The evening scan flags a name you have never traded. The daily chart shows a tidy two-month base and today’s candle poking out of the top of it. The intraday chart looks better still: flat all morning, then a decisive move at half past one that holds into the close. The RSI has gone from the middle of its range to 81. You look at the volume bar under the breakout candle and it is the tallest on the chart. Everything a breakout is supposed to look like, it looks like. Then you check the day’s total traded quantity and it is 3,400 shares — the whole session, every buyer and every seller, about ₹6.5 lakh at this price. That tallest volume bar is four hundred of them. And the flat morning was not the market waiting — it was a hundred and twenty one-minute candles in which not one share changed hands, drawn by your platform at the last price anybody paid, which was yesterday afternoon.
A watchman is meant to note the temperature every hour through the night. He falls asleep at eleven and wakes at four, and rather than leave the sheet blank he copies the eleven o’clock reading into the twelve, one, two and three o’clock rows. The sheet is now complete, every row has a number, and the night looks unusually stable. Nothing on the sheet says which four numbers are observations and which four are the same observation written out four times.
A bar in which nothing traded is that copied row. Your chart has no way of marking it as an absence, because the format only has room for four prices, and the platform has to put something in them. The one column that tells you the truth is volume — and it is the column people look at last.
Three ways to draw nothing, and none of them is wrong
When a bar’s time slot passes with no trades in it, the platform must decide what to show. There are three conventions in common use. They are all defensible, they are all in the field, and they produce visibly different charts of the same security on the same day.
| Convention | What the bar looks like | What it does to your reading |
|---|---|---|
| Flat bar at the last price | Open = high = low = close, a single horizontal line, volume zero | The chart is continuous and the time axis is honest, but the series is now full of manufactured prices. Every rolling calculation over that window is averaging a number nobody traded at |
| Omit the bar entirely | The empty slots simply do not exist; the next traded bar sits immediately beside the last one | Every price on the chart is real, and the time axis has quietly become a list of the moments this stock happened to trade. Two adjacent candles may be four minutes or four hours apart |
| Carry the previous bar forward | The whole previous bar, sometimes including its volume, is repeated | The worst of the three, because the fabricated volume defeats the one check that would have caught the problem. Rare in equities, still met in some end-of-day files and in vendor data for thin instruments |
What flat bars do to an oscillator
Take Wilder’s RSI, because the arithmetic is short and the result is startling. RSI compares an average of the up moves with an average of the down moves over fourteen bars. A bar in which nothing traded contributes a change of zero — it is neither an up bar nor a down bar — so it adds nothing to either average while still counting as one of the fourteen. Both averages are therefore dragged towards zero, and an oscillator whose reading is a ratio of two very small numbers becomes violently sensitive to the next real trade.
The daily chart has the same problem, more slowly
A thin stock does not trade at all on some days. The exchange’s end-of-day file handles this cleanly — for a security with no trades there is simply nothing to report, and for band purposes the previous close is carried as the reference. Data vendors then have to decide again, and the same three conventions reappear on the daily series, where they are much harder to notice because a flat daily candle in a stock that moves ₹0.50 a day does not look odd at all.
- A 200-day moving average may not be 200 days of anything. Under the flat-bar convention the window still holds 200 rows, but forty of them repeat a price that is already in there, so only 160 are observations and the repeated ones are counted several times over. Under the omit convention the distortion runs the other way: 200 rows of a stock that trades five sessions in six reach back 240 sessions, so the oldest price inside a "200-day" average is about a year old rather than the ten months you had in mind.
- The “current price” on a screener may be days old. A [[Last traded price]] is exactly that and nothing more. For an instrument that last traded on Tuesday, the figure quoted on Friday is Tuesday’s, and any calculation you do against it — market capitalisation, a percentage from a high, a valuation multiple — is being done against a stale price.
- The gap that is not a gap. On a platform that omits empty bars, the first trade after a long absence sits immediately beside the last one. It looks like a normal adjacent candle at a very different price. Nothing gapped; the market was simply shut for that stock, in the middle of the session, because nobody was there.
- Percentage moves become uninformative. A stock that has not traded for three days and then trades once, ten per cent up, has not risen ten per cent in a day. It has risen ten per cent over an interval of unknown length, against a reference price that may itself have been a single trade.
- Your fill has nothing to do with the chart. In a name where the whole day was forty shares here and sixty there, the price on the screen is not a price you can transact at in size. That is the [[Bid-ask spread]] and [[Slippage]] problem this track covers elsewhere, and thin bars are how you spot it before you meet it.
The breakout on the tallest volume bar on the chart
The scan has flagged a smallcap. Two-month base, today’s daily candle closes above it, RSI 81, and the volume bar under the breakout is the tallest of the last sixty sessions. Then you notice the day’s total traded quantity is 3,400 shares against a price of ₹190 — about ₹6.5 lakh of turnover for the entire session.
Two stocks both show RSI(14) = 50. In the thin one, ten of the last fourteen bars had no trades and were printed flat; the other four were two up ₹1 and two down ₹1. In the liquid one, the average gain and average loss are both ₹2. Both stocks now rise ₹6 in one bar. What do the two oscillators read?
Dukaan band hai par bahar ka rate board abhi bhi kal shaam ka bhaav dikha raha hai. Koi har ghante uss board ko likhta jaaye toh register mein poora din bhar jaayega — har ghante ek number, sab ek jaise, sab "stable". Par saara din ek bhi sauda nahi hua. Patli scrip ke chart pe candle aise hi bante hain: jis minute koi sauda nahi hua, platform ko phir bhi kuch dikhana hai, toh woh aakhri bhaav pe ek chapti si line kheench deta hai. Ab arithmetic dekho. Chhaudah bar mein se das aise chapte, aur baaki chaar mein do ₹1 upar, do ₹1 neeche — average gain 2 ÷ 14 = ₹0.14, average loss bhi ₹0.14, RSI theek 50. Ab ek aadmi 40 share khareedta hai aur bhaav ₹6 upar jaata hai: gain (13 × 0.14 + 6) ÷ 14 = ₹0.56, loss ₹0.13, ratio 4.2 — RSI seedha 81. Wahi ₹6 ki chaal ek chalu stock pe (jahan gain aur loss dono ka average ₹2 tha) RSI ko 50 se sirf 55 pe le jaati. Farq stock ka dum nahi tha, farq yeh tha ki das bar mein koi tha hi nahi. Isiliye RSI pe chhaana hua scan sabse upar wahi naam laata hai jinme sabse kam trade hota hai — pattern se pehle volume ka column padho.
- A bar with no trades in it must still be drawn, and platforms choose between flat, omitted and carried-forward.
- Flat bars keep the clock honest and corrupt the prices; omitted bars keep the prices honest and corrupt the clock.
- Flat bars hollow out the averages inside RSI and understate ATR, so thin names dominate the extremes of any scan.
- A last traded price on a screener can be days old, and every calculation made against it inherits that staleness.
- The volume column is the only field that distinguishes an observation from a placeholder — read it before the pattern.
Mark it done to track your progress through the curriculum.