Chapter 15 · Evidence Literacy

证据素养：这一切长期到底值不值得做Evidence Literacy: Was Any of This Worth Doing

前十四章教的是怎么把观点做成结构。这一章问一个所有人都该在开始之前问、而几乎所有人都在亏钱之后才问的问题：这些策略在几十年的真实数据里，到底跑出了什么结果？本章是全书唯一一章没有任何示意值——所有数字都来自公开发布的基准指数，可以逐个核对。 The first fourteen chapters taught how to turn a view into a structure. This one asks the question everybody should ask before starting and almost everybody asks only after losing money: across decades of real data, what did these strategies actually produce? It is the only chapter in this book with no illustrative values at all — every figure comes from a published benchmark index and can be checked line by line.

15.1 为什么这一章放在最后15.1 Why This Chapter Comes Last

本书用了十四章讲结构，而每一个案例都是单次事件：NVDA 那一夜、COVID 那五周、CPI 那一天。单次事件能说明机制，但它说明不了期望。第 8 章已经点破过这个取样偏差——那些案例都是幸存下来的故事。要回答"长期值不值得做"，你需要的不是又一个精彩案例，而是一条不加挑选、连续几十年、包含所有好年份和所有坏年份的记录。 Fourteen chapters of structures, and every case was a single episode: that one NVIDIA night, those five COVID weeks, that one CPI day. Episodes demonstrate mechanism; they cannot establish expectancy. Chapter 8 named the sampling bias — those cases are all stories that survived. To answer "is this worth doing over a lifetime," you need not another vivid case but an unselected, continuous, decades-long record containing every good year and every bad one.

这样的长期基准是存在的，但先要说清它是什么。Cboe 发布的 BXM 策略基准指数假设持有标普 500，并按规则每月卖出平值 Call。其历史序列追溯到 1986 年，指数在 2002 年正式发布；发布前的数据属于按后来方法回算的假设结果。它回答的是“同一套公开规则在这段历史数据上会呈现什么回报”，不是某位投资者四十年的真实成交记录。这条边界不会让基准失去价值，反而决定了我们能从它推出什么、不能推出什么。 A long benchmark does exist, but its nature matters. Cboe's BXM strategy index assumes an S&P 500 position and a rules-based monthly sale of an at-the-money call. Its historical series reaches back to 1986, while the index itself launched in 2002; the pre-launch history is a hypothetical back-calculation under the later methodology. It answers “what would these published rules have produced on this historical dataset?”, not “what did one investor actually earn through forty years of fills?” That boundary does not make the benchmark useless; it defines what can and cannot be inferred from it.

BXM 的结构就是第 2 章那张图is the shape from Chapter 2

15.2 四十年的假设性基准记录15.2 Forty Years of Hypothetical Benchmark History

CASE FILE · 1986–2026

备兑开仓的长期基准：波动更小，回撤更浅，然后呢？The Long Buy-Write Benchmark: Less Volatility, Shallower Drawdowns, and Then?

Cboe 官方假设性基准数据（历史序列自 1986 年 6 月 20 日起，截至 2026 年 9 月 30 日）： Official Cboe hypothetical benchmark data (historical series from June 20, 1986, as of September 30, 2026):

年化收益Annualized return: BXM 8.6% · 标普全收益 11.2%BXM 8.6% · S&P 500 TR 11.2%
年化波动率Annualized volatility: BXM 10.6% · 标普 15.2%BXM 10.6% · S&P 15.2%
最大回撤Max drawdown: BXM −35.8% · 标普 −50.9%BXM −35.8% · S&P −50.9%
Beta: BXM 0.62 · 标普 1.0BXM 0.62 · S&P 1.0

在 BXM 的特定规则与这段历史样本里，买写基准的年化波动更低，最大回撤也浅了 15.1 个百分点；与此同时，年化收益少 2.6 个百分点。这证明的是该基准在该窗口中的风险收益形态，不是对所有 Covered Call 规则、所有未来窗口或任何实际产品的普遍定律。 Under BXM's particular rules and this historical sample, the buy-write benchmark had lower annualized volatility and a maximum drawdown 15.1 percentage points shallower, while annualized return was 2.6 points lower. That establishes the benchmark's risk-return shape in this window, not a universal law for every covered-call rule, every future period or any investable product.

2.6 个百分点听起来不多。把它按四十年复利展开就不是了。用同一份数据里 2011–2025 的逐年收益，从 $100 起算： 2.6 points a year sounds small. Compounded over decades it is not. Using the calendar-year returns from the same source, 2011–2025, starting from $100:

$100 从 2010 年底起（Cboe 逐年收益，全收益口径）from year-end 2010 (Cboe calendar-year returns, total return)

十五年后：BXM $283，标普 $720——相差 2.54 倍。这十五年里 BXM 只在 3 年跑赢（2011、2015、2022），跑输 12 年。 Fifteen years later: BXM $283, the S&P $720 — 2.54 times apart. Across those fifteen years BXM outperformed in only three (2011, 2015, 2022) and trailed in twelve.

但"跑输"本身不是判决——第 9 章讲过，低波动策略本来就该用风险调整后的口径衡量。所以看这份数据里最该看的两个数字： But "trailed" is not the verdict on its own — Chapter 9 established that a lower-volatility strategy deserves to be judged risk-adjusted. So look at the two numbers in this dataset that matter most:

夏普比率Sharpe ratio: BXM 0.56 · 标普 0.56BXM 0.56 · S&P 0.56
索提诺比率Sortino ratio: BXM 0.76 · 标普 0.84BXM 0.76 · S&P 0.84

这组数据最值得读的是：BXM 与标普的夏普比率同为 0.56；BXM 的索提诺比率为 0.76，对照为 0.84。在这套定义和历史窗口里，BXM 没有显示出优于直接持有指数的风险调整收益。这里不能再多走一步说“市场证明交换绝对公平”或“少掉的风险正好等于少掉的收益”——单一历史样本与两个比率不支持如此精确的因果结论。可支持的结论只有：这份长期基准没有提供免费午餐或持久超额收益的证据。 The most informative comparison is that BXM and the S&P have the same Sharpe ratio, 0.56, while BXM's Sortino ratio is 0.76 versus 0.84. Under these definitions and this historical window, BXM does not show superior risk-adjusted return to simply owning the index. One cannot take the extra step of claiming that the market proved the exchange perfectly fair or that every unit of avoided risk exactly matched foregone return — one historical sample and two ratios do not support that precise causal conclusion. The defensible result is narrower: this long benchmark provides no evidence of a free lunch or durable excess return.

数据来源声明：以上全部数字取自 Cboe Global Indices 的 BXM 官方 factsheet（数据截至 2026 年 9 月 30 日），包括年化收益、波动率、最大回撤、Beta、夏普与索提诺比率，以及 2011–2025 的逐年收益。累计增长曲线由该逐年收益直接复利计算。本章不含任何由本书推算的期权价格。需要注意：夏普与索提诺比率按该 factsheet 说明基于 1989 年 7 月起的数据；指数收益为税前，且不含基金管理费、税负等投资者层面的费用（见 15.4 节）。 Source statement: every figure above is taken from Cboe Global Indices' official BXM factsheet (data as of September 30, 2026) — annualized return, volatility, maximum drawdown, beta, Sharpe and Sortino ratios, and the 2011–2025 calendar-year returns. The cumulative growth curve is compounded directly from those annual returns. This chapter contains no option prices computed by this book. Note that the factsheet bases its Sharpe and Sortino ratios on data from July 1989 and reports a pre-tax index return without investor-level costs such as fund management fees and taxes (see §15.4).

基准结果：最大回撤 −50.9%（标普）→ −35.8%（BXM），浅 15.1 个百分点；年化收益少 2.6 点，且未显示更高的风险调整收益（夏普同为 0.56）Benchmark result: maximum drawdown moved from −50.9% (S&P) to −35.8% (BXM), 15.1 points shallower; annualized return was 2.6 points lower, with no superior risk-adjusted return shown (Sharpe 0.56 for both)

15.3 同类策略，两组不能直接比较的数据15.3 Related Strategies, Two Incomparable Datasets

上一节描述 BXM 的一条长期样本。2009 年另一份权威研究则报告 PUT 指数在另一个窗口跑赢其基准。两份摘要看起来方向相反，但它们不是对同一个策略、同一窗口和同一基准的配对实验。 The previous section describes one long BXM sample. A separate authoritative 2009 study reports that the PUT Index beat its benchmark over a different window. The summaries appear to point in opposite directions, but they are not a paired experiment on the same strategy, window, and benchmark.

2009 年 1 月研究Jan 2009 study: PUT 指数 10.32% · 标普 8.77%（卖方 +1.55 点）PUT 10.32% · S&P 8.77% (writer +1.55)
窗口Window: 1986-07 → 2008-10Jul 1986 → Oct 2008
2026 年 9 月数据As of Sep 2026: BXM 8.6% · 标普 11.2%（卖方 −2.6 点）BXM 8.6% · S&P 11.2% (writer −2.6)
窗口Window: 1986-06 → 2026-09Jun 1986 → Sep 2026

2008 年 10 月处于金融危机深处，终点选择当然可能影响结果；但这里策略从 PUT 换成 BXM、窗口延长近十八年、基准口径也未被这张并列表统一。这些变量同时改变，因此不能把两组结果相减或归因于单一差异。上一节的 Beta 0.62 属于 BXM 资料，也不能套到 2009 年的 PUT 样本。这个并列只负责触发可比性审查，不负责证明截止日效应。 October 2008 sits deep in the financial crisis, so endpoint choice may matter. But here the strategy changes from PUT to BXM, the window extends by nearly eighteen years, and this side-by-side does not harmonize benchmark conventions. Because those variables change together, the two results cannot be subtracted or attributed to one difference. The 0.62 beta in the previous section belongs to the BXM factsheet and cannot be applied to the 2009 PUT sample. This juxtaposition triggers a comparability audit; it does not prove an endpoint effect.

在你比较任何两个收益率之前Before you compare any two return figures

上面两组数字严格来说甚至不能直接相减，而这本身就是本章要教的东西。它们至少有三处不可比：① 指数不同——PUT 卖 Put、BXM 卖 Call，是两个不同的策略指数；② 窗口不同——差了近十八年，且一个终止于崩盘谷底；③ 对照口径可能不同——标普 500 的"价格收益"与"全收益"相差每年约两个百分点（股息），而很多材料不写清楚用的是哪一个。 Strictly, those two sets of numbers cannot even be subtracted from one another — and that is itself the lesson. They differ in at least three ways: ① different indices — PUT sells puts, BXM sells calls; two different strategies; ② different windows — nearly eighteen years apart, one ending at a crash bottom; ③ possibly different benchmark variants — the S&P 500's "price return" and "total return" differ by roughly two points a year (dividends), and a great deal of material never says which one it used.

所以这一节真正的产出不是"卖方到底赢没赢"，而是一个习惯：看到任何一个"某策略年化跑赢 X 个百分点"的说法，先问三个问题——哪个指数、哪个窗口、对照的是价格收益还是全收益。三个问题里只要有一个答不上来，那个数字就不能用来做决策。 So this section's output isn't "did writers win" but a habit: whenever you meet a claim that some strategy beat the market by X points a year, ask three questions first — which index, which window, and price return or total return. If even one has no answer, that number cannot be used to make a decision.

15.4 指数不是基金15.4 An Index Is Not a Fund

还有一层，在上面所有数字之外。BXM 是一个指数，不是一只你能买的基金。Cboe 自己的免责声明写得毫不含糊：指数值是理论值，不反映实际交易结果，也不反映买卖、持有相关投资所产生的费用、税负或其他开支。 There is one more layer beneath all of the above. BXM is an index, not a fund you can buy. Cboe's own disclaimer is unambiguous: index values are theoretical, do not reflect the results of actual trading, and do not reflect the fees, taxes or other expenses associated with buying, holding or selling investments.

把 8.6% 变成你账户里的数字，至少要另外核算四层：① 产品或管理费；② 券商佣金、报价与真实成交执行；③ 税务——月度卖 Call 会持续产生潜在税务事件，具体处理因账户与辖区而异；④ 跟踪偏差——真实产品的展期时间、现金管理、持仓约束和操作选择未必逐点复制指数。这里还要纠正一个常见误解：按 Cboe 当前 BXM 方法文件，到期 Call 用 SOQ 结算，新 Call 的权利金按展期日 11:30–13:30 ET 的成交 VWAP 计入；它不是简单假设“所有腿都按结算价成交”。但指数仍不等于任何投资者可直接获得的税后、费后回报。 Turning 8.6% into an account return requires four additional reconciliations: ① product or management fees; ② brokerage commissions, quotes and actual execution; ③ taxes — monthly call writing can create recurring taxable events whose treatment depends on account and jurisdiction; and ④ tracking difference — a real product's roll timing, cash management, constraints and discretionary operations may not replicate the index point for point. One common misconception also needs correcting: under Cboe's current BXM methodology, the expiring call settles to the SOQ and the new call premium is incorporated using the 11:30 a.m.–1:30 p.m. ET trading VWAP on the roll date. The index does not simply assume that every leg fills at a settlement price. It still is not an after-tax, after-fee return directly available to an investor.

这一层为什么是课程内容，而不是脚注Why this layer is the lesson, not a footnote

因为这些摩擦对比较两边未必相同。低成本广泛指数产品通常只需很少交易，而买写产品需要持续管理期权腿；在其他条件相同时，额外费用与执行摩擦会给 BXM 一侧增加向下拖累。税务则必须按账户与辖区单独计算，不能宣称对每位投资者都“只会更差”。所以严谨的结论是：8.6% 对 11.2% 是指数层的比较；任何产品或个人策略都必须用自己的费后、税后、可实现回报重新比较，不能直接把 BXM 指数收益当到账户收益。 Because those frictions need not affect both sides equally. A low-cost broad-index product generally trades very little, whereas a buy-write product must continually manage an options leg; all else equal, additional fees and execution frictions create extra downward drag on the BXM side. Taxes, however, must be calculated for the actual account and jurisdiction, so “it can only get worse” is not a universal statement for every investor. The defensible conclusion is narrower: 8.6% versus 11.2% is an index-level comparison. Any fund or personal implementation must be re-evaluated using its own realizable after-fee and after-tax returns rather than treating the BXM index return as an account return.

15.5 那么第 2 章错了吗15.5 So Was Chapter 2 Wrong?

没有。而弄清楚"为什么没错"，是本章最后、也最重要的一步。 No. And working out why not is this chapter's final and most important step.

CASE FILE · 2018 vs 2022

两个下跌年份，同一个策略，相反的结果Two Down Years, One Strategy, Opposite Results

2011–2025 这十五年里，标普全收益只有两年为负。备兑策略号称在下跌时提供缓冲，那就看这两年： Across 2011–2025, the S&P's total return was negative in only two years. Buy-writing claims to cushion declines, so look at exactly those two:

2022: BXM −11.4% · 标普 −18.1%（少亏 6.7 点）BXM −11.4% · S&P −18.1% (6.7 pts better)
2018: BXM −4.8% · 标普 −4.4%（多亏 0.4 点）BXM −4.8% · S&P −4.4% (0.4 pts worse)

同一个指数规则，同样是下跌年，一次少亏 6.7 个百分点，一次反而多亏 0.4 个百分点。这两个被选中的年份与备兑策略的路径依赖机制一致：持续下行期间，权利金可能提供部分缓冲；急跌后快速反弹时，卖出的 Call 又可能限制修复收益。但逐年收益本身不能证明这个解释，更不能证明未来能提前识别哪种路径。 The same index rule, in two down years, was 6.7 points better in one and 0.4 points worse in the other. These selected years are consistent with the path-dependent mechanism of a covered call: premium may cushion part of a sustained decline, while sold calls may limit recovery after a sharp fall and rebound. The annual returns alone do not prove that explanation, however, and they certainly do not show that the future path can be identified in advance.

因此，第 2 章的结构机制与 BXM 的长期基准证据回答的是不同问题。前者说明损益如何随路径变化；后者显示一套固定规则在这段假设性指数历史中的结果。两者都不能证明“读 Regime”本身有可交易的优势。若要提出状态切换策略，必须先定义可观测信号，再做样本外、扣费后的比较。 Chapter 2's structural mechanism and BXM's long-run benchmark evidence therefore answer different questions. The former explains how payoff changes with the path; the latter records one fixed rule over this hypothetical index history. Neither proves that "reading the regime" is itself a tradable edge. A regime-switching strategy would require observable signals specified in advance and an out-of-sample, after-cost comparison.

数据来源声明：2018 与 2022 的逐年收益取自 Cboe BXM factsheet（全收益口径，与标普 500 全收益指数并列公布）。对两年市场形态的描述属于解释，不属于该数据源的结论。 Source statement: the 2018 and 2022 calendar-year returns are from the Cboe BXM factsheet (total return, published alongside the S&P 500 Total Return Index). The characterisation of the two years' market shapes is interpretation, not a finding of that source.

历史观察：BXM 在 2022 年少亏 6.7 个百分点，在 2018 年却多亏 0.4 个百分点；这说明结果依赖路径，不证明存在可事前交易的 Regime 信号Historical observation: BXM lost 6.7 points less in 2022 but 0.4 points more in 2018; this illustrates path dependence, not a regime signal that can be traded in advance

思考Think 在这份历史指数样本里，BXM 与标普的夏普比率同为 0.56。如果一个基准没有显示更高的风险调整后收益，它仍可能满足什么投资目标？先想想第 9 章那条回撤曲线：−35.8% 和 −50.9% 对真实投资者的持有体验并不相同。更平缓的路径是否值得较低的历史收益，取决于投资者约束；仅凭两个比率不能证明交换“恰好公平”。 In this historical index sample, BXM and the S&P both have a Sharpe ratio of 0.56. If a benchmark does not show a higher risk-adjusted return, what investor objective might it still serve? Revisit Chapter 9's drawdown curve: −35.8% and −50.9% do not create the same holding experience for a real investor. Whether a smoother path is worth the lower historical return depends on investor constraints; two ratios cannot prove that the exchange was "exactly fair."

15.6 本章测验15.6 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.

四十年里 BXM 年化 8.6%、标普全收益 11.2%，但 BXM 波动率和回撤都低得多。最准确的结论是？Over forty years BXM returned 8.6% annualized against the S&P's 11.2% total return, but with much lower volatility and drawdown. The most accurate conclusion:

C。A 和 B 都只看了一半：比较收益必须同时看风险。样本中的夏普同为 0.56，索提诺是 0.76 对 0.84，因此没有显示 BXM 提供了更高的风险调整后收益。这不等于证明两者“恰好公平”或未来也会如此；它只描述这条假设性指数历史。投资者是否愿意用部分收益换取较低波动和较浅回撤，还取决于负债、税务与行为约束。 C. A and B each see only half the comparison: return must be considered together with risk. In this sample the Sharpe ratios are equal at 0.56 and the Sortino ratios 0.76 versus 0.84, so BXM did not show a higher risk-adjusted return. That does not prove the exchange was "exactly fair" or that the future will match; it describes this hypothetical index history. Whether lower volatility and drawdown justify giving up some return depends on an investor's liabilities, taxes and behavioural constraints.

一份 2009 年 1 月发布的研究显示卖权策略年化跑赢标普 1.55 个百分点，窗口是 1986-07 至 2008-10。最该警惕什么？A study published in January 2009 shows a put-writing strategy beating the S&P by 1.55 points annualized, over July 1986 to October 2008. What should you be most wary of?

B。2008 年 10 月这个终点值得做敏感性分析，但不能拿 PUT 的该窗口与 BXM 的另一窗口直接相减：策略、结束日期和基准口径同时改变。上一节的 0.62 Beta 也只属于 BXM，不能移植到 PUT。要检验截止日效应，应固定同一策略与同一基准，改变终点并报告滚动窗口或端点敏感性。A 与 D 只讨论样本长度，C 则攻击来源而没有解决可比性。 B. The October 2008 endpoint deserves sensitivity analysis, but the PUT window cannot be directly subtracted from a different BXM window: strategy, endpoint, and benchmark convention all change together. The previous section's 0.62 beta belongs only to BXM and cannot be transferred to PUT. Testing an endpoint effect requires holding strategy and benchmark fixed, varying the endpoint, and reporting rolling-window or endpoint sensitivity. A and D discuss length only; C attacks the source without resolving comparability.

你打算按 BXM 的方式自己每月卖备兑 Call。相对那 8.6% 的指数年化，你实际拿到的大概率会？You plan to run BXM's monthly covered-call program yourself. Relative to that 8.6% index return, what you actually receive will most likely be:

D。Cboe 的免责声明写明指数值是理论值，不反映买卖、持有投资产生的费用、税负或其他开支。真实账户还会受到佣金、点差、相对指数规则的执行偏差与账户税制影响；这些都必须单独测算。指数方法本身也不是“任意时点按一个结算价成交”：到期 Call 使用 SOQ，新 Call 使用规定时段的 VWAP。在其他条件相同的情况下，费用与交易摩擦会拖累可实现收益；税务和实际执行造成的差异则取决于账户与实施方式，不能从 8.6% 这个毛指数数字直接推出。 D. Cboe's disclaimer says index values are theoretical and do not reflect the fees, taxes or other expenses of buying, holding or selling investments. A real account is also affected by commissions, spreads, implementation drift from the index rules and its own tax regime, all of which require separate measurement. The methodology itself is not an arbitrary single settlement-price fill: it uses the SOQ for the expiring call and a prescribed VWAP window for the new call. All else equal, fees and trading friction reduce realizable returns; differences from taxes and execution depend on the account and implementation and cannot be inferred directly from the 8.6% gross index figure.

2022 年 BXM 少亏 6.7 个百分点，2018 年却多亏 0.4 个百分点。为什么？BXM was 6.7 points better in 2022 but 0.4 points worse in 2018. Why?

A。这正是第 2 章与第 9 章那条主线在四十年数据里的显影：Covered Call 的缓冲来自"月月能顺利收租"，而不是来自"市场下跌"这件事本身。2022 年全年阴跌，每个月的租金都落袋；2018 年的跌幅集中在第四季度急挫、随后迅速反弹，反弹会击穿卖出的 Call，于是缓冲没兑现、封顶却兑现了。C 把一个条件性结论说成了必然，而这两年恰好证伪了它。决定它有没有用的是 Regime，不是"涨还是跌"。 A. This is the through-line of Chapters 2 and 9 developing in forty years of data: a covered call's cushion comes from collecting rent cleanly month after month, not from the market falling as such. 2022 ground lower all year and every month's rent was banked; 2018's decline was concentrated in a sharp Q4 drop followed by a rapid rebound, and rebounds breach the calls you sold — so the cushion never arrived while the cap did. C states a conditional result as a certainty, and these two years are precisely what falsifies it. What decides whether it helps is the regime, not merely up-or-down.

读到"某策略年化跑赢市场 X 个百分点"时，本章建议先问哪三个问题？On meeting a claim that "strategy X beats the market by N points a year," which three questions does this chapter say to ask first?

C。这三个问题决定了两个数字能不能被放在一起比较——本章那两组权威数据严格来说甚至不能相减：指数不同（PUT 卖 Put vs BXM 卖 Call）、窗口不同（差近十八年，一个终止于崩盘谷底）、对照口径可能不同（标普价格收益与全收益每年差约两个百分点）。B 和 D 都是好问题，但它们是第二步：在确认两个数字可比之前，算风险调整或扣费用都没有意义。A 关注来源可信度，而本章那两份数据都可信——问题不在可信度，在可比性。 C. Those three decide whether two numbers can be compared at all — the chapter's two authoritative datasets strictly cannot even be subtracted: different indices (PUT sells puts, BXM sells calls), different windows (nearly eighteen years apart, one ending at a crash bottom), possibly different benchmark variants (the S&P's price and total return differ by about two points a year). B and D are good questions but they are step two: risk-adjusting or netting out fees is meaningless before you've established the figures are comparable. A asks about credibility, yet both of this chapter's sources are credible — the problem isn't credibility, it's comparability.

本章要点Key Takeaways

案例可以说明机制，却不能单独证明长期期望。Cboe BXM 提供了一条有价值的假设性基准历史：该指数 2002 年推出，1986 年至推出前的部分属于回溯计算；截至资料日期，年化 8.6% 对标普全收益 11.2%，波动率 10.6% 对 15.2%，最大回撤 −35.8% 对 −50.9%。夏普同为 0.56、索提诺 0.76 对 0.84，说明这份样本没有显示 BXM 提供更高的风险调整后收益，但也不能证明收益与平稳之间存在精确、普适的“公平交换”。指数还不包含投资者层面的费用、税负与全部执行偏差；真实账户必须按自身成本和税制重算。 Cases can illustrate mechanisms, but they cannot by themselves establish long-run expectancy. Cboe BXM provides a valuable hypothetical benchmark history: the index launched in 2002, so the segment from 1986 to launch is back-tested; as of the factsheet date it returned 8.6% annualized versus 11.2% for the S&P total-return benchmark, with 10.6% volatility versus 15.2% and a maximum drawdown of −35.8% versus −50.9%. Equal Sharpe ratios of 0.56 and Sortino of 0.76 versus 0.84 show no higher risk-adjusted return for BXM in this sample, but they do not prove an exact or universal "fair exchange" between return and smoothness. The index also omits investor-level fees, taxes and all implementation drift; a real account must be recalculated using its own costs and tax regime.

2022 与 2018 两个被选中的年份展示了路径依赖：持续阴跌与急跌后反弹会让备兑策略呈现不同结果。它们不能证明投资者能够事前识别 Regime，更不能证明择时规则拥有超额收益。任何“读状态”的规则都需要预先定义，并用样本外、扣除费用后的结果检验。至于窗口不同、结构不同的两份研究，它们首先提醒我们：看到任何跑赢多少个百分点的说法，先问哪个指数、哪个窗口、价格收益还是全收益；只要口径不同，就不能直接相减或据此决策。最后要学会的不是相信某个结构，而是审查数字的来源、定义与可比性。 The selected 2022 and 2018 episodes illustrate path dependence: a grinding decline and a sharp fall followed by a rebound can produce different outcomes for a covered-call benchmark. They do not show that an investor can identify the regime in advance, much less that a timing rule has excess return. Any regime rule must be specified beforehand and tested out of sample after costs. The two studies with different windows and structures chiefly teach a measurement habit: when a claim says a strategy beat the market by N points, first ask which index, which window, and whether the benchmark is price return or total return; unlike definitions cannot be subtracted directly or used as a decision rule. The final lesson is not to trust a particular structure, but to inspect each number's source, definition and comparability.