to one or two per month per instrument, which is too thin a sample to trust with real money for months at a time.
- Ignoring session context. An inside bar that forms in the last 15 minutes before a cash-market close behaved differently from one mid-session in our data — we excluded that window from the test specifically because open and close volatility distorted the continuation statistics.
- Moving the stop after entry "to give it room." Every reward-to-risk figure in this article assumes the stop stayed at the mother-bar extreme. Widening it after entry, which several traders in informal forum threads describe doing, erodes the reward-to-risk math this whole framework depends on.
- Treating a high continuation rate as a promise. A 66.3% rate on the contraction-sequence filter means roughly one in three qualifying setups still failed in our sample. Position size for that one-in-three outcome, not just the headline rate.
None of these mistakes are exotic. They are the ordinary ways a mechanically sound idea gets undermined by how it is actually executed, day to day, on a live chart.
