Risk, return and asset classes

An asset that earns 10 % a year on average can still lose money in almost three years out of ten. In 10,000 synthetic years of a made-up asset whose returns swing by 18 % a year, 28.9 % ended below zero, and the worst one lost 56.9 %.

This lesson names what each asset class owns and the risk it carries, measures how far returns swing, separates a fall you can wait out from a loss that is permanent, matches a goal’s date to the classes that suit it, and shows why a promise of high, guaranteed returns is a warning. It is education only: every asset in it is synthetic or invented, it names no fund, scheme or security, and no number in it is a forecast or a recommendation.

Each asset class earns its return by taking a different risk

Picture five homes for the same ₹1,00,000: a share in a company, a loan to a company or a government, gold, a flat and a bank deposit. Each pays in its own way and can disappoint in its own way. A group of investments that pay and fail alike is an asset classasset class: A group of investments that pay and fail in the same way, such as equity (shares), debt (bonds and deposits), gold, real estate and cash. Each earns its return by carrying its own kind of risk.. SEBI’s investor education site describes the main ones: a share is ownership in a company, with returns that depend on how the company does and with higher risk and volatility; a bondbond: A loan to a company or a government in the form of a security: it pays interest and returns the amount lent on a set date. Its price moves with interest rates, and its issuer may fail to pay. is a loan that pays interest and carries the risk of interest-rate changes and of the issuer failing to pay; real estate earns rent or a rise in price but lacks liquidity; gold is held as a store of value and a hedge against inflation, with a price moved by economic, geopolitical and supply-and-demand forces.

Read the last column as the price of the third. Equity pays the most when things go well because nothing is promised back; a bond’s interest is promised, so its main danger is a borrower who cannot keep the promise; cash promises its balance, and its risk is the quiet one of inflation. Bank deposits are insured by DICGC up to ₹5 lakh per depositor in each bank, as who regulates your money shows, while shares, bonds and gold have no such cover. Mutual funds and exchange-traded funds are not classes of their own: they are ways of holding these classes, and a later lesson of this module takes them apart.

Volatility measures how far returns swing

Two assets can share an average and differ in everything else. The program below draws 10,000 synthetic years for three made-up assets: Steady averages 4 % a year and rarely strays from it, Middle averages 6.5 %, and Swinging averages 10 % but strays far. Volatilityvolatility: How far an investment's returns swing around their average, usually measured as the standard deviation of its returns over a year. It describes the size of the ups and downs, not a lasting loss. is the size of those strays, measured as the standard deviationA measure of how far values spread around their average: the square root of the average squared distance from it. For a bell curve, about 68 % of values lie within one standard deviation of the average. of yearly returns:

"""Ten thousand synthetic years for three made-up assets.

Each asset's yearly return is drawn from a bell curve (a normal distribution) with an assumed average and an assumed
standard deviation, the usual measure of volatility. The draws are synthetic and seeded, so every run prints the same
numbers. No real security, fund or index is described, and nothing here is a forecast of any return.
"""

import math
import random
import statistics

# name, assumed average yearly return, assumed standard deviation (volatility)
ASSETS = (
    ("Steady", 0.04, 0.01),
    ("Middle", 0.065, 0.05),
    ("Swinging", 0.10, 0.18),
)
YEARS = 10_000
SEED = 7


def pct(x, places=1):
    """A fraction as a percentage with a true minus sign: -0.234 -> "−23.4 %"."""
    value = round(x * 100, places)
    return ("−" if value < 0 else "") + f"{abs(value):.{places}f} %"


def share_below(mean, sd, level):
    """The bell curve's probability of a year below `level` (the normal distribution's CDF)."""
    return 0.5 * (1 + math.erf((level - mean) / (sd * math.sqrt(2))))


rng = random.Random(SEED)
print(f"{YEARS:,} synthetic years for each asset (seed {SEED})")
print(f"{'Asset':<9}{'Average':>9}{'Volatility':>12}{'Worst year':>12}{'Best year':>11}{'Years below 0':>15}")
results = []
for name, mean, sd in ASSETS:
    draws = [rng.gauss(mean, sd) for _ in range(YEARS)]
    below = sum(r < 0 for r in draws) / YEARS
    below_20 = sum(r < -0.20 for r in draws) / YEARS
    results.append((name, mean, sd, below, below_20))
    print(
        f"{name:<9}{pct(statistics.fmean(draws)):>9}{pct(statistics.stdev(draws)):>12}"
        f"{pct(min(draws)):>12}{pct(max(draws)):>11}{pct(below):>15}"
    )

print()
print("The same shares from the bell curve itself, for comparison")
print(f"{'Asset':<9}{'Below 0, drawn':>16}{'Below 0, curve':>16}{'Below −20 %, drawn':>20}{'Below −20 %, curve':>20}")
for name, mean, sd, below, below_20 in results:
    print(
        f"{name:<9}{pct(below, 2):>16}{pct(share_below(mean, sd, 0), 2):>16}"
        f"{pct(below_20, 2):>20}{pct(share_below(mean, sd, -0.20), 2):>20}"
    )

10,000 synthetic years for each asset (seed 7)
Asset      Average  Volatility  Worst year  Best year  Years below 0
Steady       4.0 %       1.0 %       0.4 %      7.7 %          0.0 %
Middle       6.5 %       5.0 %     −14.3 %     25.1 %          9.7 %
Swinging    10.1 %      17.9 %     −56.9 %     70.2 %         28.9 %

The same shares from the bell curve itself, for comparison
Asset      Below 0, drawn  Below 0, curve  Below −20 %, drawn  Below −20 %, curve
Steady             0.00 %          0.00 %              0.00 %              0.00 %
Middle             9.70 %          9.68 %              0.00 %              0.00 %
Swinging          28.91 %         28.93 %              4.54 %              4.78 %

Steady never had a losing year, Middle lost in 9.7 % of its years and Swinging in 28.9 %. Swinging’s worst year took 56.9 % off its value and its best added 70.2 %, around the same 10 % average. The second table sets each measured share beside the bell curve the draws came from: 28.91 % of Swinging’s years fell below zero, and the curve says 28.93 %.

The curve itself is below, set to Swinging’s average of 10 and volatility of 18, with the range covering every return below zero. Lower the standard deviation to 5 and watch the share of losing years fall to 2.3 %, with the average unchanged; the table under the chart shows that 68.3 % of years land within one standard deviation of the average:

Regulators measure risk this way too. Every mutual fund scheme shows a riskometer, and for holdings of gold and silver SEBI’s consolidated circular for mutual funds scores the risk by the yearly volatility of the metal’s price over the past 15 years: under 10 % counts as Moderate, 10 to 15 % as Moderately High, 15 to 20 % as High and above 20 % as Very High.

A fall becomes a permanent loss when you sell low or the borrower fails

A fall from the highest value so far is a drawdowndrawdown: A fall in an investment's value from its highest point so far, as a share of that peak. Undoing a drawdown takes a larger gain than the fall: a 40 % drawdown needs a 66.7 % gain., and undoing it takes a larger gain than the fall itself, because the gain works on a smaller amount:

In words: divide 1 by what is left after the fall, then subtract 1. A 40 % fall leaves 0.6 of the value, and 1 ÷ 0.6 − 1 = 66.7 %. A 50 % fall needs 100 %. Now follow one synthetic 20-year path of Swinging through a fall of 37 %, for an investor who holds and one who sells:

The two investors met the same fall. The holder’s drawdown was a swing: it hurt for three years and then belonged to the past. The seller turned the same drawdown into a permanent loss by selling at the low and keeping the money where it could not recover. The program behind the figure prints the whole path, then repeats the experiment on 10,000 synthetic paths and ends with a loan whose borrower fails:

"""Falls, recoveries and the two ways a fall becomes a permanent loss.

Part 1 is exact arithmetic: the gain needed to get back to where you were after a fall.
Part 2 follows one synthetic 20-year path of the made-up "Swinging" asset (yearly returns drawn from a bell curve
with an assumed average of 10 % and volatility of 18 %, seed 50): one investor holds, the other sells the first time
the value is 25 % or more below its peak and keeps the money at an assumed 4 % a year from then on.
Part 3 repeats that for 10,000 synthetic paths. Part 4 is a loan whose borrower fails, which no waiting can repair.
Everything is synthetic or invented: no real security, fund or index is described, and nothing here is a forecast.
"""

import random
import statistics
from decimal import ROUND_HALF_UP, Decimal

MEAN, SD = 0.10, 0.18
SAFE = 0.04
YEARS = 20
START = 100_000
TRIGGER = 0.25


def inr(amount):
    """Whole rupees with Indian digit grouping: 1234567 -> "₹12,34,567"."""
    value = int(Decimal(amount).quantize(Decimal("1"), rounding=ROUND_HALF_UP))
    sign, digits = ("−" if value < 0 else ""), str(abs(value))
    head, tail = digits[:-3], digits[-3:]
    groups = []
    while len(head) > 2:
        groups.insert(0, head[-2:])
        head = head[:-2]
    if head:
        groups.insert(0, head)
    return sign + "₹" + ",".join(groups + [tail])


def pct(x, places=1):
    value = round(x * 100, places)
    return ("−" if value < 0 else "") + f"{abs(value):.{places}f} %"


def path(seed):
    rng = random.Random(seed)
    return [rng.gauss(MEAN, SD) for _ in range(YEARS)]


def follow(returns):
    """Year-end values for a holder, the deepest fall from a peak, and the seller's year (or None) and end value."""
    values, value, peak, deepest, sold = [START], START, START, 0.0, None
    for year, r in enumerate(returns, 1):
        value *= 1 + r
        values.append(value)
        peak = max(peak, value)
        deepest = max(deepest, 1 - value / peak)
        if sold is None and value <= (1 - TRIGGER) * peak:
            sold = year
    seller = values[sold] * (1 + SAFE) ** (YEARS - sold) if sold else value
    return values, deepest, sold, seller


print("Part 1: the gain that undoes a fall")
for fall in (0.10, 0.20, 0.30, 0.40, 0.50, 0.60):
    print(f"  a fall of {pct(fall, 0):>5} needs a gain of {pct(1 / (1 - fall) - 1)}")

print()
SEED = 50
returns = path(SEED)
values, deepest, sold, seller = follow(returns)
print(f"Part 2: one synthetic path of {inr(START)} (seed {SEED})")
print(f"{'Year':>4}{'Return':>10}{'Holder':>12}{'Below peak':>12}")
peak = START
for year in range(1, YEARS + 1):
    peak = max(peak, values[year])
    print(f"{year:>4}{pct(returns[year - 1]):>10}{inr(values[year]):>12}{pct(1 - values[year] / peak):>12}")
print(f"The seller sells at the end of year {sold}, at {inr(values[sold])}, and earns {pct(SAFE, 0)} a year after that.")
old_peak = max(values[:sold])
back = next(y for y in range(sold, YEARS + 1) if values[y] > old_peak)
print(
    f"The holder is back above the old peak of {inr(old_peak)} at the end of year {back}, with {inr(values[back])};"
    f" the seller then has {inr(values[sold] * (1 + SAFE) ** (back - sold))}."
)
print(f"After {YEARS} years: holder {inr(values[-1])}, seller {inr(seller)}.")

print()
PATHS = 10_000
holders, sellers, deep, holder_lost, seller_ahead, triggered = [], [], [], 0, 0, 0
for seed in range(1, PATHS + 1):
    v, d, s, end = follow(path(seed))
    deep.append(d)
    holder_lost += v[-1] < START
    if s:
        triggered += 1
        holders.append(v[-1])
        sellers.append(end)
        seller_ahead += end > v[-1]
print(f"Part 3: {PATHS:,} synthetic 20-year paths (seeds 1 to {PATHS:,})")
print(f"  median deepest fall from a peak: {pct(statistics.median(deep))}")
print(f"  paths that fell {pct(TRIGGER, 0)} or more below a peak at some point: {pct(triggered / PATHS)}")
print(f"  in those paths, median end value: holder {inr(statistics.median(holders))}, seller {inr(statistics.median(sellers))}")
print(f"  in those paths, the seller ended ahead of the holder in {pct(seller_ahead / triggered)}")
print(f"  paths where even the holder ended below {inr(START)}: {pct(holder_lost / PATHS, 2)}")

print()
print("Part 4: a loan of ₹1,00,000 at 8 % a year for 5 years, whose borrower fails after year 2")
promised = 5 * 8_000 + START
received = 2 * 8_000 + 40_000
print(f"  promised: five interest payments of ₹8,000 and the {inr(START)} back = {inr(promised)}")
print(f"  received: two interest payments, then 40 % of the loan recovered = {inr(received)}")
print(f"  short of the promise by {inr(promised - received)}; against the {inr(START)} lent, {inr(received - START)}")
print("  Waiting does not help: the money the borrower could not repay is gone.")

Part 1: the gain that undoes a fall
  a fall of  10 % needs a gain of 11.1 %
  a fall of  20 % needs a gain of 25.0 %
  a fall of  30 % needs a gain of 42.9 %
  a fall of  40 % needs a gain of 66.7 %
  a fall of  50 % needs a gain of 100.0 %
  a fall of  60 % needs a gain of 150.0 %

Part 2: one synthetic path of ₹1,00,000 (seed 50)
Year    Return      Holder  Below peak
   1    −4.2 %     ₹95,840       4.2 %
   2    10.2 %   ₹1,05,634       0.0 %
   3     1.3 %   ₹1,06,989       0.0 %
   4    −0.2 %   ₹1,06,779       0.2 %
   5   −37.1 %     ₹67,199      37.2 %
   6    18.0 %     ₹79,302      25.9 %
   7    23.5 %     ₹97,961       8.4 %
   8    18.0 %   ₹1,15,617       0.0 %
   9     6.6 %   ₹1,23,261       0.0 %
  10     3.2 %   ₹1,27,204       0.0 %
  11    29.2 %   ₹1,64,337       0.0 %
  12    37.6 %   ₹2,26,139       0.0 %
  13    25.7 %   ₹2,84,168       0.0 %
  14     8.0 %   ₹3,06,884       0.0 %
  15    28.0 %   ₹3,92,944       0.0 %
  16    54.7 %   ₹6,07,975       0.0 %
  17    −1.7 %   ₹5,97,921       1.7 %
  18    31.5 %   ₹7,86,524       0.0 %
  19    −9.9 %   ₹7,08,307       9.9 %
  20    −6.0 %   ₹6,65,968      15.3 %
The seller sells at the end of year 5, at ₹67,199, and earns 4 % a year after that.
The holder is back above the old peak of ₹1,06,989 at the end of year 8, with ₹1,15,617; the seller then has ₹75,589.
After 20 years: holder ₹6,65,968, seller ₹1,21,021.

Part 3: 10,000 synthetic 20-year paths (seeds 1 to 10,000)
  median deepest fall from a peak: 28.5 %
  paths that fell 25 % or more below a peak at some point: 60.7 %
  in those paths, median end value: holder ₹3,72,452, seller ₹2,12,100
  in those paths, the seller ended ahead of the holder in 20.9 %
  paths where even the holder ended below ₹1,00,000: 1.83 %

Part 4: a loan of ₹1,00,000 at 8 % a year for 5 years, whose borrower fails after year 2
  promised: five interest payments of ₹8,000 and the ₹1,00,000 back = ₹1,40,000
  received: two interest payments, then 40 % of the loan recovered = ₹56,000
  short of the promise by ₹84,000; against the ₹1,00,000 lent, −₹44,000
  Waiting does not help: the money the borrower could not repay is gone.

Big falls were common: 60.7 % of the synthetic paths dropped 25 % or more below a peak at some point, and the median path’s deepest fall was 28.5 %. In those paths the holders ended with a median ₹3,72,452 and the sellers with ₹2,12,100, yet the sellers came out ahead in 20.9 % of them, so holding on is no promise. Part 4 shows the other way to lose for good: when a borrower fails and pays back only ₹40,000 of a ₹1,00,000 loan, no amount of waiting returns the rest. That is credit risk, and it is why the issuer matters as much as the price.

Your horizon decides which swings you can wait out

A fall can be waited out only if the money can wait. SEBI’s investor guide on managing risk says it plainly: for short-term investments, avoid risky investments such as equity; for long-term ones, choose investments with the potential to beat inflation; and money needed in the near future should avoid volatile or illiquid investments. The program takes each time horizontime horizon: How long it is until you need the money for a goal. Money needed within about a year must be reachable on the day at a value that cannot fall; a long horizon leaves time to adjust a plan. More in “Setting money goals you can measure” from one year to twenty, counts how many of 10,000 synthetic paths ended below the money put in, and then how many ended buying less than it once prices rise an assumed 5 % a year:

"""How often each synthetic asset ends a holding period below where it started, by length of the period.

The three made-up assets are those of synthetic_years.py: yearly returns drawn from bell curves with assumed averages
of 4 %, 6.5 % and 10 % and assumed volatilities of 1 %, 5 % and 18 %. For each holding period the program draws
10,000 synthetic paths (seed 11) and counts the paths that end below the money put in. It also counts the paths whose
end value buys less than the money put in once prices rise an assumed 5 % a year. Synthetic and invented throughout:
no real security, fund or index is described, and nothing here is a forecast.
"""

import random

ASSETS = (("Steady", 0.04, 0.01), ("Middle", 0.065, 0.05), ("Swinging", 0.10, 0.18))
HORIZONS = (1, 3, 5, 10, 20)
PATHS = 10_000
INFLATION = 0.05
SEED = 11


def pct(x, places=1):
    value = round(x * 100, places)
    return ("−" if value < 0 else "") + f"{abs(value):.{places}f} %"


rng = random.Random(SEED)
growth = {}  # asset -> list of paths, each a list of growth factors after 1..20 years
for name, mean, sd in ASSETS:
    paths = []
    for _ in range(PATHS):
        factor, factors = 1.0, []
        for _ in range(max(HORIZONS)):
            factor *= 1 + rng.gauss(mean, sd)
            factors.append(factor)
        paths.append(factors)
    growth[name] = paths

print(f"Share of {PATHS:,} synthetic paths that end below the money put in (seed {SEED})")
print(f"{'Years held':<11}" + "".join(f"{name:>10}" for name, _, _ in ASSETS) + f"{'Swinging, worst path a year':>30}")
for h in HORIZONS:
    row = f"{h:<11}"
    for name, _, _ in ASSETS:
        row += f"{pct(sum(p[h - 1] < 1 for p in growth[name]) / PATHS):>10}"
    worst = min(p[h - 1] for p in growth["Swinging"]) ** (1 / h) - 1
    print(row + f"{pct(worst):>30}")

print()
print(f"Share of paths whose end value buys less than the money put in, with prices up {pct(INFLATION, 0)} a year")
print(f"{'Years held':<11}" + "".join(f"{name:>10}" for name, _, _ in ASSETS))
for h in HORIZONS:
    prices = (1 + INFLATION) ** h
    print(f"{h:<11}" + "".join(f"{pct(sum(p[h - 1] < prices for p in growth[name]) / PATHS):>10}" for name, _, _ in ASSETS))

Share of 10,000 synthetic paths that end below the money put in (seed 11)
Years held     Steady    Middle  Swinging   Swinging, worst path a year
1               0.0 %     9.3 %    29.4 %                       −58.9 %
3               0.0 %     1.2 %    19.9 %                       −31.3 %
5               0.0 %     0.2 %    14.8 %                       −19.3 %
10              0.0 %     0.0 %     7.1 %                       −13.9 %
20              0.0 %     0.0 %     1.9 %                        −6.9 %

Share of paths whose end value buys less than the money put in, with prices up 5 % a year
Years held     Steady    Middle  Swinging
1              84.0 %    37.7 %    39.7 %
3              96.0 %    30.5 %    35.3 %
5              98.9 %    26.2 %    32.8 %
10            100.0 %    18.9 %    26.8 %
20            100.0 %    10.7 %    19.3 %

Swinging ended below its start in 29.4 % of one-year paths, 14.8 % of five-year paths, 7.1 % of ten-year paths and 1.9 % of twenty-year paths: fewer as the horizon grows, never none, and its worst five-year path still lost 19.3 % a year. Steady never fell in rupees, yet with prices rising 5 % a year it bought less than at the start in 84.0 % of one-year paths and in every twenty-year path. Short money fears the swing; long money fears the slow loss to prices.

The table describes asset classes by their traits, never products: within a year it leaves only cash, deposits and debt that is repaid in time; beyond five years it warns that cash loses to prices. Need for cash at short notice (liquidityliquidity: How quickly and cheaply something can be turned into cash without losing value. A savings balance is very liquid; money fixed for a term, or a house, is much less so. More in “Setting money goals you can measure”) rules out real estate at any horizon.

A promise of high, guaranteed returns is a warning sign

Everything above ties return to risk: Swinging’s higher average came with losing years, deep falls and paths that ended below their start. An offer that promises a high return and no risk at once claims what no asset class here can give. SEBI’s investor guide on frauds lists the red flags: promises of quick and high returns, guaranteed returns, pressure to decide at once, unsolicited offers by phone, e-mail or social media, a strategy nobody will explain, and promoters who are not registered with a regulator. SEBI’s Investor Survey 2025, which covered over 90,000 households, found fear of losses among the main reasons people stay away from investing, and a scheme that promises safety and high returns together speaks to exactly that fear.

The arithmetic of an “assured” monthly return gives it away. The program compounds three such promises and counts how many of 10,000 synthetic five-year paths of Swinging, the most volatile asset of this lesson, get as far:

"""What "assured" monthly returns add up to, set against the synthetic Swinging asset.

The monthly promises are invented examples of the kind SEBI's investor guide warns about. Swinging is the made-up asset
of synthetic_years.py: yearly returns drawn from a bell curve with an assumed average of 10 % and an assumed volatility
of 18 %. The program counts how many of 10,000 synthetic five-year paths (seed 5) reach what each promise claims.
Nothing here describes a real investment, and nothing is a forecast.
"""

import random
from decimal import ROUND_HALF_UP, Decimal

START = 100_000
YEARS = 5
PATHS = 10_000
SEED = 5


def inr(amount):
    value = int(Decimal(amount).quantize(Decimal("1"), rounding=ROUND_HALF_UP))
    digits = str(value)
    head, tail = digits[:-3], digits[-3:]
    groups = []
    while len(head) > 2:
        groups.insert(0, head[-2:])
        head = head[:-2]
    if head:
        groups.insert(0, head)
    return "₹" + ",".join(groups + [tail])


rng = random.Random(SEED)
ends, losing = [], []
for _ in range(PATHS):
    value, down = START, 0
    for _ in range(YEARS):
        r = rng.gauss(0.10, 0.18)
        value *= 1 + r
        down += r < 0
    ends.append(value)
    losing.append(down)

print(f"{'Promise':<16}{'A year':>9}{'₹1,00,000 in 5 years':>23}{'Swinging paths that get there':>32}")
for monthly in ("0.02", "0.03", "0.05"):
    m = Decimal(monthly)
    yearly = (1 + m) ** 12 - 1
    target = START * (1 + m) ** (12 * YEARS)
    reached = sum(e >= target for e in ends) / PATHS
    label = f"{m * 100:.0f} % a month"
    print(f"{label:<16}{float(yearly) * 100:>7.1f} %{inr(target):>23}{reached * 100:>30.2f} %")
best = max(range(PATHS), key=lambda i: ends[i])
print()
print(f"The best of the {PATHS:,} synthetic paths turned {inr(START)} into {inr(ends[best])},")
down = "no losing year" if losing[best] == 0 else f"{losing[best]} losing years"
print(f"{((ends[best] / START) ** (1 / YEARS) - 1) * 100:.1f} % a year, with {down} among its five.")
print(f"Paths with at least one losing year: {sum(d > 0 for d in losing) / PATHS * 100:.1f} %.")

Promise            A year   ₹1,00,000 in 5 years   Swinging paths that get there
2 % a month        26.8 %              ₹3,28,103                          1.32 %
3 % a month        42.6 %              ₹5,89,160                          0.00 %
5 % a month        79.6 %             ₹18,67,919                          0.00 %

The best of the 10,000 synthetic paths turned ₹1,00,000 into ₹5,46,184,
40.4 % a year, with no losing year among its five.
Paths with at least one losing year: 82.1 %.

“3 % a month” is 42.6 % a year and would turn ₹1,00,000 into ₹5,89,160 in five years. Not one synthetic path got there, and the best of them, ₹5,46,184, needed five years without a single loss, while 82.1 % of the paths had at least one losing year. The table reads four common pitches the same way:

The last row has a regulatory edge. SEBI’s frequently asked questions on education and advice say that a person engaged solely in education does not use the market price data of the past thirty days to name a security or hint at its future price, which is why this track uses only synthetic and invented assets.

Practise: risks, swings and horizons

The questions run from naming the risk behind an invented event to matching a goal with its asset classes. The first comes from the idea you have found hardest, and each numeric one says how many decimals it wants:

Interview questions

Warm-up (fresher): What is the difference between volatility and a permanent loss?

Volatility is how far an investment’s returns swing around their average, measured as the standard deviation of its yearly returns. A synthetic asset that averages 10 % a year with a volatility of 18 % ends about 29 % of its years below zero. A swing is a loss only on paper while you still own something that can recover: in one synthetic path, a 37 % fall was back above its old peak three years later. A loss becomes permanent in two ways. You sell after the fall and keep the money where it cannot recover, or the company or borrower fails, as when a loan of ₹1,00,000 returns only ₹40,000. So the questions to ask are how long you can wait, and what can fail.

Warm-up (fresher): Why is a promise of guaranteed high returns a warning sign?

Because return and risk come together: an asset earns more on average only by accepting losing years and deep falls. A promise of a high return with no risk claims what no asset class can give. SEBI’s investor guide lists guaranteed returns, quick and high returns, pressure to decide at once, unsolicited offers and unregistered promoters among the red flags of fraud. The arithmetic helps too: an “assured 3 % a month” is 42.6 % a year once it compounds. The first check is the regulator’s register: anyone advising on securities must be registered with SEBI.

Key takeaways

• Each asset class pays in its own way and carries its own risk: equity swings and can fail, debt carries credit and interest-rate risk, gold swings with no income, real estate is hard to sell, and cash loses to rising prices.
• Volatility is the standard deviation of returns: at the same 10 % average, a volatility of 18 % puts 28.9 % of years below zero, and 5 % only 2.3 %.
• A fall needs a larger gain to undo, 66.7 % after a 40 % fall, and it becomes permanent only when you sell low or the borrower or company fails.
• The horizon decides which swings you can wait out: money needed within a year cannot ride a fall, and money left for decades faces the slow loss to rising prices.
• A promise of high returns with no risk is a warning sign, and SEBI lists guaranteed returns and pressure to decide at once among the red flags of fraud.