Inside Day: The Exact Rule, and What Happens Next
An inside day is a session that never escapes the one before it. Its high stops short of yesterday's high, its low holds above yesterday's low, and the entire bar nests within its predecessor — the outer bar being what traders call the mother bar.
Across 50,774 Binance daily candles the shape prints on roughly one bar in five. That makes it the most common thing on this site, and also the most confidently mis-taught, because the rule a real engine runs beside it is not the rule you have been shown.
What an inside day is
Two bars, one swallowed by the other. Range is the whole story: high against high, low against low, and nothing else consulted.
Consider what the compression means. Yesterday the market discovered a price high enough to exhaust buyers and one low enough to exhaust sellers, and today it declined to revisit either boundary — a session spent inside somebody else's verdict, with no new information arriving to justify the argument.
The median case is not subtle. Over the pairs I measured, the typical inside day covers 55.9% of its parent's range, so it really is about half the size of the thing containing it.
That pause is the signal, and it points nowhere. An inside day tells you volatility contracted; it says nothing whatsoever about which way the release goes, which is why the trade lives at the mother bar's edges rather than inside the quiet bar itself.
How to identify an inside day
Two comparisons, no others.
- The bar's high is below the previous bar's high.
- Its low is above the previous bar's low.
Both have to hold. Miss either and you have an ordinary session, however tight the bar looks on screen.
Strict or inclusive? Plenty of references let an equal high or an equal low count. That choice barely moves the number: 9,508 bars qualify under the strict reading and 9,780 under the loose one, so 272 pairs across nine years hang on an exact tie. Everything below uses the strict version.
Notice what the test never mentions. Bodies are irrelevant here — a green inside day counts, a red one counts, a doji counts, a spinning top counts, and so does a bar that is almost entirely wick. Hold onto that, because the detector in the next section inverts it completely.
What Vike's engine actually flags
batch_harami reads — and the dashed lines are its high and low, all an inside day reads. Each pair satisfies one rule and breaks the other.This is where the page parts company with every other guide on the topic, and it opens with an admission.
Vike's engine has no inside-day detector. Search vike-indicators for a range-containment rule and nothing comes back. Its nearest relative is batch_harami, which measures something genuinely different: the body, never the range.
Three clauses, copied from the crate rather than from anybody's article:
- The current body's top sits strictly below the prior body's top.
- The current body's base sits strictly above the prior body's base.
- The two bars are opposite colors — red parent with green child scores +100, green parent with red child scores −100.
Highs and lows go unread. Ties are rejected too, since the function bails on >= and <=, so a body edge landing exactly on its parent's disqualifies the pair — the mirror image of the engulfing rule, whose inclusive comparisons wave an exact match through.
So the honest phrasing is this. One rule describes a range that shrank; the other describes a body that shrank and changed hands. Overlapping populations, not identical ones — and the section below counts both.
Inside day vs harami, counted on the same bars
Both rules, one tape: 50,754 adjacent pairs from 20 Binance markets, daily, reaching back to August 2017.
Inside days come to 9,508, or 18.7% of bars. batch_harami fires 4,122 times, or 8.1%. Only 1,666 pairs satisfy both, meaning 17.5% of inside days are haramis, while 59.6% of haramis fail to be inside days because the child's wick pokes out of the parent's range even though its body stayed home.
Here is the part that surprised me. Of the 7,842 inside days the crate ignores, 7,818 fall at body containment rather than at color, and 2,614 of those miss by under 0.01% of price.
Why so fine? Crypto never shuts, so a daily bar opens exactly where the last one stopped — 54.7% of these bars open within a basis point of the previous close — and that seam is one of the two body edges, leaving a single tick to decide the clause.
Watch it happen. Bitcoin's inside day on 2026-07-22 opened at $66,556.15, one cent below the prior close of $66,556.16, so the body clause clears by 0.00002% and batch_harami fires. Three weeks earlier Solana printed an inside day that opened one cent above its parent's close, the same clause failed, and the crate said nothing. One seam, opposite ticks, opposite verdicts.
Color, meanwhile, is nearly free: 4,207 pairs pass body containment and 4,122 already have opposite colors, so that third clause rejects 2.0% of what reaches it.
What happens after an inside day
Behind the search term sits one real question, so here is the tape's answer without adjectives.
Over 9,457 inside days the mother bar's range gave way within five sessions 93.6% of the time. Downside first: 46.4%. Upside first: 43.5%. Another 3.7% took out both ends during a single session, and 6.4% simply never resolved.
Direction is a coin flip.
Follow-through looks worse still. Among the breaks that did pick a side, a mere 48.5% were still holding two sessions later — 49.8% of upside breaks against 47.3% of downside ones — and the resolution agreed with the previous three days' drift on 51.1% of occasions, which is a long way of saying prior trend adds nothing.
Timing is the exception, and it is genuinely useful: 62.6% of those breaks land on the very next session. So the shape makes a dependable statement about when volatility returns and a worthless one about where price then goes.
How to trade an inside day
Trade the mother bar, never the quiet one. Everything else is bookkeeping.
Bracket the parent's extremes. A buy stop rests above the mother bar's high and a sell stop below its low, whichever fills first becoming the position. Because 3.7% of sessions take out both ends, kill the opposite order the instant one triggers.
Anchor the stop at the far side. Price traveling back through the opposite extreme has erased the compression that created the setup, which makes it a clean invalidation — expensive on a wide parent, and that expense is the honest cost of the structure.
Size against that width rather than your conviction. The BNB session flagged on 2026-05-29 broke $649.38 and ran 16.0% inside five days; the LINK one from 2026-05-11 broke the other way and closed back over its trigger within two. Identical shape, identical rule, opposite result.
Insist on a reason beyond geometry. An inside day at the boundary of a multi-week range, or one printing while RSI sits at an extreme, beats the same bar floating in the middle of nothing. When I pass on one of these, that missing second reason is nearly always why.
Where inside days fail
Its failure mode has a name every breakout trader has personally funded: the false break.
Study the LINK chart below. Compression sat at 61.0% of the parent range, the next session took out the mother bar's low cleanly, and two days on price had closed back above the trigger — a whipsaw that pays the spread and collects the stop. My first year of trading was essentially a subscription to that sequence.
Then there is frequency. One bar in five carries this shape, so a chart of any length is littered with them, and treating each as a signal manufactures far more trades than edge. Thin books make everything worse: on an illiquid market a quiet session produces the geometry almost mechanically, with no compression story underneath.
I've learned to grade them first. Most get discarded.
Frequently asked questions
Is an inside day the same as a harami? No. One compares ranges while the other compares bodies and additionally demands opposite colors. They agreed 1,666 times out of 9,508 candidates on the tape here, so using the two words interchangeably will mislead you roughly four times in five.
Is an inside day bullish or bearish? Neither, as far as the numbers go. The parent range broke downward first on 46.4% of occasions and upward first on 43.5%, which is about as close to nothing as a directional signal can get. Read the breakout, not the bar.
What is the difference between an inside day and an inside bar? Timeframe, and that is all. The daily version picked up the "day" label; identical containment on an hourly or weekly chart is simply an inside bar. Vike's detectors never look at the clock, so one rule serves whatever candles you hand it.
What is the opposite of an inside day? An outside bar, where the newer session covers both of its predecessor's extremes. When the bodies do the covering as well, that arrangement becomes a bearish engulfing or its bullish twin — a genuinely separate detector with a genuinely separate rule.
How reliable is an inside day breakout? Fewer than half of the breaks in this sample survived a further two days. That is exactly why the sane version of the trade is bracketed rather than directional, with protection parked at the opposite end of the parent bar.
This is educational material, not financial advice. Candlestick patterns describe probabilities, not certainties, and every real trade carries real risk — size positions so a wrong read costs little.