Chapter 13 · Hedging Engineering

对冲的工程学：尾部保险的成本核算Hedging Engineering: Costing Out Tail Protection

第 2 章说 Protective Put 是“特定时期的战术工具，不是常态配置”，并留下一个预算问题：连续买保险可能需要多大的毛支出？本章先做一个明确受限的情景计算。在固定标的、30 天到期、VIX≈20 与指定偏斜等静态输入下，单月模型保费乘以 12 得到 8.8% 的年化毛支出情景——它与标普 500 四十年 11.2% 的年化总回报处在同一个量级，但不是实际滚动策略的年成本。 Chapter 2 called the protective put “a tactical tool for specific windows, not a permanent allocation,” and left a budgeting question: how large might the gross outlay for continuous insurance be? This chapter starts with an explicitly bounded scenario. With spot fixed, 30-day maturity, VIX≈20, and the stated skew inputs, one model month's premium × 12 produces an 8.8% annualized gross-outlay scenario — the same order of magnitude as the S&P 500's 11.2% annualized total return over forty years, but not the annual cost of an actual rolling strategy.

13.1 先把账单摆出来13.1 The Bill, Up Front

最简单的持续对冲方案：每月买入一张 5% 价外的 Put，到期后再买下一张，永不间断。用本书的引擎按不同的市场波动水平定价（偏斜使价外 Put 的 IV 高于平值，此处统一加 3 个百分点）： The simplest continuous hedge: buy a 5% out-of-the-money put every month, roll it at expiry, never stop. Priced with this book's engine across volatility regimes (skew lifts an OTM put's IV above the at-the-money level, so 3 points are added throughout):

平静 VIX≈15Calm, VIX≈15: 年化 4.5%4.5%/yr
常态 VIX≈20Normal, VIX≈20: 年化 8.8%8.8%/yr
紧张 VIX≈30Tense, VIX≈30: 年化 19.3%19.3%/yr
恐慌 VIX≈45Panic, VIX≈45: 年化 37.2%37.2%/yr

对冲年成本 vs 波动率水平annual hedging cost vs. volatility level

先说清楚这张表算的是什么：它是毛保费支出——你每年为这些 Put 付出去的钱，还没有减去它们的赔付。在指数单月跌破行权价时，Put 的赔付会抵消部分乃至超过累计保费，因此净拖累可能远低于毛支出。本章没有运行逐月、含展期与赔付的历史回测，所以不能给出净拖累的数值或方向。年度收益为正或负也不能推断月度尾部赔付；要回答这个问题，必须使用逐月路径、实际行权价、当时 IV、成交成本和赔付。 First, be clear what that table measures: it is gross premium outlay — what you pay each year for the puts, before subtracting what they pay back. When the index finishes a month below the strike, payouts offset part of the premium and can even exceed cumulative outlay, so net drag may be far below gross spending. This chapter does not run a monthly historical backtest with rolls and payouts, so it cannot state the amount or even the sign of net drag. Positive or negative annual returns do not identify monthly tail payouts; answering that question requires the monthly path, actual strikes, contemporaneous IV, execution costs, and payouts.

所以正确的说法不是“11.2% 减 8.8%”这样一道减法，而是：在这组静态模型输入下，8.8% 是值得进入预算压力测试的毛支出量级。它与标普长期 11.2% 年化总回报的并列只能提醒你成本可能很大，不能推出对冲后的期望收益、净拖累或风险调整结果。这些结论必须由完整净损益回测支持，而本章没有提供这样的回测。 So the correct statement is not the subtraction “11.2 minus 8.8.” It is this: under these static model inputs, 8.8% is a gross-outlay magnitude that belongs in a budget stress test. Placing it beside the S&P's 11.2% long-run annualized total return only signals that the spending may be material; it does not establish hedged expected return, net drag, or risk-adjusted performance. Those claims require a complete net-P/L backtest, which this chapter does not provide.

这张表的计算口径How the table is computed

每月保费由本书引擎按 Black-Scholes 计算（标的指数化为 100、30 天到期、利率 4%、价外 Put 的 IV 在平值基础上加 3 个百分点以反映偏斜），年化情景值 = 单月保费 × 12。这一步隐含两个简化：① 假设十二次买入都发生在标的价格不变（100）的水平上，真实市场里标的会移动、每次的 5% 价外行权价也随之改变；② 只计保费支出、不计赔付，所以它描述的是这组固定输入下的毛支出情景，不是净成本。它也不是实际毛支出的上界：真实 IV、偏斜、路径和成交价格都可能令全年保费高于或低于这个静态数值。这些数字只用于展示量级及其随 IV 变化的形状，不是任何具体对冲方案的回测结果。 Each month's premium is a Black-Scholes value from this book's engine (index normalised to 100, 30 days, 4% rate, with 3 points added to the OTM put's IV for skew), and the annualized scenario is one month's premium × 12. That step embeds two simplifications: ① it assumes all twelve purchases occur with the underlying unchanged at 100, whereas in a real market the index moves and each 5%-OTM strike moves with it; ② it counts premium outflows only, never payouts, so it is a gross-outlay scenario under fixed inputs, not a net-cost estimate. Nor is it an upper bound on actual gross spending: realized IV, skew, path, and execution prices can make annual premiums higher or lower than this static figure. The numbers show magnitude and sensitivity to IV; they are not a backtest of any specific hedging programme.

而且它在你最需要的时候最贵And it costs most exactly when you want it most

注意这张情景表的形状：在其他输入不变时，Put 毛保费随 IV 上升——从 IV≈15 情景的 4.5% 到 IV≈45 情景的 37.2%，相差八倍多。这不是说压力中“无法”买保险，而是说当压力已被 IV 定价后，同样规则的保护可能需要完全不同的预算。第 3 章的 2024 年 8 月示意也展示了时点对保费的敏感性。是否此时仍值得买，取决于剩余风险、价格与约束，不能只凭“已经出事”决定。 Note the shape of the scenario table: holding other inputs fixed, gross put premium rises with IV — from 4.5% in the IV≈15 case to 37.2% in the IV≈45 case, more than eightfold. That does not make insurance “unavailable” during stress; it means once stress is priced into IV, the same protection rule may require a completely different budget. Chapter 3's August 2024 illustration also shows premium sensitivity to timing. Whether buying then is still worthwhile depends on residual risk, price, and constraints, not on “trouble has started” alone.

13.2 三个把账单压下来的杠杆13.2 Three Levers That Lower the Bill

如果这组输入下的 8.8% 毛支出超过预算，工程问题就变成了：怎么在保留有意义保护的前提下，把毛支出压到可承受的范围。下面是三个常见杠杆；每个都要用完整净损益验证。 If the 8.8% gross outlay under these inputs exceeds the budget, the engineering question becomes: how can gross spending be reduced while retaining meaningful protection? Three common levers follow, and each needs validation on complete net P/L.

① 买得更远① Further OTM: 便宜得多，但只保尾部Far cheaper, tail only
② 卖掉一部分② Sell some back: 价差 / 领口（第 4、10 章）Spreads / collars (Ch.4, 10)
③ 有条件部署③ Conditional deployment: IV 分位可作候选输入，非已证实信号IV rank may be a candidate input, not a proven signal

第 ① 条是尾部对冲的核心思路：不再保护“下跌 5% 以后的全部损失”，而只保护“下跌 20% 以后的部分”。深度价外 Put 的初始保费通常更低，代价是许多回撤没有到期赔付。第 ② 条对应第 10 章发行人公开的 Put-spread Collar 方法：卖出上行和更深的下行来补贴保费，同时引入封顶与保护失效区间。第 ③ 条只是待检验假设：低 IV 分位不保证保护“便宜”或随后下跌，必须与常态配置和不对冲方案做样本外、扣费后的完整净损益比较。 Lever ① is the essence of tail hedging: stop insuring “everything below −5%” and insure only “everything below −20%.” Deep OTM puts usually have lower initial premium, at the cost of no expiry payout in many drawdowns. Lever ② corresponds to Chapter 10's issuer-published put-spread collar method: sell upside and deeper downside to subsidize premium while adding a cap and a protection gap. Lever ③ is only a hypothesis to test: low IV rank does not guarantee that protection is “cheap” or that a decline follows; compare it out of sample and after costs against always-on and unhedged alternatives using complete net P/L.

TAIL HEDGE 深度价外 Put · 拖动天数看保费怎么流走deep OTM put · drag the days and watch the premium bleed

读懂"+4,144%"这种数字Reading a "+4,144%" headline

2020 年 3 月后，多家媒体转述 Universa Investments 投资者信中的管理人声称：其 Black Swan Protection Protocol 在当月约 +3,612%、一季度约 +4,144%。这是对非公开投资者信的二手转述，不是公开审计报告，也不是整个客户组合的已验证回报。本书保留数字只为了教你不要直接使用它。 After March 2020, multiple outlets relayed a manager claim from a Universa Investments investor letter: roughly +3,612% for the Black Swan Protection Protocol in March and about +4,144% for the quarter. This is secondary reporting of a non-public investor letter, not a public audit and not a verified return for an entire client portfolio. The figures remain only to teach why they cannot be used directly.

原因是第 15 章那三个问题里的第一个：分母是什么？媒体转述将它描述为对对冲切片的“投入资本”回报，而不是客户整体组合回报。没有原始信件、统一的分母定义、费用口径与独立审计，这两个百分比不能用来做任何配置决策。它们最多提醒你凸性策略可以产生极端的切片回报，不能说明对整个组合的贡献。来源与方法 The reason is Chapter 15's first question: what is the denominator? Media retellings describe the figures as return on "invested capital" in the hedge sleeve, not on a client's whole portfolio. Without the original letter, a consistent denominator, fee convention, and independent audit, neither percentage can inform an allocation decision. At most they show that a convex sleeve can post an extreme sleeve-level return; they do not establish its contribution to a whole portfolio. Sources and methodology

13.3 不要用会衰减的工具做长期保险13.3 Never Use a Decaying Instrument as Long-Term Insurance

既然买 Put 这么贵、又要不停展期，一个自然的想法是：能不能直接买一个"做多波动率"的 ETF 拿着？它看起来省事得多——不用挑行权价、不用管到期日、像股票一样买入即可。 Given how expensive puts are and how relentlessly they must be rolled, a natural thought: why not just buy a long-volatility ETF and hold it? It looks far simpler — no strike to choose, no expiry to manage, buy it like a stock.

PRODUCT DOCUMENT · VXX

VXX 不是 VIX：先读它真正承诺跟踪什么VXX Is Not VIX: Read What It Actually Promises to Track

Cboe 的产品监管通知说得很明确：VXX 的回报链接至 S&P 500 VIX Short-Term Futures Index Total Return，该指数持有近月与次近月 VIX 期货，并逐日调整以维持约一个月的恒定到期暴露。它没有承诺按点对点复制现货 VIX，更没有承诺对你的股票组合进行固定赔付。 Cboe's product regulatory circular is explicit: VXX returns are linked to the S&P 500 VIX Short-Term Futures Index Total Return, which holds first- and second-month VIX futures and rebalances daily to maintain roughly one month of constant maturity. It does not promise point-for-point replication of spot VIX, and it does not promise a fixed payout against your equity portfolio.

当次近月期货高于近月（contango）时，这个逐日滚动会把暴露从较便宜的近月转向较贵的次近月，形成不利的展期项；backwardation 时方向可以反过来。正确的问题不是“VIX 会不会涨”，而是“这条期货指数路径与我要保护的损失路径是否匹配”。 When the second-month future is above the first (contango), the daily roll shifts exposure from the cheaper near contract toward the dearer next contract, creating an adverse roll component; in backwardation the direction can reverse. The right question is not "will VIX rise?" but "does this futures-index path match the loss path I want to insure?"

VIX: 由 SPX 期权报价计算的 30 天预期波动率指标30-day expected-volatility measure from SPX option quotes
VXX: 链接短期 VIX 期货指数的 ETNETN linked to a short-term VIX futures index
调整Rebalance: 近月/次近月期货之间逐日滚动daily roll between first- and second-month futures
不承诺Not promised: 现货 VIX 或股票组合的点对点对冲point-for-point hedge of spot VIX or an equity portfolio

这个区分不是语义游戏。一张股票 Put 在到期日跌破 Strike 时有明确的合约赔付；VXX 只给你一条短期期货指数的回报路径。即使 VIX 在某个时点上升，期货曲线、每日滚动与你持有期间都会影响 VXX 的实际结果。因此它可以是交易波动率期货暴露的工具，却不能被不经验证地当成股票保单。 That distinction is not semantic. An equity put has a defined contractual payout below its strike at expiry; VXX gives you the return path of a short-term futures index. Even if VIX rises at some point, the futures curve, daily roll, and your holding period all affect VXX's realized result. It can be a vehicle for trading short-term volatility-futures exposure, but it cannot be assumed to be an equity insurance policy without testing the mapping.

证据声明：本节只使用 Cboe 监管通知与官方 VIX 风险披露中可验证的产品映射，不再使用任意截止日、衔接两代 ETN 后得出的累计涨跌或年化数字。来源与方法 Evidence statement: this section uses only the product mapping verifiable in Cboe's regulatory circular and official VIX risk disclosures. It no longer uses cumulative or annualized returns dependent on an arbitrary cutoff date and a chained series across two ETNs. Sources and methodology

映射风险降低：把 VXX 误当成“现货 VIX”或“股票 Put” → 按产品文件将它识别为每日滚动的短期 VIX 期货指数暴露Mapping risk reduced: treating VXX as "spot VIX" or an "equity put" → identifying it from product documents as daily-rolled short-term VIX futures-index exposure

13.4 那么什么时候对冲是理性的13.4 So When Is Hedging Rational?

即使完整回测才能决定净成本，一个更重要的问题仍然成立：如果对冲可能降低收益，为什么还要付费？ Even though only a complete backtest can establish net cost, a more important question remains: if hedging may reduce return, why pay for it?

答案不仅在期望值，也在路径与约束里。第 9 章那条回撤曲线说过：−50% 需要 +100% 才能回本，而这个算术对一个必须在特定时间用钱的人尤其危险。如果你还有三十年才退休，可能有时间等待恢复；如果你明年要用这笔钱，同样的回撤可能迫使你在低点变现。对冲会改变回报分布，通常也会改变期望收益；它的价值要与负债期限、可承受回撤和成本一起评估。 The answer lies not only in expectancy but in path and constraints. Chapter 9's drawdown curve established that −50% requires +100% to recover, and that arithmetic is especially dangerous for someone who must spend the money on a particular date. Thirty years before retirement, there may be time to recover; if the money is needed next year, the same drawdown may force liquidation near a low. Hedging changes the return distribution and usually changes expected return too; its value must be evaluated alongside liability horizon, drawdown capacity, and cost.

决策输入Decision input: 资金的期限与用途Liability date and purpose
候选规则Candidate rule: 低 IV 分位部署；仍需样本外净损益验证Deploy at low IV rank; still needs out-of-sample net-P/L validation
预算压力Budget stress: 压力进入 IV 后，同一保护可能更贵The same protection may cost more once stress enters IV
映射风险Mapping risk: 长期持有会衰减的工具前先验证路径匹配Validate path fit before holding a decaying vehicle long term

思考Think 13.1 的静态情景表显示，IV 从 15 升到 45 时，同一规则的毛保费可相差八倍。压力时期 IV 往往会上升，因此一个自洽的对冲计划需要事先定义预算、触发条件与“不买”的边界，而不是在情绪最强时临时决定。把这个问题和第 9 章“连续成功会放大仓位”放在一起想：你的规则能否在平静与压力两种状态下都被执行和审计？ §13.1's static scenarios show that when IV rises from 15 to 45, gross premium for the same rule can differ eightfold. IV often rises during stress, so a coherent hedge plan needs a pre-defined budget, trigger, and no-buy boundary rather than an improvised decision at peak emotion. Consider this alongside Chapter 9's observation that a run of success inflates position size: can the rule be executed and audited in both calm and stressed states?

13.5 本章测验13.5 Chapter Quiz

点击选项即可作答，答案与解析即时显示。Click an option to answer; the correct answer and explanation appear immediately.

按本章静态输入，每月滚动买 5% 价外 Put 的毛支出年化约 8.8%。关于这个数字，哪句话准确？Under the chapter's static inputs, rolling a monthly 5%-OTM put annualizes to about 8.8% of gross outlay. Which statement is accurate?

C。8.8% 是单月模型保费 × 12 的毛支出，没有计入任何赔付、实际展期、路径或交易成本。A 把它误称净成本；B 把不可比的毛支出与历史净总回报直接相减。年度收益正负不能推断月度 Put 赔付，所以本章能得出的只是预算量级，不是净拖累或对冲后期望收益。D 与情景表直接矛盾：同一模型的毛支出从 4.5% 到 37.2% 随 IV 大幅变化。 C. The 8.8% is one model month's premium × 12, with no payouts, actual rolls, path, or trading costs. A mislabels it as net cost; B directly subtracts an incomparable gross outlay from a historical net total return. Annual gain or loss does not identify monthly put payouts, so this chapter establishes a budget magnitude, not net drag or hedged expected return. D contradicts the scenario table: gross outlay in the same model changes from 4.5% to 37.2% as IV changes.

在本章固定其他输入的模型中，IV≈15 与 IV≈45 情景的毛保费年化为 4.5% 和 37.2%。这能推出什么？In the chapter's model with other inputs fixed, gross premium annualizes to 4.5% at IV≈15 and 37.2% at IV≈45. What follows?

B。本题只支持条件性结论：在其他输入相同的模型里，IV≈45 的毛保费约为 IV≈15 时的八倍。压力和 IV 上升常常同现，但并非机械同步，也不表示压力中绝对不能买。正确动作是重新比较剩余尾部风险、当时报价、预算和替代方案，而不是把“反应够快”当成价格不会变化。 B. The question supports a conditional conclusion only: with other model inputs held fixed, gross premium at IV≈45 is about eight times the IV≈15 case. Stress and rising IV often coincide, but not mechanically, and this does not make buying during stress impossible. The correct action is to reassess residual tail risk, current quotes, budget, and alternatives—not assume that reacting quickly leaves price unchanged.

VXX 的产品文件实际承诺跟踪什么？What do VXX's product documents actually promise to track?

D。Cboe 文件定义的是第一、第二月 VIX 期货之间逐日滚动的指数暴露。A 错在现货 VIX 不是 VXX 的直接标的；B 和 C 都是产品没有做出的保证。先把工具映射到真正的标的，再谈它能否对冲你的风险。 D. Cboe documents define daily-rolled exposure between first- and second-month VIX futures. A is wrong because spot VIX is not VXX's direct underlying; B and C are guarantees the product never makes. Map the instrument to its actual reference first, then ask whether it hedges your risk.

媒体转述某尾部对冲管理人声称 2020 年一季度回报 +4,144%。在用它做判断之前，最该先问什么？Media relay a tail-hedge manager's claim of +4,144% for Q1 2020. Before using it, what must you ask first?

A。媒体将这类回报描述为投入资本（对冲切片）回报，而不是整个组合回报。在能说出分母、费用口径和证据性质之前，这个数字不能说明它对你组合的贡献。B 也必须问：这里只有对管理人声称的二手转述，没有公开审计记录；但即使日后验真，分母问题仍然使 A 成为配置前的首要问题。 A. Media describe such a figure as return on invested capital in the hedge sleeve, not on the whole portfolio. Until you can state the denominator, fee convention, and evidence type, the number cannot establish contribution to your portfolio. B must also be asked: this is a secondary retelling of a manager claim, with no public audit. But even if later verified, the denominator still makes A the first allocation question.

即使对冲可能降低收益，什么情况下它仍可能是理性的？Even if hedging may reduce return, when can it still be rational?

C。对冲会改变回报分布，通常也会改变期望收益；关键是这种改变是否匹配负债期限与可承受回撤。还有三十年才用的钱可能有时间等待恢复，明年要用的钱却可能被迫在低点变现。B 依赖预测能力，D 则忽略了高 IV 下保险更贵。理性与否要比较路径风险的减少和完整净成本，不能由“对冲”这个名称预先决定。 C. Hedging changes the return distribution and usually expected return as well; the question is whether that change fits the liability horizon and drawdown capacity. Money needed in thirty years may have time to recover, while money needed next year may be forced out near a low. B depends on forecasting, and D ignores that insurance is more expensive at high IV. Rationality requires comparing reduced path risk with complete net cost; the label “hedge” cannot decide it in advance.

本章要点Key Takeaways

按本章明示的 Black-Scholes 假设，每月滚动 5% 价外 Put 的毛保费支出会随 IV 大幅变化：平静、常态、紧张和恐慌四个示意情景分别为年化 4.5%、8.8%、19.3% 与 37.2%。这些不是历史策略回报，尚未扣除到期赔付、交易成本和路径影响。降低毛支出的常见结构手段包括买得更远、卖出保护腿形成价差，或只在预先定义的条件下部署；是否改善组合结果必须用完整净损益检验。VXX 也不能直接替代股票 Put：产品文件定义它跟踪逐日滚动的短期 VIX 期货指数，而不是现货 VIX 或你的股票损失。对冲的决定最终取决于负债期限、可承受回撤、成本和工具与风险的映射，不是媒体回报标题或单一波动率观点。 Under this chapter's disclosed Black–Scholes assumptions, the gross premium outlay for rolling a monthly 5%-OTM put changes sharply with IV: the calm, normal, tense and panic scenarios annualize to 4.5%, 8.8%, 19.3% and 37.2%. These are not historical strategy returns and do not include option payouts, trading costs or path effects. Common structural ways to lower gross outlay include buying farther out, selling a protective leg to form a spread, or deploying only under pre-defined conditions; whether any choice improves the portfolio must be tested on complete net P/L. VXX is not a direct substitute for an equity put either: its product documents define exposure to a daily-rolled short-term VIX futures index, not spot VIX or losses on your equities. The hedging decision ultimately depends on liability horizon, drawdown capacity, cost and the instrument-to-risk mapping—not a media return headline or a single volatility view.