[{"data":1,"prerenderedAt":1173},["ShallowReactive",2],{"\u002Fzh\u002Fglossary\u002Fwhat-is-standard-deviation-in-trading":3,"\u002Fzh\u002Fglossary\u002Fwhat-is-standard-deviation-in-trading-related":299,"\u002Fzh\u002Fglossary\u002Fwhat-is-standard-deviation-in-trading-langs":1170},{"id":4,"title":5,"body":6,"category":281,"date":282,"description":283,"draft":284,"extension":285,"meta":286,"navigation":287,"path":288,"related":289,"seo":294,"stem":295,"term":296,"updated":297,"__hash__":298},"glossary\u002Fzh\u002F5.glossary\u002Fwhat-is-standard-deviation-in-trading.md","交易中的标准差是什么？公式与数据",{"type":7,"value":8,"toc":273},"minimark",[9,13,17,28,48,87,90,93,96,110,113,140,143,153,156,166,170,197,230,233,265],[10,11,12],"p",{},"标准差衡量的是单笔交易的结果通常离账户的平均交易结果有多远。ShowMyTrades 以账户货币显示它，所以它回答的问题很直接：在这个账户上，一笔交易和平常的交易通常相差多少？数值小，说明结果比较集中；数值大，说明这段历史是由离群值堆出来的。",[14,15,16],"h2",{"id":16},"工作原理",[18,19,24],"pre",{"className":20,"code":22,"language":23},[21],"language-text","标准差 = √( Σ (rᵢ − 均值)² \u002F n )\n","text",[25,26,22],"code",{"__ignoreMap":27},"",[10,29,30,31,34,35,39,40,43,44,47],{},"其中 ",[25,32,33],{},"rᵢ"," 是第 ",[36,37,38],"em",{},"i"," 笔已平仓交易的净结果，",[25,41,42],{},"均值"," 是这些结果的平均值，",[25,45,46],{},"n"," 是已平仓交易笔数。这个数字代表什么，取决于四个细节：",[49,50,51,62,68,81],"ul",{},[52,53,54,58,59,61],"li",{},[55,56,57],"strong",{},"每笔交易都按扣除成本后的净额计入","。",[25,60,33],{}," 等于盈利 + 佣金 + 库存费，所以佣金或库存费高的交易，压缩了多少盈利，就同样放大了多少离散程度。",[52,63,64,67],{},[55,65,66],{},"只统计已平仓的买入和卖出交易","。入金和出金在生成序列之前就已剔除，所以往账户里注资不会变成一个巨大的离群值。",[52,69,70,73,74,76,77,80],{},[55,71,72],{},"它用的是总体标准差","，除以 ",[25,75,46],{}," 而不是 ",[25,78,79],{},"n − 1","。已平仓交易不足两笔时，显示为 0.00。",[52,82,83,86],{},[55,84,85],{},"它以货币计，不是百分比","。同样是 $50 的标准差，在 $2,000 的账户和 $200,000 的账户上，说明的是完全不同的交易方式。一定要对照账户余额来看。",[10,88,89],{},"由于每个距离都先平方再平均，一笔异常交易对结果的影响远超十几笔普通交易，所以它特别容易揪出那笔格格不入的交易。",[14,91,92],{"id":92},"为什么重要",[10,94,95],{},"“稳定”这个词是真是假，标准差一测便知：谁都可以说自己的净值曲线很平稳，标准差则用一个货币数值告诉您，这些交易到底有多平稳。",[10,97,98,99,104,105,109],{},"它也是我们的",[100,101,103],"a",{"href":102},"\u002Fglossary\u002Fwhat-is-the-sharpe-ratio","夏普比率","的分母：夏普比率就是用这个数字作单位来衡量的平均交易结果。它还是",[100,106,108],{"href":107},"\u002Fglossary\u002Fwhat-is-expectancy","期望值","的天然搭档：期望值给出每笔交易的平均结果，却看不出这个平均结果有多靠得住。",[10,111,112],{},"最有用的读法是比较，拿它和几行之外的两个平均值对照：",[49,114,115,121,134],{},[52,116,117,120],{},[55,118,119],{},"接近平均盈利","：结果很集中，账户大部分时候表现一致。",[52,122,123,126,127,130,131,58],{},[55,124,125],{},"是平均盈利的好几倍","：整段历史靠少数几笔交易撑着。相信总数之前，先看",[55,128,129],{},"最佳交易（盈亏）"," 和",[55,132,133],{},"最差交易（盈亏）",[52,135,136,139],{},[55,137,138],{},"远大于平均亏损","：至少有一笔亏损远远超出平常水平。",[14,141,142],{"id":142},"数据显示了什么",[10,144,145,146,149,150,58],{},"以下数字描述的是 ShowMyTrades 上已发布的账户，不代表全体交易者。在有交易历史的已发布账户中（2026 年 8 月），",[55,147,148],{},"夏普比率中位数为 0.05","，",[55,151,152],{},"已平仓交易数中位数为 171 笔",[10,154,155],{},"按我们自己的定义倒推，这组数字能把离散程度说得很准：夏普比率是平均交易结果除以标准差，中位数 0.05 意味着，中间水平的账户平均每笔交易只值交易标准差的二十分之一左右。也就是说，一笔交易和下一笔之间的正常波动，大约是从中赚到的优势的二十倍。",[10,157,158,165],{},[55,159,160,164],{},[100,161,163],{"href":162},"\u002Fglossary\u002Fwhat-is-average-trade-length","持仓时间","中位数为 2.4 小时","，所以看一整天的结果，交易者学不到任何东西：一天的结果完全落在正常波动范围之内。同样道理，新账户的标准差低，也证明不了风险受控：样本里可能只是还没出现这个策略会产生的那个离群值，而距离要平方，它一旦出现，动静就会很大。",[14,167,169],{"id":168},"在-showmytrades-上的位置","在 ShowMyTrades 上的位置",[10,171,172,173,176,177,180,181,184,185,187,188,191,192,196],{},"账户页面的",[55,174,175],{},"高级统计","模块里，",[55,178,179],{},"表现指标","一栏中有一行，标签正是",[55,182,183],{},"标准差","，以账户货币显示，并附有提示说明。它紧挨在以它为分母的",[55,186,103],{},"上方、",[55,189,190],{},"盈利因子","下方，由我们的服务器根据存储的已平仓交易计算。在已发布的账户页面上，这个模块默认显示，不过",[100,193,195],{"href":194},"\u002Fguides\u002Fpublic-profile-and-custom-slugs","账户主人可以隐藏它","；可嵌入的小组件里没有这个模块。",[10,198,199,200,203,204,207,208,207,211,207,213,130,215,218,219,222,223,229],{},"把它和同一模块的",[55,201,202],{},"交易","一栏对照着看：",[55,205,206],{},"平均盈利","、",[55,209,210],{},"平均亏损",[55,212,129],{},[55,214,133],{},[55,216,217],{},"总交易数","为这个数字提供了参照。模块里的",[55,220,221],{},"持续时间","视图能把离散程度直接画出来：每笔已平仓交易是一个点，横轴是持仓小时数，纵轴是盈亏，离散程度就从一个数字变成了一个形状。",[100,224,226],{"href":225},"\u002Fguides\u002Fcustom-analysis-and-filter-presets",[55,227,228],{},"自定义分析","会在筛选出的子集上重新计算整个模块，可按日期范围、品种、魔术号、交易类型、手数等筛选，由此看出离散程度来自策略本身，还是来自某一个交易品种。",[14,231,232],{"id":232},"常见误解",[49,234,235,242,253,259],{},[52,236,237,238,241],{},"“",[55,239,240],{},"它是一个百分比。","”我们的标准差是每笔交易的金额。旁边没有余额作参照，在不同规模的账户之间比较毫无意义。",[52,243,237,244,247,248,252],{},[55,245,246],{},"它衡量的是市场波动率。","”它衡量的是您交易结果的",[100,249,251],{"href":250},"\u002Fglossary\u002Fwhat-is-volatility","波动率","。市场再平静，只要仓位越加越大，数值照样会很大。",[52,254,237,255,258],{},[55,256,257],{},"标准差低，风险就低。","”只有已平仓交易计入其中。扛着亏损单不平的系统，已平仓结果个个都小而整齐，真正的风险敞口却藏在浮动仓位里。",[52,260,237,261,264],{},[55,262,263],{},"30 笔交易下数值很低，可以放心。","”不到几百笔交易时，只能把它当作暂时的结果。",[10,266,267,268,272],{},"一组统计数据要有多少笔交易才有意义，请参阅",[100,269,271],{"href":270},"\u002Fguides\u002Fhow-much-history-a-track-record-needs","一份业绩记录需要多长的历史","指南。",{"title":27,"searchDepth":274,"depth":274,"links":275},2,[276,277,278,279,280],{"id":16,"depth":274,"text":16},{"id":92,"depth":274,"text":92},{"id":142,"depth":274,"text":142},{"id":168,"depth":274,"text":169},{"id":232,"depth":274,"text":232},"指标","2026-08-19T00:00:00.000Z","标准差衡量单笔交易的结果离您的平均交易结果有多远。本文介绍我们使用的公式（已扣除成本），以及如何在数千个账户上解读它。",false,"md",{},true,"\u002Fzh\u002Fglossary\u002Fwhat-is-standard-deviation-in-trading",[290,291,292,293],"what-is-the-sharpe-ratio","what-is-volatility","what-is-expectancy","what-is-the-sortino-ratio",{"title":5,"description":283},"zh\u002F5.glossary\u002Fwhat-is-standard-deviation-in-trading","标准差 (Standard Deviation)","2026-09-17T00:00:00.000Z","d23H7esNsSG175eWPyoZzoJTWWR3SEfy5D-ZKP-x9wU",[300,541,749,959],{"id":301,"title":302,"body":303,"category":281,"date":282,"description":529,"draft":284,"extension":285,"meta":530,"navigation":287,"path":531,"related":532,"seo":537,"stem":538,"term":539,"updated":297,"__hash__":540},"glossary\u002Fzh\u002F5.glossary\u002Fwhat-is-expectancy.md","什么是期望值？公式、示例与真实数据",{"type":7,"value":304,"toc":522},[305,308,310,316,323,326,332,335,337,344,352,360,362,374,392,411,413,425,465,482,484,515],[10,306,307],{},"期望值就是一笔交易的平均结果，可以用货币或点数表示：账户每开一笔仓位，平均能赚多少。它由胜率、平均盈利、亏损率、平均亏损四个数字算出，回答的是“这套系统每做一笔到底赚不赚钱”，而且结果可以直接乘以交易笔数。正期望值重复得足够多，就是一门生意；负期望值则是在倒数计时。",[14,309,16],{"id":16},[18,311,314],{"className":312,"code":313,"language":23},[21],"期望值 = (胜率 × 平均盈利) − (亏损率 × 平均亏损)\n\n亏损率 = 1 − 胜率\n",[25,315,313],{"__ignoreMap":27},[10,317,318,319,322],{},"假设一个系统胜率 40%，每笔盈利单赚 $300，每笔亏损单亏 $100：(0.40 × $300) − (0.60 × $100) = $120 − $60 = ",[55,320,321],{},"每笔交易 +$60","。一年做四百笔，预期结果就是 $24,000，这还没考虑仓位大小的变化，也没算上任何意外。",[10,324,325],{},"把同一个公式改用 R 倍数来写（R 即一笔平均亏损），账户货币的影响就去掉了，不同系统也能放在一起比较：",[18,327,330],{"className":328,"code":329,"language":23},[21],"期望值 (R) = (胜率 × 盈亏比) − (1 − 胜率)\n\n盈亏比 = 平均盈利 \u002F 平均亏损\n",[25,331,329],{"__ignoreMap":27},[10,333,334],{},"算出来的数字有没有意义，取决于两点。第一，输入里必须已经扣除成本：点差、佣金和库存费每笔都要付，盈利单因此变小，亏损单因此变大。第二，平均值背后要有足够多的交易，不能让某一笔特别大的结果单独左右它。",[14,336,92],{"id":92},[10,338,339,340,343],{},"统计数字要变成交易计划，靠的就是期望值。",[100,341,190],{"href":342},"\u002Fglossary\u002Fwhat-is-a-profit-factor","告诉您赚的钱和亏的钱之比；期望值告诉您再多做一笔交易值多少钱，把它乘以交易频率，就能得出一个月的预期结果。",[10,345,346,347,351],{},"胜率掩盖的取舍，也会在期望值里暴露出来。胜率只是优势的一半，另一半是",[100,348,350],{"href":349},"\u002Fglossary\u002Fwhat-is-risk-reward-ratio","风险回报比","，期望值正是两者的交汇点。动其中任何一个，结果都会跟着变；两个都不动，故事讲得再好，结果也不会变。",[10,353,354,355,359],{},"期望值还划出了成本的底线。如果一个策略每笔的期望是 +$8，而一个来回的",[100,356,358],{"href":357},"\u002Fglossary\u002Fwhat-is-a-spread","点差","加佣金就要 $7，剩下的那点东西已经算不上策略了。",[14,361,142],{"id":142},[10,363,364,365,149,368,149,371,58],{},"以下数字描述的是 ShowMyTrades 上已发布的账户，不代表全体交易者。在有交易历史的已发布账户中（2026 年 8 月）取中间水平：",[55,366,367],{},"胜率中位数 68.8%",[55,369,370],{},"盈利因子中位数 1.28",[55,372,373],{},"已平仓交易数中位数 171 笔",[10,375,376,377,380,381,384,385,58],{},"有了这几个数字，这个“中间账户”的期望值就能算出来。把盈利因子公式反推回去，平均盈利约为平均亏损的 ",[55,378,379],{},"0.58 倍","；剩下的用 R 形式一算便知：(0.688 × 0.58) − 0.312 = ",[55,382,383],{},"+0.087R","。也就是说，每做一笔交易，只赚到一笔平均亏损的百分之九。171 笔交易下来，全部利润加起来大约只相当于十五笔平均亏损，所以尽管近七成交易以盈利平仓，",[55,386,387,391],{},[100,388,390],{"href":389},"\u002Fglossary\u002Fwhat-is-time-weighted-return","时间加权收益率","的中位数也只有 +3.2%",[10,393,394,395,398,399,402,403,406,407,410],{},"优势薄到这个程度，成本就不是可以忽略的零头了。已发布账户累计支付了 ",[55,396,397],{},"$4,782,670 佣金","和 ",[55,400,401],{},"$862,547 库存费","；在产生过库存费的账户中，",[55,404,405],{},"86.3%"," 的净库存费为负。持仓时间中位数只有 ",[55,408,409],{},"2.4 小时","，在这种节奏下，0.087R 的优势要想保住，一个来回的成本相对平均亏损就必须足够小。",[14,412,169],{"id":168},[10,414,415,416,418,419,421,422,424],{},"每个已发布账户页面的",[55,417,175],{},"模块里都有",[55,420,108],{},"这一行，位于",[55,423,179],{},"分组。它在同一行里给出两种单位（先点数，后账户货币）：点数便于在不同品种之间比较，货币金额便于在不同仓位大小之间比较，单看哪一个都不完整。",[10,426,427,428,430,431,207,434,207,436,438,439,441,442,447,448,438,451,454,455,207,458,438,461,464],{},"它的四个输入在同一模块的",[55,429,202],{},"分组里：",[55,432,433],{},"胜率",[55,435,206],{},[55,437,210],{},"和",[55,440,217],{},"，旁边还有",[100,443,444],{"href":162},[55,445,446],{},"平均持仓时间","，能看出这点薄薄的单笔优势是几小时赚到的，还是几周才赚到的。已经扣除的成本显示在",[55,449,450],{},"佣金合计",[55,452,453],{},"已付库存费合计","两行，每笔交易的明细则在交易表格的",[55,456,457],{},"盈利（毛额）",[55,459,460],{},"库存费",[55,462,463],{},"佣金","列。",[10,466,467,468,472,473,207,476,438,479,481],{},"想检验一个期望值到底是一种优势，还是几种优势平均出来的，可以打开",[100,469,470],{"href":225},[55,471,228],{},"侧边面板：按品种、魔术号、方向、日期范围、星期或小时筛选出子集，整个区块就会重新计算。高级统计还提供",[55,474,475],{},"星期",[55,477,478],{},"小时",[55,480,221],{},"视图，不用筛选就能把同一结果拆开来看。",[14,483,232],{"id":232},[49,485,486,497,503,509],{},[52,487,237,488,491,492,496],{},[55,489,490],{},"正期望值就意味着我会赚钱。","”它只说明平均每笔交易是盈利的。仓位定错了，正期望值的系统照样会亏光：能不能活下来，是",[100,493,495],{"href":494},"\u002Fglossary\u002Fwhat-is-position-sizing","仓位管理","的问题，不是期望值的问题。",[52,498,237,499,502],{},[55,500,501],{},"期望值就是我的平均每笔交易。","”前提是平均值按净额计算。按毛利润算就漏掉了佣金和库存费，而薄优势恰恰是在这里消失的。",[52,504,237,505,508],{},[55,506,507],{},"高胜率就意味着正期望值。","”单看胜率说明不了任何问题。盈亏比为 0.58 时，60% 的胜率是个亏钱的系统。",[52,510,237,511,514],{},[55,512,513],{},"这个数字很稳定。","”只有几百笔交易时并不稳定。一笔特别大的盈利单就能把整个记录的期望值抬上去，把它去掉再算一遍，是个公平的检验。",[10,516,517,518,58],{},"哪些指标值得长期追踪、该按什么顺序看，请参阅",[100,519,521],{"href":520},"\u002Fguides\u002Ftracking-trading-performance","交易表现追踪完全指南",{"title":27,"searchDepth":274,"depth":274,"links":523},[524,525,526,527,528],{"id":16,"depth":274,"text":16},{"id":92,"depth":274,"text":92},{"id":142,"depth":274,"text":142},{"id":168,"depth":274,"text":169},{"id":232,"depth":274,"text":232},"期望值是单笔交易的平均盈亏：(胜率 × 平均盈利) − (亏损率 × 平均亏损)。本文给出公式、完整算例和公开账户的真实数据。",{},"\u002Fzh\u002Fglossary\u002Fwhat-is-expectancy",[533,534,535,536],"what-is-a-profit-factor","what-is-risk-reward-ratio","what-is-position-sizing","what-is-drawdown",{"title":302,"description":529},"zh\u002F5.glossary\u002Fwhat-is-expectancy","期望值 (Expectancy)","4gLLgYBgHrrvWKTKFixk2tUfM9kXODaMx4LcyeirnhQ",{"id":542,"title":543,"body":544,"category":281,"date":282,"description":739,"draft":284,"extension":285,"meta":740,"navigation":287,"path":741,"related":742,"seo":745,"stem":746,"term":747,"updated":297,"__hash__":748},"glossary\u002Fzh\u002F5.glossary\u002Fwhat-is-the-sharpe-ratio.md","什么是夏普比率？公式、局限与真实数据",{"type":7,"value":545,"toc":732},[546,549,551,554,560,563,569,576,579,581,584,590,598,600,610,624,627,629,665,668,686,688,725],[10,547,548],{},"夏普比率是平均收益除以这些收益的标准差，也就是账户每承受一单位波动，换来多少收益。两个账户一年下来增长一样，夏普比率却可能差很多：比率高的那个，一路走得更平稳。",[14,550,16],{"id":16},[10,552,553],{},"教科书上的定义用的是周期收益，并减去无风险利率：",[18,555,558],{"className":556,"code":557,"language":23},[21],"夏普比率 = (平均收益 − 无风险利率) \u002F 收益的标准差\n",[25,559,557],{"__ignoreMap":27},[10,561,562],{},"ShowMyTrades 按已平仓交易计算，而不是按日历周期，也不减去无风险利率：",[18,564,567],{"className":565,"code":566,"language":23},[21],"夏普比率 = 每笔交易结果的均值 \u002F 每笔交易结果的标准差\n",[25,568,566],{"__ignoreMap":27},[10,570,571,572,575],{},"每笔交易按“盈利 + 佣金 + 库存费”计入序列，所以输入已经扣除了成本。只统计已平仓的买入和卖出交易：入金、出金和余额操作在生成序列之前就已剔除。分母是同一序列的总体标准差（距离平方和除以交易笔数，而不是笔数减一）。所以在我们的统计表里，",[100,573,183],{"href":574},"\u002Fglossary\u002Fwhat-is-standard-deviation-in-trading","就排在这个比率正上方：它就是那个除数。已平仓交易不足两笔的账户显示 0.00。",[10,577,578],{},"按笔而不是按月计算，意味着这个数字没有年化，常见的经验标准也就用不上。“高于 1.0 算好，高于 2.0 算优秀”，说的是基于周期收益的年化夏普比率。在逐笔序列上，1.00 意味着平均每笔交易值整整一个交易结果标准差：交易笔数足以说明问题的账户，几乎没有能达到的。",[14,580,92],{"id":92},[10,582,583],{},"只看增长，只知道账户最后停在哪里；夏普比率还能说明它是怎么走过来的。优势小但可靠、反复执行的策略，比率不高，但为正；靠两笔运气交易撑起来的策略，均值很大，离散程度却更大，比率随之一落千丈。",[10,585,586,587,589],{},"有些图表只是看起来平滑，夏普比率能诚实地戳穿它，而它用的正是计算",[100,588,190],{"href":342},"的同一批已平仓交易。",[10,591,592,593,597],{},"它做不到的，是区分好的波动和坏的波动。一笔向上大幅偏离的交易，和一笔向下大幅偏离的交易，把分母抬高的幅度完全一样。",[100,594,596],{"href":595},"\u002Fglossary\u002Fwhat-is-the-sortino-ratio","索提诺比率","正是为了回应这一质疑而诞生的。",[14,599,142],{"id":142},[10,601,145,602,149,604,149,606,58],{},[55,603,148],{},[55,605,152],{},[55,607,608,164],{},[100,609,163],{"href":162},[10,611,612,613,615,616,619,620,623],{},"用我们自己的公式来读，0.05 的含义很具体：在中位数账户上，平均每笔交易约值交易结果标准差的二十分之一，差不多是二十份噪音对一份信号。这并不说明这些账户都很失败：",[100,614,390],{"href":389},"中位数为 ",[55,617,618],{},"+3.2%","，长期为正的占 ",[55,621,622],{},"63.0%","。真实的优势落到单笔交易上，就是这个样子。",[10,625,626],{},"这也说明了短样本为什么什么都证明不了。在这样的信噪比下，一个毫无优势的账户连赢二十笔也很平常，而一个确有优势的账户亏上一周，也说明不了任何问题。",[14,628,169],{"id":168},[10,630,172,631,176,633,180,635,637,638,640,641,644,645,207,647,207,650,207,652,655,656,659,660,664],{},[55,632,175],{},[55,634,179],{},[55,636,103],{},"，并附有提示说明。它的正上方是它自己的分母",[55,639,183],{},"，正下方是",[55,642,643],{},"夏普比率（年化）","（以净值的日收益率计算，再乘以 √252 年化；账户历史不足 30 个工作日时显示为 n\u002Fa），旁边是",[55,646,190],{},[55,648,649],{},"Z 值（概率）",[55,651,108],{},[55,653,654],{},"AHPR"," 和 ",[55,657,658],{},"GHPR","。它由我们的服务器根据存储的已平仓交易计算，保留两位小数，1.00 及以上显示绿色，0 到 1 显示黄色，低于零显示红色。在已发布的账户页面上，这个模块默认显示，不过账户主人可以隐藏它；",[100,661,663],{"href":662},"\u002Fguides\u002Ftrading-widgets-and-embeds","可嵌入的小组件","里没有这个模块。",[10,666,667],{},"关于我们自己的页面，要提醒一点：逐笔那一行标签的提示说明里，引用的仍是常规区间，即高于 1.0 算好、高于 2.0 算优秀。这些区间适用于下面的年化那一行，而不是旁边显示的逐笔数值。",[10,669,670,671,207,674,207,676,438,678,680,681,685],{},"同一模块提供",[55,672,673],{},"表格",[55,675,475],{},[55,677,478],{},[55,679,221],{},"视图；",[100,682,683],{"href":225},[55,684,228],{},"会在筛选出的子集上重新计算整个模块，可按日期范围、品种、魔术号、交易类型、交易时段、手数等筛选。把账户表现最好的品种去掉再算一次夏普比率，就能看出这个比率描述的是一套系统，还是一段插曲。",[14,687,232],{"id":232},[49,689,690,702,713,719],{},[52,691,237,692,695,696,698,699,701],{},[55,693,694],{},"高于 1.0 就是好。","”这话适用于基于周期收益的年化夏普比率，也就是",[55,697,643],{}," 那一行。普通的",[55,700,103],{},"是逐笔计算的，这个门槛对它不适用。",[52,703,237,704,707,708,712],{},[55,705,706],{},"夏普比率衡量的是风险。","”它衡量的是已平仓交易结果的离散程度，至于账户曾经跌得多深，它一无所知：那是",[100,709,711],{"href":710},"\u002Fglossary\u002Fwhat-is-drawdown","回撤","，另有单独的数字。",[52,714,237,715,718],{},[55,716,717],{},"夏普比率高，账户就安全。","”网格和摊平系统的已平仓结果之所以紧凑均匀，恰恰是因为它们不肯把亏损单平掉。浮动亏损永远进不了这个序列。",[52,720,237,721,724],{},[55,722,723],{},"我的夏普比率可以和任何账户比较。","”只有交易频率和持仓周期相当时才行：每月平仓 200 笔的剥头皮交易者和每月平仓 8 笔的波段交易者，用的根本不是同一把尺子。",[10,726,727,728,272],{},"夏普比率和账户页面上其他数字怎么对照着看，请参阅",[100,729,731],{"href":730},"\u002Fguides\u002Freading-a-trading-account-dashboard","如何读懂交易账户控制台",{"title":27,"searchDepth":274,"depth":274,"links":733},[734,735,736,737,738],{"id":16,"depth":274,"text":16},{"id":92,"depth":274,"text":92},{"id":142,"depth":274,"text":142},{"id":168,"depth":274,"text":169},{"id":232,"depth":274,"text":232},"夏普比率是平均收益除以这些收益的波动率。本文介绍公式、我们如何逐笔计算，以及为什么我们的中位数是 0.05。",{},"\u002Fzh\u002Fglossary\u002Fwhat-is-the-sharpe-ratio",[743,293,744,533],"what-is-standard-deviation-in-trading","what-is-recovery-factor",{"title":543,"description":739},"zh\u002F5.glossary\u002Fwhat-is-the-sharpe-ratio","夏普比率 (Sharpe Ratio)","rF1ahLb2_s2YlvPIZt9lNpcU3FX-JK93QxZ6RXkVz7s",{"id":750,"title":751,"body":752,"category":281,"date":282,"description":950,"draft":284,"extension":285,"meta":951,"navigation":287,"path":952,"related":953,"seo":955,"stem":956,"term":957,"updated":297,"__hash__":958},"glossary\u002Fzh\u002F5.glossary\u002Fwhat-is-the-sortino-ratio.md","什么是索提诺比率？下行风险详解",{"type":7,"value":753,"toc":943},[754,757,759,765,771,777,780,794,797,799,810,813,821,823,841,844,859,861,893,909,911,937],[10,755,756],{},"索提诺比率可以看作分母更窄的夏普比率：收益不再除以全部结果的波动率，而是只除以低于某个选定目标的那部分结果的波动率。向上的波动不算风险。有人批评说，交易者因为大赚而被扣分，扣错了地方；索提诺比率回应的正是这一点。",[14,758,16],{"id":16},[18,760,763],{"className":761,"code":762,"language":23},[21],"索提诺比率 = (平均收益 − 目标收益) \u002F 下行偏差\n",[25,764,762],{"__ignoreMap":27},[18,766,769],{"className":767,"code":768,"language":23},[21],"下行偏差 = √( Σ min(rᵢ − 目标收益, 0)² \u002F n )\n",[25,770,768],{"__ignoreMap":27},[10,772,773,774,776],{},"高于目标的观测值，对分母的贡献都是零；低于目标的，则贡献差额的平方，做法和",[100,775,183],{"href":574},"完全一样。",[10,778,779],{},"这个公式里藏着两个选择，每一个都会改变结果：",[49,781,782,788],{},[52,783,784,787],{},[55,785,786],{},"目标","。可以是零、无风险利率，或者一个要求达到的收益率。以 0% 为目标和以每年 4% 为目标算出的索提诺比率，名字一样，其实是两种不同的统计量。",[52,789,790,793],{},[55,791,792],{},"除数","。有的实现除以全部观测值的个数，有的只除以低于目标的观测值个数。后一种算出的比率总是更大、更好看。",[10,795,796],{},"这两个选择都没有统一标准，所以不注明目标和计算惯例的索提诺比率，根本没法拿来比较。这不是吹毛求疵，而是实际问题：两个平台对同一个账户报出不同的索提诺比率，而且都没算错，原因就在这里。",[14,798,92],{"id":92},[10,800,801,802,806,807,809],{},"索提诺比率的出发点没错。投资者不会把上涨 5% 的月份当成风险，把它当成风险的指标，就会错估正偏度策略的价值。比如",[100,803,805],{"href":804},"\u002Fglossary\u002Fwhat-is-trend-following","趋势跟踪","：少数几笔特大盈利撑起全部结果，同时也把",[100,808,103],{"href":102},"的分母撑大了。",[10,811,812],{},"但把向上的波动当作没有代价，只是一种建模选择，不是定律；而对一整类交易系统来说，它恰恰用反了。网格、马丁格尔和逆势加仓摊平策略，涨和跌都出自同一个机制：逆着持仓方向不断加码。用索提诺比率给它们打分，平常的小额盈利留在分子里，大部分离散程度却从分母里消失了，直到这个机制失灵的那一天。索提诺比率美化的，恰恰是风险最难从已平仓交易里看出来的那些策略。",[10,814,815,816,820],{},"稳妥的看法是：索提诺比率可以作为收益序列的第二意见，但绝不是定论，更不能取代",[100,817,819],{"href":818},"\u002Fglossary\u002Fwhat-is-maximum-drawdown","最大回撤","，最大回撤衡量的，才是投资者真正要熬过去的东西。",[14,822,142],{"id":142},[10,824,145,825,828,829,832,833,836,837,840],{},[55,826,827],{},"最深回撤的中位数为 9.7%","，但关键在分布：",[55,830,831],{},"38.5%"," 的账户跌幅从未超过 5%，而 ",[55,834,835],{},"38.2%"," 曾下跌超过 20%，",[55,838,839],{},"17.6%"," 亏掉了峰值的一半以上。",[10,842,843],{},"两个规模几乎相当的群体，分处两端。正因为是这种形态，只看下行的统计量才在同一批数据上既诱人又危险：深尾部的账户并不是左右对称地晃到那里的，而是朝一个方向一路跌下去的。任何指标要丢掉一半分布，就必须丢对那一半。",[10,845,846,847,850,851,853,854,858],{},"原始数据里本身就已经有另一种不对称。已发布账户共支付了 ",[55,848,849],{},"$862,547"," 库存费，在产生过库存费的账户中，",[55,852,405],{}," ",[100,855,857],{"href":856},"\u002Fguides\u002Fswap-and-carry-trade","净库存费为负","。还没等任何策略做决定，平均持仓成本就已经在往下拖了。",[14,860,169],{"id":168},[10,862,863,864,867,868,870,871,873,874,438,876,878,879,882,883,885,886,892],{},"哪里都没有，这一点值得说清楚。",[55,865,866],{},"ShowMyTrades 上没有任何面板、图表、小组件或导出文件显示索提诺比率","，我们的统计引擎也不计算它：界面背后并没有一个存好了、只等显示的数值。",[55,869,175],{},"模块在",[55,872,179],{},"一栏中公布",[55,875,103],{},[55,877,183],{},"；",[55,880,881],{},"账户统计","面板公布",[55,884,711],{},"（最差的净值回撤）和",[100,887,889],{"href":888},"\u002Fglossary\u002Fwhat-is-drawdown-on-balance",[55,890,891],{},"余额回撤","（只按已平仓结果计算的最差回撤）。这些就是我们页面目前提供的下行指标。",[10,894,895,896,207,899,207,902,655,905,908],{},"如果想要某个已发布账户的索提诺比率，可以自己算。账户页面的交易表格支持把已平仓交易列表下载为 CSV，每笔交易一行，包含 ",[55,897,898],{},"Profit",[55,900,901],{},"Swap",[55,903,904],{},"Commission",[55,906,907],{},"Net profit"," 列：最后一列正是我们计算夏普比率所用的序列。选好目标并写明，任何人拿到同一份文件都能算出同样的数字。这比页面上挂一个来历不明的比率要好，也是我们自己发布这个指标之前必须达到的标准。",[14,910,232],{"id":232},[49,912,913,919,925,931],{},[52,914,237,915,918],{},[55,916,917],{},"索提诺比率就是更好的夏普比率。","”它回答的是另一个问题，不是同一个问题的更好答案。它衡量相对目标的不足，夏普比率衡量离散程度，两者都不管回撤有多深。",[52,920,237,921,924],{},[55,922,923],{},"索提诺比率高，说明亏损小。","”它只说明低于目标的结果少或小。一笔还没发生的灾难性交易，在这两个比率里都看不到。",[52,926,237,927,930],{},[55,928,929],{},"不同网站的索提诺比率可以互相比较。","”只有目标和除数惯例都一致时才行，而这两项通常都不公布。",[52,932,237,933,936],{},[55,934,935],{},"ShowMyTrades 把自己的索提诺比率藏起来了。","”没什么可藏的：我们根本不计算它。用已平仓交易的导出文件，您可以自己算。",[10,938,939,940,272],{},"我们确实公布的下行指标是什么、该怎么看，请参阅",[100,941,819],{"href":942},"\u002Fguides\u002Fmaximum-drawdown-explained",{"title":27,"searchDepth":274,"depth":274,"links":944},[945,946,947,948,949],{"id":16,"depth":274,"text":16},{"id":92,"depth":274,"text":92},{"id":142,"depth":274,"text":142},{"id":168,"depth":274,"text":169},{"id":232,"depth":274,"text":232},"索提诺比率用下行偏差取代总波动率。本文介绍公式、公式里藏着的建模选择，以及我们为什么不公布这个指标。",{},"\u002Fzh\u002Fglossary\u002Fwhat-is-the-sortino-ratio",[290,743,954,744],"what-is-maximum-drawdown",{"title":751,"description":950},"zh\u002F5.glossary\u002Fwhat-is-the-sortino-ratio","索提诺比率 (Sortino Ratio)","yJBnJvTj398d0Ye0udTJCiWcMfFgnOW7pDyPZ_h4c0o",{"id":960,"title":961,"body":962,"category":281,"date":282,"description":1160,"draft":284,"extension":285,"meta":1161,"navigation":287,"path":1162,"related":1163,"seo":1165,"stem":1166,"term":1167,"updated":1168,"__hash__":1169},"glossary\u002Fzh\u002F5.glossary\u002Fwhat-is-volatility.md","什么是波动率？交易中如何衡量",{"type":7,"value":963,"toc":1153},[964,967,969,972,978,981,988,990,996,1002,1005,1007,1016,1032,1046,1048,1068,1094,1115,1117,1147],[10,965,966],{},"波动率是收益围绕自身平均值的离散程度：看的是单个结果离均值有多远，而不是朝哪个方向。它用标准差来衡量，而且是对称的：偶尔单日赚 8% 的策略，和偶尔单日亏 8% 的策略，波动率完全一样。所以波动率不等于风险。它是一个乘数，决定了某个仓位大小会变成多大的风险。",[14,968,16],{"id":16},[10,970,971],{},"取一组收益（按日、按月或按笔），求出均值，再看各个结果离均值的典型距离。",[18,973,976],{"className":974,"code":975,"language":23},[21],"σ = √( Σ (rᵢ − r̄)² \u002F (n − 1) )\n\nrᵢ = 第 i 期的收益\nr̄  = 全部 n 期的平均收益\nn  = 期数\n",[25,977,975],{"__ignoreMap":27},[10,979,980],{},"标准差的单位和输入一致：输入百分比，得到百分比；输入每笔交易的盈亏，得到账户货币。要在不同时间周期之间比较，就得按时间的平方根换算，所以用日数据算年化数值时要乘以 √252，也就是一年的交易日数。",[10,982,983,984,987],{},"实践中有两个特性很重要。第一，波动率",[55,985,986],{},"不是恒定的","，它会扎堆：平静的几周后面往往还是平静的几周，剧烈的日子则接二连三。第二，它是所有风险调整指标的分母：夏普比率这一类指标，正是拿收益除以这个数字，所以两个增长相同的策略，评价可能天差地别。",[14,989,92],{"id":92},[10,991,992,993,995],{},"离开波动率，仓位大小就没有意义。每天波动 40 点的货币对上的 1 手，和每天波动 140 点的货币对上的 1 手，标签相同，却是两笔完全不同的押注。按手数而不是按预期波动来定仓位，最容易让账户的风险状况在交易者什么都没决定的情况下悄悄走样，这也是",[100,994,495],{"href":494},"规则按承担风险的金额来写、而不是按手数来写的原因。",[10,997,998,999,1001],{},"波动率最终会变成",[100,1000,711],{"href":710},"。下跌的深度，大致等于波动率乘以风险敞口，再乘以判断错误的持续时间。把所交易品种的波动率减半，仓位不变，您要扛过的回撤也会减半，收益也跟着减半。",[10,1003,1004],{},"最后，波动率会在您不知不觉中改变。在平静的季度里仓位刚好合适的策略，波动率一翻倍就成了重仓，它不用做任何新决定，就已经变得危险。止损拉宽，点差拉宽，同样的手数，承担的风险已经是当初允许的两倍。",[14,1006,142],{"id":142},[10,1008,1009,1010,615,1012,1015],{},"在 ShowMyTrades 上有交易历史的公开账户中（2026 年 8 月），",[55,1011,103],{},[55,1013,1014],{},"0.05","。这样的收益和周围的噪音几乎分不出来：中间水平的账户，每期的平均结果和这些期间的离散程度比起来微不足道。",[10,1017,1018,1019,1021,1022,149,1025,1028,1029,1031],{},"分布的其他数字也是一样的结论：时间加权收益率中位数 ",[55,1020,618],{},"，最深回撤中位数 ",[55,1023,1024],{},"9.7%",[55,1026,1027],{},"171 笔已平仓交易","，持仓时间中位数 ",[55,1030,409],{},"。优势很小，承受的风险却是所得结果的好几倍。",[10,1033,1034,1035,1038,1039,1042,1043,1045],{},"离散程度还解释了这些账户为什么赢得多、最后却剩得少。胜率中位数是 ",[55,1036,1037],{},"68.8%","，盈利因子中位数是 ",[55,1040,1041],{},"1.28","：大多数交易以盈利平仓，而少数亏损单亏得够大，足以吃掉大部分收益。",[55,1044,622],{}," 的账户长期为正，算是多数，但领先的幅度，比背后结果的离散程度薄得多。",[14,1047,169],{"id":168},[10,1049,1050,1052,1053,1055,1056,438,1060,1062,1063,1067],{},[55,1051,175],{},"的",[55,1054,179],{},"部分，",[100,1057,1058],{"href":574},[55,1059,183],{},[55,1061,103],{},"上下相邻。这里的标准差以账户货币显示，衡量的是单笔交易结果的离散程度：标准差很大，",[100,1064,1065],{"href":107},[55,1066,108],{},"却很小，说明这个策略的结果很大程度上取决于样本里恰好有哪些交易。",[10,1069,1070,1072,1073,438,1076,1079,1080,130,1083,1086,1087,1093],{},[55,1071,881],{},"面板在",[55,1074,1075],{},"增长",[55,1077,1078],{},"绝对增长","下面两行显示",[55,1081,1082],{},"日均 %",[55,1084,1085],{},"月均 %","。这两个是平均值，所以要对照",[100,1088,1090],{"href":1089},"\u002Fguides\u002Fmonthly-returns-table-explained",[55,1091,1092],{},"月度收益","表来看，逐月的波动在那里一目了然：月均百分比相同的两个账户，每一行的数字可能天差地别。",[10,1095,1096,1099,1100,1103,1104,1106,1107,1109,1110,438,1112,1114],{},[55,1097,1098],{},"图表模块","是一个可以切换视图的面板：",[55,1101,1102],{},"逐笔交易增长","能清楚看出波动扎堆，",[55,1105,711],{},"则显示这些离散付出了多大代价。",[55,1108,175],{},"还提供",[55,1111,475],{},[55,1113,478],{},"视图，可以看出结果集中在什么时候，而那通常也是波动集中的时候。",[14,1116,232],{"id":232},[49,1118,1119,1125,1131,1137],{},[52,1120,237,1121,1124],{},[55,1122,1123],{},"波动率就是风险。","”风险是遭受您承受不起的亏损的可能性。波动率是双向的离散程度，完全不涉及您的资金规模。",[52,1126,237,1127,1130],{},[55,1128,1129],{},"波动率低就安全。","”卖出尾部风险的策略（网格、马丁格尔、不对冲的套息交易），在那次注定让它们输掉的事件到来之前，波动率读数一直很低。",[52,1132,237,1133,1136],{},[55,1134,1135],{},"夏普比率高，交易者就更优秀。","”它只说明在所测样本里，每单位离散程度换来的收益更高。只有 171 笔交易，这个估计噪音很大。",[52,1138,237,1139,1142,1143,58],{},[55,1140,1141],{},"波动率是交易品种本身的属性。","”它既取决于品种，也取决于时间段。同一个货币对一年里会几次切换市场状态，含同一种货币的货币对还会一起变化。请参阅",[100,1144,1146],{"href":1145},"\u002Fglossary\u002Fwhat-is-forex-pair-correlation","外汇货币对相关性",[10,1148,1149,1150,58],{},"波动率只有变成真实亏损时才真正要紧，这部分该怎么看，请参阅",[100,1151,1152],{"href":942},"最大回撤详解",{"title":27,"searchDepth":274,"depth":274,"links":1154},[1155,1156,1157,1158,1159],{"id":16,"depth":274,"text":16},{"id":92,"depth":274,"text":92},{"id":142,"depth":274,"text":142},{"id":168,"depth":274,"text":169},{"id":232,"depth":274,"text":232},"波动率是收益围绕平均值的离散程度，用标准差来衡量。它本身不是风险，但同样的手数意味着什么，由它决定。",{},"\u002Fzh\u002Fglossary\u002Fwhat-is-volatility",[536,535,954,1164],"what-is-forex-pair-correlation",{"title":961,"description":1160},"zh\u002F5.glossary\u002Fwhat-is-volatility","波动率 (Volatility)","2026-09-24T00:00:00.000Z","XCZ9I_SBNYpYPqBkPwcZX0S3YAivr3b6366b9-tzqnU",[1171,1172],{"path":574},{"path":288},1793898069732]