5D Library › Macro-Liquidity Regime Switching, Extreme Tail Dependence and Barbell Asset Allocation · 中文

Chapter 5 Barbell Allocation and Stochastic Dominance

The author’s question

The author’s framework proposes a “barbell” allocation: 90 % of the money in a cushion of safe assets such as short-term government bills, and 10 % in options with convex payoffs. Its return distribution has two attractive features: the floor is protected (the options can lose at most that 10 %), and the upper tail is convex (in a big rally the options’ returns multiply). The author hoped to prove that this allocation “stochastically dominates”.

This chapter’s conclusion is that, in general, the claim does not hold. When the options expire worthless more often than the stock falls below the floor (as in this chapter’s model), the barbell cannot first-order dominate a stock; second-order dominance depends on the price of the options; in the book’s model, under four option prices, it dominates neither a stock nor a simple “90 % bills + 10 % stock” portfolio, and when options are rich it is itself second-order dominated by the latter. The definitions come first, then the proof and the numbers.

Source text

In Antifragile, Taleb describes the idea: if you put 90 percent of your funds in “boring” cash (assuming, he adds, protection from inflation) and 10 percent in very risky, maximally risky securities, “you cannot possibly lose more than 10 percent, while you are exposed to massive upside” (Taleb 2012).

The book cites Taleb’s barbell only as the source of the idea. One correction is needed at once: “you cannot lose more than 10 %” holds only in nominal terms and only if the cash or bills do not default. In this chapter’s example the bills pay 3 % a year, so the 90 % in bills grows to 92.7 % by the year end and the nominal maximum loss is 7.3 %; but the loss of purchasing power after inflation can be larger, and the floor holds only at the end of the holding period.

Definitions of stochastic dominance

Evidence

Let FAF_A and FBF_B be the cumulative distribution functions of two returns. First-order stochastic dominance (FSD): AA first-order dominates BB if and only if FA(x)≤FB(x)F_A(x) \le F_B(x) for every xx (strictly somewhere); equivalently, every investor who prefers more wealth to less (increasing utility) likes AA at least as much. Second-order stochastic dominance (SSD): AA second-order dominates BB if and only if ∫−∞xFA(t)dt≤∫−∞xFB(t)dt\int_{-\infty}^{x} F_A(t)\,dt \le \int_{-\infty}^{x} F_B(t)\,dt for every xx; equivalently, every risk-averse investor (increasing, concave utility) likes AA at least as much (Hadar and Russell 1969; Levy 1992).

Two necessary conditions follow at once. First, both first- and second-order dominance require E[A]≥E[B]E[A] \ge E[B]: take the utility u(x)=xu(x) = x (increasing and concave), and the dominating prospect cannot have a lower expected return. Second, if BB has some probability of falling below the lowest possible value of AA, BB cannot dominate AA in either order: on that stretch FB>0=FAF_B > 0 = F_A, and the integral is larger for BB too.

Proposition and proof

Proposition (a corrected version of the framework’s claim). Let the barbell’s return have a floor mm (its return when all the options expire worthless), equal to mm with positive probability π\pi; let the stock’s return be continuously distributed with P(RS<m)>0P(R_S < m) > 0. Then:

  1. the stock cannot first- or second-order dominate the barbell;
  2. if π>FS(m)\pi > F_S(m) (the options expire worthless more often than the stock falls below the floor), the barbell cannot first-order dominate the stock;
  3. if E[Rbarbell]<E[RS]E[R_{\text{barbell}}] < E[R_S], the barbell cannot second-order dominate the stock either.

Proof. (1) For x<mx < m, Fbarbell(x)=0<FS(x)F_{\text{barbell}}(x) = 0 < F_S(x) (at least near mm), so FS≤FbarbellF_S \le F_{\text{barbell}} cannot hold everywhere, and likewise for the integrals. (2) At x=mx = m, Fbarbell(m)≥π>FS(m)F_{\text{barbell}}(m) \ge \pi > F_S(m), which violates first-order dominance. (3) Second-order dominance implies an expected return that is no lower, contradicting the assumption. ∎

The condition in (2) cannot be dropped: if the options rarely expire worthless while the stock often falls below the floor, the proposition does not rule out first-order dominance of the stock by the barbell (although under arbitrage-free pricing that would be most unusual). In this chapter’s model the options expire worthless with probability 56.4 % and the stock falls below −7.3 % with probability 20.5 % (computed separately from the same simulated years as the program below), so the condition holds.

Under the condition of (3), the two portfolios do not dominate each other: some risk-averse investors prefer the barbell and some the stock, depending on the exact shape of their utility. “The barbell stochastically dominates” can therefore be discussed at most as second-order dominance under the condition that “options are cheap enough for the barbell’s expected return to be no lower than the comparison’s”; first-order dominance is all but impossible when there is a floor and a positive chance of the options expiring worthless.

The price of options: the variance risk premium

The barbell’s expected return depends on whether the options are bought dear or cheap. Book 38 discussed how the implied volatility of index options has historically averaged more than the volatility subsequently realised, the variance risk premium, so that option buyers bear negative excess returns over the long run (Carr and Wu 2009); here we only recap that conclusion and do not repeat its programs. If options are dear on average, the 10 % in options is “paying an insurance premium” every year, dragging down the barbell’s long-run expected return.

The program below compares several portfolios using the author’s model (illustrative parameters): one-year stock returns follow Merton’s jump-diffusion (Merton 1976), with diffusion volatility 0.15, on average 0.5 jumps a year whose log size is normal with mean −0.08 and standard deviation 0.10, and an expected return of 8 % (total volatility 0.175); bills pay 3 %; the options are one-year calls struck 10 % above the current price, bought at the Black–Scholes price for a given implied volatility. Four implied volatilities are used: 0.13 (cheap), 0.175 (equal to the stock’s total volatility), 0.22 (rich) and 0.28 (very rich).

Calculation

# Does a barbell (90 % one-year bills + 10 % one-year calls) stochastically dominate?
# ILLUSTRATIVE 1-year model: bills 3 %; stock = Merton jump-diffusion (diffusion sd 0.15,
# 0.5 jumps a year with log size N(-0.08, 0.10)), expected return 8 %. Calls struck 10 %
# out of the money, bought at the Black-Scholes price with an implied volatility that is
# rich, equal or cheap relative to the stock's total volatility. 200,000 simulated years.
# First-order dominance: F_A <= F_B everywhere; second-order: integral of F_A <= of F_B.
import math, random, statistics

R, SIG, LAM, JM, JS, MU = 0.03, 0.15, 0.5, -0.08, 0.10, 0.08
K = 1.10
TOT = math.sqrt(SIG ** 2 + LAM * (JM ** 2 + JS ** 2))
KAP = math.exp(JM + JS ** 2 / 2) - 1

def stock_draw(rng):
    n, t = 0, rng.expovariate(LAM)
    while t < 1:
        n += 1; t += rng.expovariate(LAM)
    lr = (math.log(1 + MU) - LAM * KAP - SIG ** 2 / 2 + SIG * rng.gauss(0, 1)
          + sum(rng.gauss(JM, JS) for _ in range(n)))
    return math.exp(lr)

def bs_call(s, k, vol, r=math.log(1 + R), t=1.0):
    d1 = (math.log(s / k) + (r + vol * vol / 2) * t) / (vol * math.sqrt(t))
    N = statistics.NormalDist().cdf
    return s * N(d1) - k * math.exp(-r * t) * N(d1 - vol * math.sqrt(t))

def cdfs(sample, grid):
    s = sorted(sample); n = len(s); out, j = [], 0
    for x in grid:
        while j < n and s[j] <= x:
            j += 1
        out.append(j / n)
    return out

def dominance(a, b, grid, dx):
    Fa, Fb = cdfs(a, grid), cdfs(b, grid)
    fsd = all(x <= y + 0.002 for x, y in zip(Fa, Fb))
    Ia = Ib = 0.0; ssd = True
    for x, y in zip(Fa, Fb):
        Ia += x * dx; Ib += y * dx
        ssd = ssd and Ia <= Ib + 0.0005
    return "FSD" if fsd else ("SSD" if ssd else "none")

if __name__ == "__main__":
    rng = random.Random(1992)
    S = [stock_draw(rng) for _ in range(200_000)]
    grid = [-0.9 + 0.002 * i for i in range(1451)]; dx = 0.002
    stock = [s - 1 for s in S]; linear = [0.9 * (1 + R) + 0.1 * s - 1 for s in S]
    print(f"stock: total volatility {TOT:.3f}; simulated mean return"
          f" {statistics.fmean(stock):+.3f};")
    print(f"calls expire worthless in {sum(s <= K for s in S) / len(S):.1%} of years")
    rows = [("stock", stock), ("90 % bills + 10 % stock", linear)]
    for lab, vol in (("cheap, iv 0.13", 0.13), ("equal, iv 0.175", 0.175),
                     ("rich, iv 0.22", 0.22), ("very rich, iv 0.28", 0.28)):
        c = bs_call(1.0, K, vol)
        calls = [max(s - K, 0) / c - 1 for s in S]
        bar = [0.9 * (1 + R) + 0.1 * (1 + x) - 1 for x in calls]
        print(f"calls {lab:18s}: price {c:.4f}; mean return of the call sleeve"
              f" {statistics.fmean(calls):+7.1%}")
        rows.append((f"barbell, {lab}", bar))
    print("portfolio                       mean     sd   P(<-10%)  1% quantile  P(>+30%)")
    for lab, x in rows:
        s = sorted(x)
        print(f"{lab:30s} {statistics.fmean(x):+.3f}  {statistics.pstdev(x):.3f}"
              f"   {sum(v < -0.10 for v in x) / len(x):6.1%}     {s[len(s) // 100]:+.3f}"
              f"     {sum(v > 0.30 for v in x) / len(x):6.1%}")
    print("dominance (row over column):   stock     90/10 linear  bills (+3 %)")
    bills = [R] * len(stock)
    for lab, x in rows[2:]:
        dm = [dominance(x, y, grid, dx) for y in (stock, linear, bills)]
        print(f"{lab:30s} {dm[0]:8s}  {dm[1]:8s}      {dm[2]}")
    print("over the four barbells (cheap, equal, rich, very rich):")
    for lab, x in rows[:2] + [("bills (+3 %)", bills)]:
        dm = [dominance(x, y, grid, dx) for _, y in rows[2:]]
        print((f"{lab:30s} " + "  ".join(f"{v:5s}" for v in dm)).rstrip())
# stock: total volatility 0.175; simulated mean return +0.080;
# calls expire worthless in 56.4% of years
# calls cheap, iv 0.13    : price 0.0263; mean return of the call sleeve +144.4%
# calls equal, iv 0.175   : price 0.0431; mean return of the call sleeve  +49.1%
# calls rich, iv 0.22     : price 0.0606; mean return of the call sleeve   +6.1%
# calls very rich, iv 0.28: price 0.0843; mean return of the call sleeve  -23.7%
# portfolio                       mean     sd   P(<-10%)  1% quantile  P(>+30%)
# stock                          +0.080  0.187    16.3%     -0.323      11.8%
# 90 % bills + 10 % stock        +0.035  0.019     0.0%     -0.005       0.0%
# barbell, cheap, iv 0.13        +0.171  0.414     0.0%     -0.073      24.8%
# barbell, equal, iv 0.175       +0.076  0.253     0.0%     -0.073      16.0%
# barbell, rich, iv 0.22         +0.033  0.180     0.0%     -0.073       9.5%
# barbell, very rich, iv 0.28    +0.003  0.129     0.0%     -0.073       4.4%
# dominance (row over column):   stock     90/10 linear  bills (+3 %)
# barbell, cheap, iv 0.13        none      none          none
# barbell, equal, iv 0.175       none      none          none
# barbell, rich, iv 0.22         none      none          none
# barbell, very rich, iv 0.28    none      none          none
# over the four barbells (cheap, equal, rich, very rich):
# stock                          none   none   none   none
# 90 % bills + 10 % stock        none   none   SSD    SSD
# bills (+3 %)                   none   none   none   SSD

The calls expire worthless in 56.4 % of years, so the barbell has a 56.4 % chance of losing exactly 7.3 %. The mean return of the option sleeve swings with the price: +144 % when cheap, +49 % at equal volatility, +6.1 % when rich and −23.7 % when very rich. Note that even when implied volatility equals the stock’s total volatility the calls have a positive expected return, because in this model the stock carries a 5 % risk premium and a call is in effect leveraged stock; only when implied volatility is clearly above actual volatility does the calls’ expected return turn negative.

The barbell’s expected return falls from +17.1 % (cheap options) to +0.3 % (very rich options), while its 1 % quantile stays at −7.3 % throughout and its probability of a big gain (above 30 %) falls from 24.8 % to 4.4 %. The stock’s expected return is 8.0 % with a 1 % quantile of −32.3 %; “90 % bills + 10 % stock” has an expected return of 3.5 % and a 1 % quantile of −0.5 %.

The dominance checks (comparing distribution functions and their integrals on a grid of 0.002) show that under all four prices the barbell neither first- nor second-order dominates the stock, the 90/10 linear portfolio or bills. In the other direction, the stock dominates no barbell; but “90 % bills + 10 % stock” second-order dominates the barbell when options are rich or very rich (0.22 and 0.28), and bills alone second-order dominate the barbell when options are very rich. In a market where options are dear, then, every risk-averse investor would prefer the simple 90/10 linear portfolio to the barbell. Figure 4 plots the distribution functions and their integrals.

Left: cumulative distribution functions of one-year returns; right: the distribution function integrated from the left up to x. Solid (blue): stock; dotted: 90 % bills + 10 % stock; solid (purple): barbell with calls bought at implied volatility 0.22; dashed: barbell at implied volatility 0.13 (the author’s model, illustrative parameters). The barbell’s distribution function jumps to 0.56 at −7.3 %.

In the left panel of Figure 4 the barbell’s distribution function jumps vertically to 0.56 at −7.3 %, crossing those of the stock and the 90/10 portfolio: that is the geometry of the proposition. In the right panel the integral of the 90/10 portfolio stays below that of the rich-option barbell throughout (second-order dominance), while that of the cheap-option barbell crosses the 90/10 curve (neither dominates).

Note

What the barbell really offers. The barbell provides a nominal floor plus a right-skewed, convex upper tail. That appeals to some investors: institutions with hard loss limits, for example, or investors who like positive skewness (a preference for skewness corresponds to higher-order conditions such as third-order stochastic dominance (Levy 1992)). But that is a matter of preference, not a mathematical result that it is “better for every investor”. The price is that the options expire worthless in most years, the long-run return depends on what the options cost, and historical evidence suggests that options are dear on average.

Comparison: CPPI and capital-protected portfolios

The academic relative closest to the barbell is “portfolio insurance”. Constant proportion portfolio insurance (CPPI) was analysed systematically by Black and Perold (Black and Perold 1992). Let the portfolio value be VtV_t and the floor FtF_t (for example 92.7 % at maturity, discounted to today at the risk-free rate); the cushion is Ct=Vt−FtC_t = V_t – F_t, the amount invested in the risky asset is a multiplier mm times the cushion, Et=mCtE_t = m\,C_t, the rest is held in bills, and the portfolio is rebalanced periodically. In the ideal case of continuous prices and continuous rebalancing, the portfolio value never falls below the floor; but if the risky asset drops by more than 1/m1/m between rebalancing dates, the cushion is breached. The barbell is static and buys its floor with options; CPPI is dynamic and replicates the floor with a trading rule.

Calculation

# CPPI (Black & Perold 1992) against the static barbell on the SAME simulated year.
# Floor = 0.927 at year end (as for 90 % bills at 3 %), discounted back at 3 %. Exposure =
# m x (value - floor), capped at the value (no borrowing), rebalanced daily or monthly.
# ILLUSTRATIVE stock as in p10 (diffusion sd 0.15, 0.5 jumps a year of log size
# N(-0.08, 0.10), mean return 8 %); barbell calls bought at implied vol 0.22.
# 10,000 simulated years.
import math, random, statistics

R, SIG, LAM, JM, JS, MU, K, D = 0.03, 0.15, 0.5, -0.08, 0.10, 0.08, 1.10, 252
KAP = math.exp(JM + JS ** 2 / 2) - 1
RD = math.log(1 + R) / D

def path(rng):
    mu = (math.log(1 + MU) - LAM * KAP - SIG ** 2 / 2) / D; s, out = 1.0, []
    for _ in range(D):
        lr = mu + SIG / math.sqrt(D) * rng.gauss(0, 1)
        if rng.random() < LAM / D:
            lr += rng.gauss(JM, JS)
        s *= math.exp(lr); out.append(s)
    return out

def cppi(sp, m, every):
    v, prev = 1.0, 1.0; risky = 0.0
    for t, s in enumerate(sp):
        if t % every == 0:
            floor = 0.927 * math.exp(-RD * (D - t))
            risky = min(max(m * (v - floor), 0.0), v)
        safe = v - risky
        risky *= s / prev; safe *= math.exp(RD); v = risky + safe; prev = s
    return v, risky / v

def bs_call(vol, k=K, r=math.log(1 + R)):
    N = statistics.NormalDist().cdf
    d1 = (-math.log(k) + r + vol * vol / 2) / vol
    return N(d1) - k * math.exp(-r) * N(d1 - vol)

if __name__ == "__main__":
    rng = random.Random(1992); c = bs_call(0.22)
    cfg = [(3, 1), (3, 21), (6, 1), (6, 21), (10, 21)]
    res = {k: [] for k in cfg}; bar, stk = [], []
    for _ in range(10_000):
        sp = path(rng); stk.append(sp[-1] - 1)
        bar.append(0.927 + 0.1 * max(sp[-1] - K, 0) / c - 1)
        for m, e in cfg:
            res[(m, e)].append(cppi(sp, m, e))
    print("strategy                 mean   median  below floor   worst   ending with")
    print("                                                             < 5 % in stock")
    for (m, e), r in res.items():
        v = [a - 1 for a, b in r]
        below = sum(a < 0.927 - 1e-9 for a, b in r) / len(r)
        lock = sum(b < 0.05 for a, b in r) / len(r)
        lab = f"CPPI m = {m:2d}, {'daily' if e == 1 else 'monthly'}"
        print(f"{lab:23s}  {statistics.fmean(v):+.3f}   {statistics.median(v):+.3f}"
              f"    {below:6.2%}     {min(v):+.3f}     {lock:6.1%}")
    print(f"{'barbell, calls iv 0.22':23s}  {statistics.fmean(bar):+.3f}   "
          f"{statistics.median(bar):+.3f}    {0:6.2%}     {min(bar):+.3f}")
    print(f"{'stock':23s}  {statistics.fmean(stk):+.3f}   {statistics.median(stk):+.3f}"
          f"       --       {min(stk):+.3f}")
    a = [x - 1 for x, y in res[(6, 1)]]
    win = sum(x > y for x, y in zip(a, bar)) / len(a)
    print(f"CPPI m = 6 daily beats the barbell in {win:.1%} of years;"
          f" correlation {statistics.correlation(a, bar):.2f}")
# strategy                 mean   median  below floor   worst   ending with
#                                                              < 5 % in stock
# CPPI m =  3, daily       +0.045   +0.034     0.05%     -0.075       0.7%
# CPPI m =  3, monthly     +0.045   +0.035     0.08%     -0.096       0.9%
# CPPI m =  6, daily       +0.059   +0.017     6.38%     -0.255       9.2%
# CPPI m =  6, monthly     +0.059   +0.018     7.02%     -0.252       9.1%
# CPPI m = 10, monthly     +0.066   +0.014    15.83%     -0.345      20.3%
# barbell, calls iv 0.22   +0.031   -0.073     0.00%     -0.073
# stock                    +0.079   +0.072       --       -0.658
# CPPI m = 6 daily beats the barbell in 77.3% of years; correlation 0.96

The program simulates ten thousand years day by day with the same stock model as before (including jumps). With multiplier 3, breaching the floor requires a one-day fall of more than 33 %, and only 0.05 % (daily rebalancing) and 0.08 % (monthly) of years end below the floor. With multiplier 6 the threshold drops to about 17 %, the share of years ending below the floor rises to 6.38 % and 7.02 %, and the worst year loses 25.5 %; with multiplier 10 the threshold is 10 %, 15.83 % of years end below the floor and the worst loses 34.5 %. The last column shows another risk of CPPI: after a large fall the cushion is close to zero and the portfolio sits almost entirely in bills (“cash lock”); with multiplier 6, 9.2 % of years end with less than 5 % in the risky asset, unable to take part even if the market then recovers. The barbell (calls bought at 0.22) never ends below its floor, but its expected return is only 3.1 % and its median −7.3 %. In this model, CPPI with multiplier 6 beats the barbell in 77.3 % of years, and the correlation between the two returns is 0.96. Each kind of insurance has its price: for the barbell, a certain option premium; for CPPI, the risk of a gap through the floor and path dependence.

Chapter summary

A barbell of “90 % bills + 10 % options” has a nominal floor and a convex upper tail, but “stochastic dominance” does not hold in general. The book proves that a barbell with a floor cannot be first- or second-order dominated by a stock, and that when its options expire worthless more often than the stock falls below the floor (56.4 % against 20.5 % in this chapter’s model) it cannot first-order dominate the stock either; second-order dominance requires an expected return that is no lower, which depends on option prices. In the author’s jump-diffusion model, under four option prices, the barbell shows neither first- nor second-order dominance over the stock or the 90/10 linear portfolio; when options are rich, it is itself second-order dominated by the 90/10 portfolio. Options that are dear over the long run (the variance risk premium) mean the barbell’s option sleeve is paying an insurance premium in most years. CPPI is a dynamic way to protect a floor; the larger the multiplier, the greater the risk that a gap breaches the floor.

References

Black, Fischer, and André F. Perold. 1992. “Theory of Constant Proportion Portfolio Insurance.” Journal of Economic Dynamics and Control 16 (3–4): 403–26. https://doi.org/10.1016/0165-1889(92)90043-E.
Carr, Peter, and Liuren Wu. 2009. “Variance Risk Premiums.” The Review of Financial Studies 22 (3): 1311–41. https://doi.org/10.1093/rfs/hhn038.
Hadar, Josef, and William R. Russell. 1969. “Rules for Ordering Uncertain Prospects.” The American Economic Review 59 (1): 25–34.
Levy, Haim. 1992. “Stochastic Dominance and Expected Utility: Survey and Analysis.” Management Science 38 (4): 555–93. https://doi.org/10.1287/mnsc.38.4.555.
Merton, Robert C. 1976. “Option Pricing When Underlying Stock Returns Are Discontinuous.” Journal of Financial Economics 3 (1–2): 125–44. https://doi.org/10.1016/0304-405X(76)90022-2.
Taleb, Nassim Nicholas. 2012. Antifragile: Things That Gain from Disorder. Random House.

Fan, Y. (2026). Macro-Liquidity Regime Switching, Extreme Tail Dependence and Barbell Asset Allocation: Vine Copulas, Hidden Markov Models and Sovereign Buffer Dynamics [宏观流动性体制转换、极值尾部相依与杠铃资产配置] (Version 1.0.0). 5D Library. https://doi.org/10.5281/zenodo.23828049 · Licensed under Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International (CC BY-NC-ND 4.0).