Charts & Candlesticks

1. What you'll learn

This chapter teaches you to read a price chart. By the end you will be able to:

• Say what technical analysis (TA) can and cannot do, and which assumptions it rests on.
• Read the four numbers behind every candle (open, high, low, close) and draw the candle yourself.
• Pick a chart type and a timeframe that suit your style. For swing trading that usually means daily candles checked against the weekly chart. For positional trades and investing, weekly and monthly charts.
• Recognise all 15 candlestick patterns from Varsity's Module 2, using rules written as numbers rather than eyeballing. For each one you'll know where Varsity says to enter and where to place the stop.
• Find those patterns automatically in pandas, and test whether they had any edge on 37 NSE stocks over 10 years.

2. What technical analysis is

Before you buy or sell anything you need a point of view. Varsity says a complete view has five parts: a direction (up or down), an entry price, the risk you'll accept (a stop), the reward you expect (a target), and how long you expect to hold. Technical analysis is a way to build that view from the market's own trading record, meaning prices and volumes, without studying the business.

Varsity explains it with a story about a food street in a foreign country. You can't read the menus, so you have two ways to choose dinner:

1. Research each stall: look at ingredients, taste samples, judge the cooking. You know exactly what you're getting, but there's only time to check a handful of stalls. That's fundamental analysis (FA): deep study of a few companies.
2. Watch the crowd: stand on a corner, see which stall draws the biggest queue, and assume the crowd knows something. It scales to hundreds of stalls in minutes, but sometimes the crowd is wrong. That's technical analysis: scanning many stocks for what buyers and sellers are actually doing.

On a chart, what buyers and sellers did shows up as shapes, and some shapes recur. A technical analyst learns to recognise them, form a view, and set entry, stop and exit levels.

TA works on any asset that has a price history: stocks, indices, commodities, currencies, bonds. Once you've learnt it, you can apply it anywhere, much like a driver can drive any car. FA is different for each asset. Analysing a bank's balance sheet has little in common with analysing pepper harvests or crude inventories. This portability is one of TA's biggest practical advantages.

3. The four assumptions behind TA

Every method rests on assumptions. If you use TA without knowing them, you won't know when it is likely to fail.

Varsity adds a qualification to assumption 4 that matters a lot: history repeats only when the same set of factors repeats. Its example is a stock that has fallen for four sessions, falls a fifth time on lower volume with a narrow range, then rallies on day six. If you later see the same three conditions together, you may expect the same outcome. One condition on its own isn't enough. That's why candlestick rules always come with conditions attached (a prior trend, a size requirement), and why Section 19 checks whether adding context helps.

4. TA vs FA, and what to expect from TA

It's pointless to argue whether TA or FA is "better". They answer different questions. FA asks what to own: is the business good, and is the price reasonable? TA asks when to act and where you're wrong. Good market participants learn both.

Varsity sets these expectations for TA, and they're worth taking seriously:

• Trades: TA is best for short-term trades. Pick long-term investments with fundamentals, then use TA to time your entries and exits.
• Return per trade: expect frequent, modest wins, not a 50% gain in a week.
• Holding period: a TA trade usually lasts from a few minutes to a few weeks. This course skips intraday, so for us it's days to weeks.
• Risk: when a trade goes against you, exit at your stop. Don't hold on and hope. A TA trade that has turned into an "investment" because it's losing money is the most common way beginners lose.

5. OHLC: summarising a trading day

NSE's normal equity session runs from 9:15 am to 3:30 pm, after a pre-open call auction that starts at 9:00 am. In that time a busy stock can trade millions of times. Plotting every trade would be useless, so we summarise each period with four prices:

• Open (O): the first price of the period.
• High (H): the highest price traded.
• Low (L): the lowest price traded.
• Close (C): the final price. It's the most important of the four. It tells you who won the session, and it's the reference everyone carries into the next day. If \(C > O\) the day is positive (bullish). If \(C < O\) it's negative (bearish).

Add Volume, the number of shares traded, and you have an OHLCV bar. That's exactly what the course data files contain.

6. Chart types

A normal chart plots one value per point in time. TA needs four values per point. Here's how each chart type handles that, all drawn from the same RELIANCE data for Jul–Sep 2026:

6.1 Line chart

A line chart joins the closing prices. It's clean and shows the general trend at a glance. Weekly line charts are good for a quick sweep through many stocks. Its weakness is that it throws away the open, high and low. It can't tell you that a day opened at its high and sold off, or that buyers rescued a deep intraday fall.

6.2 OHLC bar chart

Each bar is a vertical line from the low to the high, with a small tick on the left for the open and a tick on the right for the close. If the right tick is higher than the left, the close beat the open (a positive day). The length of the line is the day's range, \(H - L\). Bar charts carry all four prices, but patterns are hard to see in them, especially when you're scanning dozens of charts. A minority of traders prefer them.

6.3 Japanese candlestick chart

Candlesticks are thought to have been used by the 18th-century Japanese rice trader Homma Munehisa. They reached Western traders through Steve Nison's book Japanese Candlestick Charting Techniques (1991). That's why most pattern names are Japanese. A candle draws the open-to-close range as a filled rectangle, the real body, coloured by direction. Thin lines called shadows or wicks run up to the high and down to the low. They show the same information as a bar, but the body makes the battle between buyers and sellers visible at a glance, which is why candles are the default for nearly everyone. Section 8 covers how to read them.

6.4 Two derived chart types (brief)

Varsity's core module sticks to line, bar and candles. You'll still meet these two on every charting platform:

• Heikin-Ashi ("average bar") smooths ordinary candles. Each HA candle uses the average of the current bar and the previous HA candle, so trends show up as long runs of one colour.
• Renko ignores time completely. A new brick of fixed size (say ₹20) is drawn only when the close moves a full brick beyond the last one. A reversal needs two bricks. Quiet days produce nothing, so the chart shows only meaningful moves.

7. Timeframes and resampling

A timeframe is the period each candle covers. The method for building a candle is the same at every timeframe:

Varsity uses 252 days as a convention. Our NIFTY file has 249 daily candles for 2025, and our 10-year RELIANCE file has 2,468 real sessions (after dropping 6 zero-volume holiday rows that Yahoo stores), about 247 a year.

Information vs noise

Smaller timeframes give you more candles, but much of the extra detail is noise: random wiggles that don't change the bigger picture. Pick a timeframe that matches your holding period, and confirm with the next higher one:

• Swing trader: find setups on the daily chart, and check that the weekly trend isn't fighting you. Most of the setups in Chapter 5 are long trades taken only when the weekly trend is up.
• Positional / futures trader: read the trend on the weekly chart and time the entry on the daily.
• Investor: the monthly and weekly charts. A daily candle should almost never change an investment decision.

8. Reading a single candle: the buyer–seller story

Every candle tells you how a fight between buyers (bulls) and sellers (bears) went over that period. Once you can read that story, you don't need to memorise patterns as pictures. You can work out what they mean.

Varsity puts it simply: a long candle shows intense buying or selling, and a short candle shows subdued trading. Candles are relative. A 2% range is a big day for HINDUNILVR and an ordinary one for ADANIENT. That's why this chapter often compares a candle with its own 20-day average body.

Varsity groups patterns by how many candles they need:

• Single-candle: marubozu (bullish/bearish), spinning top, doji, paper umbrella (hammer / hanging man), shooting star.
• Multiple-candle: engulfing (bullish/bearish), piercing pattern, dark cloud cover, harami (bullish/bearish), morning star, evening star.

9. The ground rules for candlestick patterns

Before any individual pattern, here are Varsity's three rules, plus the entry, stop and size rules that apply to every pattern in this chapter.

9.1 Three rules

1. Buy strength, sell weakness. Buy on a green (bullish) candle day and sell or short on a red one. If you buy into a falling day, you're buying while sellers are still winning.
2. Be flexible, but quantify and verify. Real candles rarely match the textbook. A "marubozu" may have a tiny shadow. Allow some leeway, but write the leeway down as a number (say "body ≥ 90% of range") so your decisions are consistent and can be tested.
3. Look for a prior trend. A bullish reversal pattern only counts after a downtrend, and a bearish one only after an uptrend. There's nothing to reverse otherwise. (The marubozu is the one exception.)

Varsity judges the prior trend by eye. A computer needs a definition, so here is the one used for every detection in this chapter:

9.2 Entries: the risk-taker and the risk-averse

Varsity describes two kinds of trader:

• Risk-taker: enters on the day the pattern completes, near the close, after checking around 3:20 pm that the candle really fits the rules. (For F&O stocks under the closing auction, do this check just before continuous trading ends at 3:15 pm and enter through the auction, or act on the stock's futures after the official close and before 3:40 pm. See Section 5.) The entry price is better and the stop is closer, but the pattern hasn't been confirmed.
• Risk-averse: waits for the next day and enters near its close only if that day's candle points the right way (green for a long, red for a short). This confirmation applies rule 1. The cost is a worse entry and a deeper stop, and sometimes missing the trade completely.

Varsity's rule of thumb: the more candles a pattern contains, the more reasonable it is to enter on the completion day, because the extra candles already provide some confirmation. For the three-candle stars, Varsity says both types enter on day 3.

9.3 Stop-loss: built into the pattern

Varsity points out that candlestick patterns come with their own risk control:

• Long trades: stop at the lowest low of the pattern's candles.
• Short trades: stop at the highest high of the pattern's candles.

Once you're in, stay in until either the stop or the target is hit. The only change allowed is moving the stop in your favour (a trailing stop). Patterns give no target. Targets come from support and resistance (Chapter 3) or from a reward-to-risk multiple (Chapter 11). Taking a loss at the stop is part of the job, and sometimes the stock reverses straight after stopping you out. Section 16 has a real example. Stick to the rules anyway.

9.4 The trade trap: skip candles that are too small or too big

Varsity says to avoid trading off candles whose range is below about 1% or above about 10%. A tiny candle shows no conviction, and its direction is close to random. A huge candle forces a very deep stop, so a failed trade costs a lot. In our 37-stock dataset, 4.4% of daily candles had a range under 1% of the previous close and only 0.7% had a range over 10%. The median range was 2.15%. Every single-candle rule in this chapter applies this 1–10% filter.

10. Marubozu

Marubozu means "bald" in Japanese. It's a candle with (nearly) no shadows: all body, no hair.

The logic. In a bullish marubozu, buyers were willing to pay higher prices all session, right up to the close. Nobody managed to push it back. That's a sudden, one-sided shift in sentiment, and the expectation is that it carries on for a few sessions. A bearish marubozu is the mirror image. Varsity treats the marubozu as the only pattern that doesn't need a prior trend. It means the same thing wherever it appears. Varsity's own example tolerated a 0.17% gap between high and close, which is rule 2 at work.

11. Spinning top and doji

11.1 Spinning top

A spinning top has a small real body with upper and lower shadows of roughly equal length. The colour doesn't matter, because open and close are close together either way.

The logic. The upper shadow shows buyers tried to push higher and failed. The lower shadow shows sellers tried to push lower and failed. The small body shows that neither side won. It's indecision. On its own it means little. What matters is where it appears:

• In a downtrend: sellers may be pausing before another leg down, or buyers may be starting to take control. Varsity treats both outcomes as roughly equally likely. If you've been waiting to buy, consider taking half your intended size (say 250 of 500 shares) and adding only if price turns up.
• In an uptrend: buyers may be pausing before another leg up, or tiring. If you're holding a profit, consider booking part of it (say 50%) and riding the rest. If the trend continues you still own half. If it reverses you've already sold half higher.

Varsity calls it "the calm before the storm". A move is coming, but the direction isn't known.

11.2 Doji

A doji is a spinning top with (virtually) no body: open ≈ close. The shadows can be any length. Doji and spinning tops often appear in clusters, and they mean the same thing: indecision. Varsity accepts a "wafer-thin" body as a doji (rule 2 again).

12. Paper umbrella: hammer and hanging man

A paper umbrella has a small body at the top of the range, a long lower shadow, and little or no upper shadow. Its name depends entirely on the prior trend:

• At the bottom of a downtrend it's a hammer (bullish).
• At the top of an uptrend it's a hanging man (bearish).

Hammer logic. Sellers have been in control. On hammer day they push price to a new low as expected, but then buyers step in and drive price back to close near the high. Buyers have shown, for the first time in a while, that lower prices are being rejected. The longer the lower shadow, the stronger the signal. Colour doesn't strictly matter, though a green body is more reassuring. Trade: go long. The risk-taker buys near the close on hammer day after checking at about 3:20 pm that open ≈ close and the lower shadow is at least twice the body. The risk-averse buys the next day only if it closes green. Stop: the hammer's low.

Hanging-man logic. Buyers have been in control. On this day sellers manage a deep intraday drop, which shows their first real presence. The trade is a short (or a reason to protect a long), with the stop at the candle's high. Varsity admits it trusts the hanging man less than the hammer. If sellers were really strong, why did price recover to close near the high? Our data agrees (Section 19).

13. Shooting star

A shooting star is an upside-down paper umbrella: a small body near the low of the range and a long upper shadow at least twice the body. It needs a prior uptrend. A red body is a little more reliable, and the longer the upper shadow, the more bearish the signal.

The logic. Buyers have been in control. On shooting-star day price makes a new high as expected, then sellers hit it hard enough to push it back near the low by the close. The long upper shadow is a record of buyers who bought high and are now losing money. The expectation is that selling continues. Trade: short, or reduce a long, near the close (risk-taker), or after a red next day (risk-averse). Stop: the candle's high.

14. Engulfing patterns

Now for two-candle patterns. Call the first day P1 and the second P2. In an engulfing pattern, P2's body is opposite in colour to P1's and big enough to swallow P1's body completely.

Bullish engulfing logic. In a downtrend, P1 is another red day at a new low. P2 opens at or below P1's close and may even dip lower, then buyers arrive strongly enough to close above P1's open. In one session the bulls have undone the whole previous day. That surprises sellers and may start a reversal. Trade: buy near P2's close (risk-taker). Varsity personally prefers this for multi-candle patterns. Or buy at the close of the day after P2 if it's green (risk-averse). Stop: the lowest low of P1 and P2. The bearish version is the exact mirror: short near P2's close, with the stop at the highest high of P1 and P2.

14.1 A doji right after an engulfing pattern

Varsity describes a setup it finds especially profitable: a bearish engulfing at the top of a rally, then a doji the next day, then a big red candle. The reasoning is that P2 shocks the bulls, the doji leaves them uncertain, and panic plus uncertainty leads to a sharp fall. The broader lesson is to combine patterns and read the chart as a story rather than one candle at a time. We tested this specific claim. Across 37 stocks over 10 years, 323 bearish engulfings were followed by a doji or spinning top. Ten sessions after that third day, the stock was on average +1.69% higher, compared with +1.48% for the 1,230 engulfings without one. The doji did not make the bearish pattern work better in this sample. Read the story, but test the claim.

15. Piercing pattern and dark cloud cover

These are the "partial" versions of engulfing. P2 recovers more than half, but not all, of P1's body.

15.1 Choosing between similar patterns

Stocks in the same sector (TCS and INFY, ICICIBANK and HDFCBANK) often move together, but not identically. On the same day one might print a bearish engulfing and the other a dark cloud cover. All else equal, Varsity prefers the full engulfing, since the stronger move shows more conviction. It also prefers bullish engulfing over piercing. But "all else" rarely is equal. Varsity's later trading checklist asks that a trade meet most of these: (1) a recognisable candlestick pattern, (2) support or resistance near the stop, (3) volume confirmation, (4) agreement with the Dow theory trend (the primary trend defined by higher highs and higher lows, or the reverse; Chapter 3), (5) an acceptable reward-to-risk ratio, (6) indicator confirmation. A weaker pattern that meets 4 checklist items beats a stronger pattern that meets 3. Chapters 3–5 build the rest of that checklist.

16. Harami

Harami is an old Japanese word for "pregnant". A long P1 candle "carries" a small P2 candle of the opposite colour whose body sits inside P1's body.

The logic. In a downtrend, P1 is a big red candle at a new low, and sellers look completely in control. P2 then opens higher than P1's close, which sellers didn't expect, and closes green. The candle is small, but it appears when nobody expected it, and that can trigger short-covering (traders who sold short buying back to close their positions) and new buying. A bearish harami is the reverse: after a strong green P1, P2 unexpectedly opens lower and closes red. Because P2 is small, a harami is a weaker signal than an engulfing, and it needs a genuine prior trend. Varsity shows a "harami" after a flat period and explicitly rejects it.

17. Gaps, morning star and evening star

17.1 Gaps

A gap on a daily chart means a stock opens away from the previous close, with no trades in between. A gap-up (open above yesterday's close) shows eager buyers, for example after unexpectedly good results announced in the evening. A gap-down shows eager sellers. For the stars below, "gap" has a specific meaning: the middle candle's body sits beyond the first candle's close, and the third candle opens back across the middle candle's body.

17.2 Morning star (bullish, 3 candles)

1. P1: a long red candle in a downtrend. Selling is speeding up.
2. P2: gaps down (opens below P1's close), then goes nowhere and forms a doji or spinning top. Sellers had the gap but couldn't extend it. If P2 is a doji, it's called a morning doji star.
3. P3: gaps up above P2's body and closes green above P1's open, wiping out P1 completely. (Without P2, P1 and P3 would form a bullish engulfing.)

Trade: both risk-taker and risk-averse buy near P3's close, since the three-day pattern is its own confirmation. Stop: the lowest low of P1–P3. The evening star is the mirror image at the top of an uptrend (long green P1, a doji or spinning top that gaps up, then a red P3 that gaps down and closes below P1's open), with the stop at the highest high of the three.

18. All patterns on one page

Varsity ends the candlestick chapters with the most important point: candlesticks are a way of thinking, not a list of shapes. Once you can read the buyer–seller story, an unnamed shape tells you just as much. You don't need 50 more pattern names.

19. Do candlestick patterns work? 10 years of NSE data

Every example above was chosen to teach something, and I deliberately showed failures next to successes. Hand-picked examples can still prove anything. To find out what patterns actually do, we need every occurrence.

19.1 Method

• Universe: the 37 stocks in the course data (large and mid caps, including RELIANCE, HDFCBANK, TCS, HAL and DIXON), daily data from Oct 2016 to Sep 2026. Events run from 18 Nov 2016 to 1 Sep 2026, so that every event has 20 sessions of future data. Windows touching known data glitches (TMPV's Oct-2025 demerger, a TRENT price discontinuity on 1 Jan 2026, CDSL's listing week) are excluded.
• Detection: the exact numeric rules in this chapter (ch02_patterns.detect).
• Measurement: the close-to-close forward return \(r_{t,h} = C_{t+h}/C_t - 1\) for h = 5, 10 and 20 sessions, measured from the pattern-day close (the risk-taker's entry).
• Baselines: (a) every day for the same stocks, n = 85,490. Over 10 sessions, the average stock day returned +0.84% and rose 54.8% of the time. That's the bull-market drift. (b) A matched baseline: every day that followed the same kind of trend. Days after a downtrend averaged +0.63% (n = 19,249) and days after an uptrend +0.99% (n = 26,684).
• Mechanical trade: enter at the pattern close and exit at Varsity's stop (or at the open if price gaps through it) or after 10 sessions, measured in R. As a yardstick, the same test was run with a long entry on every day (stop at that day's low) and a short entry on every day (stop at that day's high).

19.2 Results

Returns are averages of close-to-close percentage moves after the pattern. For bearish patterns, a positive number means the stock went up, which is bad for a short. "Avg R" is before brokerage, STT (securities transaction tax) and slippage (the gap between the price you planned and the price you got); Chapter 13 covers costs.

19.3 What the numbers say

1. Bullish patterns barely beat doing nothing. Bullish engulfing averaged +0.92% over 10 sessions, against +0.63% for any day after a downtrend. The 95% interval (±0.61 points) easily includes the baseline. (A 95% interval is the range the true average could plausibly sit in, given how much the individual results vary. If it contains the baseline, the difference may just be luck.) The hammer did worse than the baseline (+0.19%). Bullish marubozus did a little better (+0.95%, up 58.7% of the time) than the +0.84% all-day average.
2. Bearish patterns, on their own, lost money for shorts. After bearish engulfing, shooting star, bearish marubozu and evening star, prices on average went up over the next 10–20 sessions. After 1,551 bearish engulfings the average was +1.34%, more than the +0.99% for any day after an uptrend. Strong stocks in a bull market tend to keep rising, and one bearish candle didn't change that. As shorts, every bearish pattern had a negative average R. They were only slightly less bad than shorting on a random day (−0.27R).
3. The R numbers don't rescue the patterns. All bullish patterns show a positive average R, but so does buying on any random day with a stop at that day's low (+0.39R, driven by the market's upward drift). The median R was −1 for most patterns. Typically the stop was hit, and a few large winners pulled the average up. This is before costs.
4. The stars look best, but the samples are tiny. 27 morning stars at +2.32% have a 95% interval of roughly +0.4% to +4.3%. That's very wide, and 27 cases are too few to rely on.
5. Stops get hit a lot. Within 10 sessions the stop was hit 53% of the time for bullish engulfing, 66% for hammers, 73% for shooting stars and 93% for hanging men. Patterns with stops very close to the entry are stopped out by ordinary noise.
6. Results drift over time. Bullish engulfing averaged +1.50% (10d) in 2016–mid-2021 and +0.48% in mid-2021–2026. Shooting star went from +1.19% to +0.69%. An "edge" that changes this much between periods is weak.

19.4 Does context help?

Varsity never meant patterns to be traded alone. They're the first item on a checklist. So we split the bullish patterns by three common filters:

Honest reading: no simple filter turned the hammer or bullish engulfing into a reliable edge. Some filters even hurt, and the samples shrink quickly. The marubozu near a 50-day low (+2.84%, 79 cases) is interesting but too small and too likely to be luck to trade on as it stands. Crude filters like these aren't proper context. Chapter 3's support and resistance are drawn more carefully, Chapter 4 adds indicators, and Chapter 12 shows how to test a full rule set (entry + trend filter + stop + target + position size + costs) without fooling yourself.

20. Try it yourself

These snippets need a copy of the course repository and Python with pandas and numpy (setup steps are in its framework/README.md). Run them from the repository's root folder, for example with .venv/bin/python, because the file paths are relative to that folder. The full detector and every figure in this chapter can be rebuilt with .venv/bin/python course/scripts/ch02_figures.py.

20.1 Resample daily candles to weekly and monthly

import pandas as pd
DATA = "data/ohlc/"
df = pd.read_csv(DATA + "IDX_NSEI.csv", index_col=0, parse_dates=True)
agg = {"Open": "first", "High": "max", "Low": "min", "Close": "last"}
weekly = df[list(agg)].resample("W-FRI").agg(agg).dropna()
monthly = df[list(agg)].resample("ME").agg(agg).dropna()
print(weekly.loc["2026-09-25"].round(2))      # the week of 21-25 Sep 2026
print(len(df.loc["2025"]), "daily candles in 2025")

Expected: Open 23,330.20, High 23,489.00, Low 23,020.95, Close 23,140.50 (matching Section 7), and 249 daily candles in 2025.

20.2 Find hammers with Varsity's rules, quantified

import pandas as pd
DATA = "data/ohlc/"
df = pd.read_csv(DATA + "BAJFINANCE.csv", index_col=0, parse_dates=True)
o, h, l, c = df.Open, df.High, df.Low, df.Close
body  = (c - o).abs()
upper = h - pd.concat([o, c], axis=1).max(axis=1)
lower = pd.concat([o, c], axis=1).min(axis=1) - l
rng   = h - l
# prior downtrend: down >= 3% over 10 sessions AND below the 20-day average (measured on yesterday's close)
prev = c.shift(1)
down = (prev / c.shift(11) - 1 <= -0.03) & (prev < c.rolling(20).mean().shift(1))
size_ok = (rng / prev).between(0.01, 0.10)          # Varsity's "trade trap" filter
hammer = (lower >= 2 * body) & (upper <= 0.10 * rng) & (body > 0.02 * rng) & size_ok & down
print(df.loc[hammer, ["Open", "High", "Low", "Close"]].loc["2026"].round(2))

Expected 2026 hammers: 22 Jan, 23 Mar (the example in Section 12), 3 Jun and 11 Sep. Try it: change 2 * body to 3 * body and watch how many disappear. That's rule 2, "quantify", in action.

20.3 Bearish engulfing and what happened 10 sessions later

import pandas as pd
DATA = "data/ohlc/"
df = pd.read_csv(DATA + "HDFCAMC.csv", index_col=0, parse_dates=True)
o, c = df.Open, df.Close
o1, c1 = o.shift(1), c.shift(1)
up = (c.shift(2) / c.shift(12) - 1 >= 0.03) & (c.shift(2) > c.rolling(20).mean().shift(2))
bear_engulf = (c1 > o1) & (c < o) & (o >= c1) & (c < o1) & up
fwd10 = c.shift(-10) / c - 1                       # return 10 sessions later
res = pd.DataFrame({"close": c, "fwd10_%": fwd10 * 100})[bear_engulf]
print(res.loc["2026"].round(2))
print("all signals:", len(res.dropna()), " mean fwd10 %:", round(res["fwd10_%"].mean(), 2),
      " baseline fwd10 %:", round(fwd10.mean() * 100, 2))

Expected 2026 signals: 6 Feb at 2,719.80 → −0.15%, and 18 Feb at 2,812.60 → −9.02%, which is the Section 14 example. Across all 41 HDFCAMC signals the average 10-session return was +1.96%, against a +0.71% baseline. On this stock the bearish signal was followed by above-average gains, matching the 37-stock result. Repeat with your favourite stocks.

21. Practical checklist: reading candles before a trade

• I've looked at the weekly chart first. I know the bigger trend and I'm not fighting it without a reason.
• On the daily chart I can name the prior trend into this pattern (≈ 3%+ over 10 sessions, on the right side of the 20-day average).
• The pattern meets my written numeric rule (body/shadow ratios, engulf or penetration conditions, gaps).
• The signal candle's range is between 1% and 10% of the previous close (trade-trap check).
• I've decided in advance whether I'm the risk-taker (enter on the completion day: ~3:20 pm for non-F&O stocks; for F&O stocks, check by 3:15 pm and use the closing auction, or the futures before 3:40 pm) or the risk-averse trader (enter only after a confirming candle).
• The stop is at the pattern's lowest low (long) or highest high (short). If that's absurdly tight (hanging man), I've widened it to a logical level and resized.
• Position size = (capital × risk%) ÷ risk per share. Risk per trade ≤ 1–2% of capital.
• There's a reason beyond the candle: support or resistance near the stop, a volume surge, a trend filter, or relative strength (Chapters 3–5).
• For short trades I know which instrument I'll use (stock futures or options). Otherwise I treat the bearish candle as an exit or trim signal for my long.
• I'll log the trade with the pattern name so I can measure my own results later (Chapter 12).

22. Key takeaways

23. Quiz and exercises

24. Sources

Zerodha Varsity, Module 2 "Technical Analysis" (concepts explained here in our own words): Ch 1 Background · Ch 2 Introducing Technical Analysis · Ch 3 Chart Types · Ch 4 Getting Started with Candlesticks · Ch 5 Single Candlestick Patterns (Part 1) · Ch 6 Single Candlestick Patterns (Part 2) · Ch 7 Single Candlestick Patterns (Part 3) · Ch 8 Multiple Candlestick Patterns (Part 1) · Ch 9 Multiple Candlestick Patterns (Part 2) · Ch 10 Multiple Candlestick Patterns (Part 3) · Ch 11 Support & Resistance (trading checklist).
NSE closing mechanism (as of Aug 2026): ICICI Direct: Closing Auction Session step-by-step guide · All India Radio News (2 Aug 2026): NSE to introduce Closing Auction Session. Check against the latest NSE circulars at nseindia.com.
Background reading: Steve Nison, Japanese Candlestick Charting Techniques (1991).
Data: daily OHLCV for 37 NSE stocks and NIFTY 50, Oct 2016 – 30 Sep 2026 (NSE via Yahoo Finance; prices adjusted for splits and bonuses, so older prices may differ from what was quoted on the day). Figures and statistics: course/scripts/ch02_figures.py and course/scripts/ch02_patterns.py.
All examples are historical analysis for education, not recommendations to buy or sell any security. Trading, especially with leverage, can lose more than you expect.