Risk Management & Trading Psychology

1. What you'll learn and why it matters

Chapters 2 to 10 were about finding trades and investments. This chapter covers the part that decides whether you are still in the game after five years: how much you bet, how much you can lose, and how you behave while losing. Most traders spend 90% of their effort on entries. Most of the damage comes from the other 10%.

The numbers are stark. SEBI's study of individual traders in equity derivatives for FY25–FY26 found that 87.7% of individual F&O traders made a net loss in FY26, with aggregate net losses of about ₹91,685 crore. Even before transaction costs, 82.1% were losing. Few of those people lacked information or chart patterns. What most lacked was position sizing, a stop they respected, and a process that kept their emotions out of the order window.

By the end of this chapter you will be able to:

• Name the kinds of risk you carry (systematic, company-specific, liquidity, gap, leverage, concentration, correlation) and say which ones diversification removes and which ones it cannot touch.
• Compute variance, covariance, correlation and portfolio volatility by hand and with a correlation matrix built from real NSE data, then trace an efficient frontier, and see why it fails out of sample.
• Estimate a portfolio's one-day Value at Risk three ways and explain why VaR broke down in March 2020.
• Size any trade: fixed amount, percentage margin, the 1–2% risk rule, ATR/volatility sizing, equity-curve sizing, Kelly and half-Kelly, and futures lots.
• Calculate R-multiples, expectancy and break-even win rate, and use Monte Carlo simulation to see your probability of a 30% or 50% drawdown.
• Recognise the biases that wreck traders (loss aversion, disposition effect, overconfidence, confirmation, recency, anchoring, FOMO, revenge trading) and put a written rule against each one.
• Keep a trading journal, run a monthly review, and write your own one-page trading plan.

About the data: every number in this chapter comes from course/scripts/ch11_figures.py, which reads the course's daily NSE data (to 30 Sep 2026), the NSE futures bhavcopy and SPAN file for 30 Sep 2026, and the 2,187 completed historical trades from the four swing setups of Chapter 5 (S1 trend pullback, S2 52-week breakout, S3 support bounce, S4 weekly trend + daily trigger, all with R net of costs; R means a trade's profit or loss measured in units of the risk planned at entry, so −1R is a normal stop-out, defined fully in Section 11). All the numbers are saved in data/extra/ch11_results.json.

2. What "risk" means, and the kinds you will face

In everyday language risk means "something bad might happen". In markets it has two working meanings. Keep both in mind:

• Risk as volatility: how widely returns swing around their average. This is what portfolio maths measures (standard deviation, variance, VaR).
• Risk as permanent loss: the chance of losing money you never get back, because you sold at the bottom, a company failed, or leverage forced you out. Traders and investors actually fear this one, and position sizing and stops are built to control it.

2.1 Systematic vs unsystematic risk

Varsity starts the module with this split, and it is the most useful one to learn first.

• Unsystematic (company-specific) risk lives inside one business: a bad quarter, a management exit, a new competitor, a regulatory order against that company. It hits the stock and usually not its peers. Diversification reduces it, because bad news at one company is offset by ordinary days at the others.
• Systematic (market) risk is common to the whole system: interest rates, a global sell-off, war, a pandemic, a tax change on all equities. Every stock is exposed. Diversification cannot remove it. You can only reduce it by holding less equity, holding cash or uncorrelated assets, or hedging with index futures or puts (Chapters 9 and 10).

2.2 The other risks a trader actually meets

The table needs one reading habit: most of these risks are invisible on a chart while you are entering. A clean breakout chart doesn't show that it is your fourth bank trade this week, that results are due in two days, or that the stock trades only ₹30 crore a day. Your risk process has to ask those questions for you.

3. Diversification maths: variance, covariance, correlation

Varsity spends four chapters building portfolio risk from first principles. The logic is simple even though the algebra looks heavy: portfolio risk depends on each stock's own volatility and on how the stocks move together. The "together" part is what diversification works on.

3.1 Variance and standard deviation of one stock

3.2 Covariance and correlation

Covariance asks whether two stocks tend to be above their averages on the same days. The raw number is hard to read, so we scale it into correlation: +1 means they move in lock-step, 0 means no linear relation, −1 means they move in opposite directions. Over the same five days, ICICIBANK's returns gave a covariance with HDFCBANK of −0.0000424 and a correlation of −0.19. Five days of noise, though. Over two years the correlation is +0.51. Always use enough data.

3.3 Two-stock portfolio: where the diversification benefit comes from

The first two terms are each stock's own risk. The third term is the interaction. If \(\rho=1\), the formula collapses to \(\sigma_p = w_1\sigma_1+w_2\sigma_2\): a plain weighted average, so no benefit at all. Any \(\rho<1\) makes \(\sigma_p\) smaller than the weighted average. That gap is the "free lunch" of diversification.

3.4 The correlation matrix: see your hidden bets

With many stocks you need every pairwise correlation. Arranged in a grid, they form the correlation matrix. It is the single most useful risk picture for a swing trader with several open positions, because it shows which "different" trades are really the same trade.

3.5 How many stocks are enough?

If you hold \(n\) equal-weight stocks, portfolio variance is approximately

As \(n\) grows, the first term (company-specific risk) shrinks towards zero. The second term (average covariance, the market risk) stays. That is the maths behind "diversification removes unsystematic risk but not systematic risk".

4. Portfolio variance for many stocks, the equity curve and expected range

Varsity uses matrix algebra to extend the two-stock formula to any number of stocks. The steps are worth knowing once, because Python and Excel then do them for you.

4.1 The equity curve

Varsity's chapter 6 normalises the portfolio to ₹100 at the start and tracks its value each day. This equity curve is the honest summary of an investment or a trading system: return, path and drawdowns in one picture. Always plot it against a benchmark.

4.2 Expected return and a realistic range

5. Portfolio optimisation and the efficient frontier, and its limits

Harry Markowitz's insight, which Varsity's chapters 8–9 walk through with Excel Solver: for any set of stocks you can try every combination of weights and plot each portfolio's risk against its expected return. Two special portfolios stand out:

• the minimum-variance portfolio, the weights that give the lowest possible volatility;
• the efficient frontier, the upper edge of the cloud: for each level of risk, the highest expected return available. Any portfolio below the frontier is "inefficient": another mix gives more return for the same risk.

6. Value at Risk: historical, variance-covariance and expected shortfall

Value at Risk (VaR) answers one question: "On a normal bad day, how much could this portfolio lose?" Formally, the 95% one-day VaR is the loss that should be exceeded on only 5% of days (about one day in twenty, or 12–13 days a year).

A note on vocabulary: Varsity's VaR chapter sorts the portfolio returns and takes the average of the worst 5%. Textbooks call that number expected shortfall or CVaR, and reserve "VaR" for the 5th-percentile cut-off itself. Both are useful. CVaR is always the larger loss, because it averages the tail beyond the cut-off. We compute both below.

6.1 When VaR fails: March 2020

7. Why position sizing comes first: gambler's fallacy and recovery maths

Varsity opens its position-sizing chapters with a poker story: lose the first two small bets, raise the third to "win it back", lose that too, and the buy-in is gone in ten minutes. Two ideas explain why that happens to traders.

7.1 Gambler's fallacy

The gambler's fallacy is the belief that after a run of losses a win is "due". If your trades are roughly independent, the odds on the next trade are the same as they were on the first. A system with a 38% win rate still has a 62% chance of losing the next trade after eight losses in a row. Raising size after losses ("martingale" thinking) leaves your odds unchanged and puts the biggest bets in the worst stretch.

7.2 Recovery maths: losses and gains are not symmetric

8. How much capital do you really have? Three equity models

Every sizing rule says "risk X% of your capital". So what is your capital while some of it is already tied up in open trades? Varsity (following Van Tharp) gives three models:

9. Position sizing methods for shares

Varsity's chapter 13 says it neatly: with 3 ways to measure equity and 4 ways to size, you have 12 combinations. Here are the methods, from crude to refined, with real numbers. The base case throughout: ₹5,00,000 account, prices as of 30 Sep 2026.

9.1 Unit per fixed amount (and fixed rupee amount)

"Trade 1 lot per ₹2 lakh" or "put ₹1 lakh into every position." Simple, and it scales with the account. Its flaw: it ignores risk. A volatile stock and a sleepy one get the same money, so a stop hit costs very different amounts.

In the table below the stop is placed 2 × ATR(14) below the close. ATR(14), the Average True Range, is the average daily trading range over the last 14 sessions including overnight gaps (Wilder's method, Chapter 4); Section 9.3 uses it for sizing.

9.2 Percentage risk: the 1–2% rule

This is the method most professionals use and the one this course recommends. Varsity notes that professional traders rarely risk more than 1–3% of capital on one trade. You decide the maximum loss per trade as a percentage of equity, and the distance to your stop then sets the quantity.

9.3 Percentage volatility (ATR) sizing

Varsity's fourth method, from Van Tharp, sizes the position so that one average day's move (ATR) equals a fixed % of equity. It uses the Average True Range rather than standard deviation, because ATR includes gaps (Chapter 4).

9.4 Equity-curve sizing

A refinement used by many systematic traders: risk your normal % while your account's equity curve is healthy, and cut size (for example in half) when the equity curve drops below its own moving average. The logic: losing streaks cluster when a system is out of sync with the market (Section 13), so betting less during those stretches limits the damage, and you return to full size once results recover.

10. Position sizing for futures lots

Futures add two complications: you trade in whole lots, and the margin is a small fraction of the exposure. Margin is what the exchange asks you to deposit. It is not your risk. Size by the rupee loss at your stop, exactly as for shares, and then check that the margin fits.

10.1 Varsity's percentage-margin model

For futures, Varsity also describes capping the margin used per position at a fixed % of equity (for example 20%), and pairing it with the core or reduced-total equity model.

11. R-multiples, expectancy and the break-even win rate

Once every trade risks a planned amount, you can measure results in R, units of initial risk. This makes trades of different sizes and prices comparable, and turns a trading method into numbers you can test.

12. Kelly criterion and half-Kelly

The Kelly criterion, which Varsity covers in chapter 14, answers: given my edge, what fraction of capital should I bet to grow fastest over the long run? Bet too little and you grow slowly. Bet too much and the losses (which hurt more than equal-sized gains help, Section 7) drag growth down and eventually into ruin.

13. Drawdowns, losing streaks and risk of ruin

13.1 Same trades, different bet size

Here is the most important experiment in this chapter. Take the exact same 906 S1 trades in the order they happened (2017 – Sep 2026) and replay them risking 1%, 2% and 5% of current equity per trade.

13.2 Monte Carlo: what could have happened

One historical sequence is one possible path. A Monte Carlo simulation reshuffles your real trade results thousands of times to show the range of paths your system could produce. We drew 200 trades at random (with replacement) from S1's 906 net-R outcomes, 10,000 times, at each risk level.

Three things to read from this table:

1. Beyond about 3% risk, more risk does not even raise the typical outcome. The median ending at 10% (0.47×) is below the median at 0.5%. Over-betting a positive edge destroys it, which is the Kelly curve again.
2. Drawdowns are larger than intuition suggests. Even at 1%, there is a 36% chance of a 20% drawdown within 200 trades. Decide now that a 20% drawdown at 1% risk is "normal bad", not a reason to abandon a tested plan.
3. A positive edge still loses money 27–29% of the time over 200 trades at low risk. A thin edge needs many trades to show itself. Re-run with S1's 2024–26 trades only (mean −0.12R): at 1% risk the median ending is 0.77× and 92% of paths lose money. A negative edge cannot be fixed by sizing.

What "risk of ruin" means here. Risk of ruin is the probability that a run of losses takes your account down to a level from which you cannot or will not carry on. If you always risk a fixed percentage of current equity, the account never mathematically reaches zero (each loss is a slice of a smaller pie), so the textbook "probability of losing everything" is not the useful number. In practice, traders quit, change systems or start revenge trading long before zero. So we define ruin as a 50% drawdown (which then needs +100% to repair) and measure its probability by simulation, as in the table above.

13.3 Losing streaks are normal

14. Portfolio heat, correlation risk and sector limits

The 1% rule controls one trade. Portfolio heat controls all of them together:

Heat is the total you would lose if every stop were hit at once. Typical caps: 4–6% for swing traders (for example 5 positions × 1%), lower in a weak market regime. Once a stop is moved to breakeven or above, that position's heat drops to zero, and that is how you make room for new trades.

Heat counts only losses to your own capital. It ignores open profit you could give back: a stock up 15% with its stop at breakeven has zero heat but can still hand back that 15%. Track that separately as open profit at risk, qty × (current price − stop), and trail stops on big winners so it stays reasonable.

14.1 Practical limits that work

These are educational starting points, not universal truths. Write your own numbers into your plan (Section 17) and test them with the Monte Carlo above.

15. Trading psychology: the biases that cost money

Everything above assumes you will follow your rules. Most traders do not, at least not when it matters. Varsity's last two chapters and its Innerworth module describe the mental shortcuts (cognitive biases) that push otherwise sensible people into bad decisions. Biases are not character flaws. They are built into how human brains handle uncertainty, so trying harder doesn't remove them. Written rules that act before the bias does are what works.

15.1 Loss aversion and the disposition effect

Loss aversion: a loss hurts roughly twice as much as an equal gain pleases. In trading it shows up as the disposition effect: we sell winners too early (to "lock in" the good feeling) and hold losers too long (to avoid making the loss "real"). That is the exact opposite of what a trend-following edge needs.

15.2 Overconfidence and attribution bias

After a run of wins we overrate our skill and raise size. That is exactly when a regime change does the most damage. Varsity's attribution bias is the mirror image: wins are "my skill", losses are "bad luck / operators / the market was irrational". Both stop you learning.

15.3 Confirmation bias

Once we own a stock we look for news that supports it and skip the rest. Varsity's example is the trader who reads only bullish reports on a stock he holds. The effect: stops get "reconsidered", warning signs get explained away, and position sizes creep up.

Rules: before entry, write one sentence on what would prove me wrong. That sentence becomes the stop. For investments, write the bear case alongside the bull case (Chapter 8) and list 2–3 measurable "thesis-broken" triggers (e.g. two consecutive quarters of falling ROCE, promoter pledge rising). Run a pre-mortem: "It's 6 months later and this lost 30%. What happened?"

15.4 Recency bias

We weight the last few weeks far more than the last few years. After a strong run every dip "will be bought". After a crash every rally "is a trap". Varsity's antidote is to keep the bigger picture in view.

15.5 Anchoring

We fix on a reference number, usually a past high or our purchase price, and judge everything relative to it. "It was ₹4,500, now it's ₹3,600, it must be cheap." The market does not know or care what you paid or where the stock used to trade.

Rules: judge a stock on its own current trend (Chapters 3–5) and valuation (Chapter 8), never on its distance from a past high. Never average down a losing trade. If an investment is to be added to on a fall, that must be pre-planned in the thesis, with a fresh valuation, not a reaction to the price.

15.6 FOMO and the bandwagon (herd) effect

Fear of missing out: a stock everyone is talking about has run, and you buy because you cannot bear to watch it go higher without you. Late buyers tend to arrive after the move, with stops far away or no stops at all.

Rules: no trade without a setup from your plan. If you "missed it", write it on a watchlist and wait for a pullback setup. Mute social media and tips during market hours.

15.7 Revenge trading and the gambler's fallacy

After a painful loss, the urge is to "win it back" immediately, with a larger size, a lower-quality setup, or an instrument you don't normally trade (often options near expiry). Varsity's poker story is a revenge-trading story. It pairs naturally with the gambler's fallacy (Section 7): "I've lost four in a row, the next one must work."

Rules: a daily loss limit (e.g. 2% or 2R): hit it and you are done for the day. A cooling-off rule: after any loss above 2R, or three losses in a row, no new trades until the next session and a journal entry. Size never increases after a loss.

15.8 Illusion of control, functional fixedness and hindsight

• Illusion of control (Varsity ch. 15): piling indicators, timeframes and data feeds onto a chart because it feels like more control. More data is not more information. Chapter 4 showed most indicators are transforms of the same price series. Rule: at most 2–3 tools per setup, each with a defined job.
• Functional fixedness (Varsity ch. 16): using a tool only the way you first learned it (RSI above 70 = sell, always), and being blind to other uses or to when it stops working. Rule: know the logic behind each tool, as this course has tried to teach, so you can tell when the context has changed.
• Hindsight bias: after the fact, the move "was obvious". It makes you overconfident about predicting the next one and harsh about past decisions that were sound but unlucky. Rule: judge decisions by whether they followed the plan, not by the outcome of a single trade.

16. The trading journal and review process

Your backtest says what should happen. Your journal records what did happen: what you did, why, and how it felt. Without one you cannot measure your expectancy, you cannot tell bad luck from bad execution, and every bias above goes unchecked.

16.1 What to record for every trade

16.2 The review cycle

• Daily (5 minutes, after the close): update open trades and stops, compute heat, log any rule breaks. No new analysis while emotional.
• Weekly (30 minutes): review every closed trade's screenshot. Were entries per plan? Any patterns in mistakes (time of day, after losses, a sector)?
• Monthly / every 20–30 trades: compute live win rate, average R, expectancy and max drawdown. Compare with the backtest. Compute expectancy separately for "followed plan" and "broke plan" trades. That gap is the price of your indiscipline in rupees.
• Quarterly: decide changes. Only change a rule if the data (30+ trades) supports it, and change one thing at a time.

17. Your written trading plan (template)

A trading plan is a one- or two-page document you write when calm and follow when not. If something isn't in the plan, you don't do it. Copy this template and fill it in. Chapter 17 turns it into your complete framework.

MY TRADING PLAN — version 1.0 — date: ________
1. Goals & capital
   Trading capital: Rs ______ (money I can lose without changing my life)
   Investment capital (separate): Rs ______   Emergency fund kept outside both: yes/no
   Realistic goal: positive expectancy over 100 trades, max drawdown under ___%

2. Markets & instruments
   Swing: NSE cash, Nifty 200 stocks, avg daily value > Rs 50 cr
   Futures: only when capital >= Rs ___ lakh per lot at 1% risk (see Section 10)
   Not traded: intraday, options buying near expiry, illiquid small caps

3. Setups (from Ch 5 / Ch 12, each backtested, with its expectancy)
   S_: __________  entry rule / stop rule / exit rule / expectancy ___R over ___ trades

4. Risk rules
   Risk per trade: ___% of core equity (start 0.5%; max 1%)
   Max position: ___% of equity      Max portfolio heat: ___%
   Max per sector / correlated cluster: ___ trades or ___% heat
   Regime: Nifty < 200-DMA -> risk halved, trend setups paused
   Futures: margin used <= ___% of capital; never add to a losing position

5. Circuit breakers
   Daily loss limit: ___R  -> stop for the day
   After 3 consecutive losses or a loss > 2R: no trade until next session + journal entry
   Monthly loss of ___% -> half size;  drawdown of ___% from peak -> stop, review, paper-trade

6. Routine
   Pre-market: check regime, events/results calendar, open positions' stops
   Entries: only at planned levels with orders placed; no chasing
   Post-close: journal, update stops, compute heat
   Weekly / monthly / quarterly reviews (Section 16)

7. Personal rules
   No trading when ill, sleep-deprived, or after a major personal upset
   No tips, no social-media trades, no changing rules during market hours

Signed: ____________   Review date: ____________

18. Stress and discipline routines

Varsity's Innerworth module comes back again and again to one theme: trading performance depends on your state of mind as much as on your method. Stress narrows attention and pushes you towards impulsive, short-term decisions, which are exactly the biases above. A few habits help:

• Treat a loss as feedback, not failure. A stopped-out trade that followed the plan is a business expense. Ask "what is this telling me?" (setup, market regime, execution) rather than "why me?".
• Be flexible and disciplined. Discipline means following the plan. Flexibility means changing the plan through the review process when data says so, never mid-trade.
• Size for sleep. If you are checking prices at night, the position is too big. Cut it until you can switch off. That is the human version of the Monte Carlo table.
• Decide before the market opens. Orders (entries, stops) placed in advance remove most in-the-moment decisions. Swing and positional trading's great advantage over intraday is that you can do your thinking when the market is closed.
• Limit screen time. Checking prices every few minutes on a swing position feeds anxiety and itchy exits. Twice a day is plenty for a daily-chart strategy.
• Physical basics matter. Sleep, exercise and an organised workspace measurably improve decision quality. Don't trade on days when you are unwell, exhausted or emotionally upset.
• Face fear and regret directly. Name the fear ("I'm afraid this gap-down will continue") and check it against the plan, instead of acting to make the feeling go away.
• Take planned breaks after a circuit breaker triggers or a big win. Both distort judgement.

19. 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. They use the course data directly. Your numbers can differ from the chapter in the second decimal: the chapter's script also drops a few zero-range holiday placeholder bars and uses population (n) rather than sample (n−1) standard deviation for the matrix example.

19.1 Correlation matrix and portfolio volatility

import pandas as pd, numpy as np
D = "data/ohlc/"
stocks = ["HDFCBANK", "INFY", "ITC", "SUNPHARMA", "TITAN"]
w = np.array([0.25, 0.20, 0.15, 0.15, 0.25])
px = pd.DataFrame({s: pd.read_csv(D + s + ".csv", index_col=0, parse_dates=True)["Adj Close"] for s in stocks})
ret = px.pct_change(fill_method=None).loc["2024-10-01":"2026-09-30"].dropna()

print(ret.corr().round(2))                       # correlation matrix
cov = ret.cov()                                  # variance-covariance matrix (daily)
port_sd = np.sqrt(w @ cov.values @ w)            # sigma_p = sqrt(w' S w)
print(f"Portfolio vol: {port_sd:.4%} per day, {port_sd*np.sqrt(252):.2%} per year")
print(f"Weighted avg of vols: {(w * ret.std()).sum()*np.sqrt(252):.2%} per year")
# -> about 0.85% per day, 13.5% per year vs ~22% weighted average

19.2 Value at Risk three ways

import pandas as pd, numpy as np
D = "data/ohlc/"
stocks = ["HDFCBANK", "INFY", "ITC", "SUNPHARMA", "TITAN"]
w = np.array([0.25, 0.20, 0.15, 0.15, 0.25])
px = pd.DataFrame({s: pd.read_csv(D + s + ".csv", index_col=0, parse_dates=True)["Adj Close"] for s in stocks})
pr = (px.pct_change(fill_method=None).dropna() @ w).loc[:"2026-09-30"].iloc[-252:]   # last year of daily returns
value = 1_000_000
hist = np.percentile(pr, 5)
cvar = pr[pr <= hist].mean()
param = pr.mean() - 1.645 * pr.std()
print(f"Historical VaR95: {hist:.2%} = Rs {-hist*value:,.0f}")
print(f"Expected shortfall (Varsity 'VaR'): {cvar:.2%} = Rs {-cvar*value:,.0f}")
print(f"Parametric VaR95: {param:.2%} = Rs {-param*value:,.0f}")
# -> about Rs 14,100 / Rs 20,300 / Rs 15,100

19.3 Position size, expectancy, Kelly and Monte Carlo with the framework

import sys; sys.path.insert(0, ".")
import numpy as np, pandas as pd
from framework.risk import position_size, kelly_fraction, risk_of_ruin_sim

# 1) Rs 5 lakh, 1% risk (risk_pct is in PERCENT), cap any one stock at 25% of capital
print(position_size(500_000, 1.0, 1220.0, 1206.41, max_position_pct=25))   # -> 102 (cap binds, not 367)

# 2) expectancy and Kelly from a list of R-multiples (here: Chapter 5's S4 trades)
R = pd.read_csv("data/extra/ch05_trades.csv")
R = R[(R.setup == "S4") & (R.exit_reason != "open at data end")].R_net
W = (R > 0).mean(); avg_w = R[R > 0].mean(); avg_l = -R[R <= 0].mean()
print(f"win {W:.1%}  payoff {avg_w/avg_l:.2f}  expectancy {R.mean():+.3f}R  Kelly {kelly_fraction(W, avg_w/avg_l):.1%}")

# 3) Monte Carlo: same edge at 1% vs 5% risk ('ruin' = 50% drawdown)
for risk in (1.0, 5.0):
    out = risk_of_ruin_sim(W, avg_w, avg_l, risk_pct=risk, n_trades=200, n_sims=5000)
    print(risk, round(out["prob_ruin"], 3), round(out["median_final"], 2), round(out["median_max_dd"], 3))

The framework's simulator uses a simple two-outcome model (every win = average win), which is kinder than reality. With S4's full-period numbers it shows a median max drawdown of about −10% at 1% risk and −42% at 5%. Replace R with your own journal's R-multiples once you have 30+ trades. Exercise: re-run with only trades from 2024 onward and see Kelly go negative.

20. Practical checklist

• Before any trade: I know my entry, my stop (a real order), and why the stop is where it is.
• Quantity comes from the formula: risk budget ÷ (entry − stop), using core or reduced-total equity.
• Risk per trade ≤ 1% (0.5% while learning or in a risk-off regime). Position ≤ my max position %.
• Futures: rupee loss at the stop for one lot ≤ my risk budget. If not, the answer is zero lots. Margin used ≤ my cap.
• Portfolio heat after this trade ≤ my cap. No more than my limit in one sector or correlated cluster.
• Results/events in the next few days? If yes, smaller size or no trade (gap risk).
• Liquidity: the order is well under 1% of the stock's average daily value.
• I am trading a setup from my written plan with known positive expectancy, not a tip, not FOMO, not revenge.
• After exit: journal entry with R, costs, "followed plan?" and one lesson.
• Monthly: live expectancy vs backtest; max drawdown vs Monte-Carlo range; circuit breakers checked.
• Investor: no stock above my cap, at least 5–6 unrelated sectors, a written "thesis-broken" rule for each holding.

21. Key takeaways

22. Quiz and exercises

23. Sources

Primary concepts (explained here in our own words): Zerodha Varsity Module 9, Risk Management & Trading Psychology: Ch 1 Orientation note · Ch 2 Risk (Part 1) · Ch 3 Risk (Part 2) – Variance & Covariance · Ch 4 Risk (Part 3) – Variance & Covariance Matrix · Ch 5 Risk (Part 4) – Correlation Matrix & Portfolio Variance · Ch 6 Equity Curve · Ch 7 Expected Returns · Ch 8 Portfolio Optimization (Part 1) · Ch 9 Portfolio Optimization (Part 2) · Ch 10 Value at Risk · Ch 11 Position Sizing for active trader · Ch 12 Position Sizing (Part 2) · Ch 13 Position Sizing (Part 3) · Ch 14 Kelly's Criterion · Ch 15 Trading Biases · Ch 16 Trading Biases (Part 2).

Psychology themes: Zerodha Varsity Module 13, Innerworth – Mind over Markets, especially Accurate Perceptions of Loss and Risk Aversion · Detailed Action Plans · Loss is Feedback, Not Failure · The Flexible and Disciplined Trader · Regret: A Powerful Emotion · It's Easier to Face Fear than Avoid It · Relieving Stress · The Herd Mentality · Everything Seems Obvious in Hindsight · Overconfidence, Risk, and Rewards · Creative, Winning Trading Plans.

Regulatory statistics: SEBI study "Profitability of Individual Traders in the Equity Derivatives Segment (FY25–FY26)" (20 Aug 2026), as reported by TaxGuru and CorpLaw Updates (87.7% of individual traders had net losses in FY26; 82.1% before costs; aggregate net loss ≈ ₹91,685 crore), announced on SEBI's official X account. Margins: NSE Clearing SPAN risk-parameter file for 30 Sep 2026 plus ELM of 2% (index) / 3.5% (stock) futures, as verified in Chapter 9.

Data: daily OHLCV for NSE stocks and indices to 30 Sep 2026 (NSE via Yahoo Finance), GOLDBEES ETF (Yahoo Finance), NSE F&O UDiFF bhavcopy (Aug–Sep 2026), Chapter 5 swing-setup trade list (data/extra/ch05_trades.csv). All figures and numbers: course/scripts/ch11_figures.py → course/img/ch11/ and data/extra/ch11_results.json. Educational material only, not investment advice; all examples are historical analysis.