[{"data":1,"prerenderedAt":1241},["ShallowReactive",2],{"\u002Fzh\u002Fglossary\u002Fwhat-is-a-magic-number":3,"\u002Fzh\u002Fglossary\u002Fwhat-is-a-magic-number-related":254,"\u002Fzh\u002Fglossary\u002Fwhat-is-a-magic-number-langs":1238},{"id":4,"title":5,"body":6,"category":236,"date":237,"description":238,"draft":239,"extension":240,"meta":241,"navigation":242,"path":243,"related":244,"seo":249,"stem":250,"term":251,"updated":252,"__hash__":253},"glossary\u002Fzh\u002F5.glossary\u002Fwhat-is-a-magic-number.md","什么是 MT4 和 MT5 中的魔术号？",{"type":7,"value":8,"toc":228},"minimark",[9,24,28,31,42,45,68,71,74,82,85,88,99,123,127,157,191,194,221],[10,11,12,13,18,19,23],"p",{},"魔术号是 ",[14,15,17],"a",{"href":16},"\u002Fglossary\u002Fwhat-is-an-expert-advisor","EA（智能交易系统）","写进每笔订单的一个整数，",[14,20,22],{"href":21},"\u002Fglossary\u002Fwhat-is-metatrader","MetaTrader"," 会在这笔交易的整个生命周期里一直保留它。有了它，多个策略可以共用一个账户，事后照样分得清。手动下的单，魔术号是零。",[25,26,27],"h2",{"id":27},"工作原理",[10,29,30],{},"这个数字是下单函数的一个参数，由代码设定，经纪商和平台都不管。",[32,33,38],"pre",{"className":34,"code":36,"language":37},[35],"language-text","MT4:  OrderSend(symbol, cmd, lots, price, slip, sl, tp, comment, magic, ...)\nMT5:  request.magic = 55566;   → 带到仓位和每一笔成交上\n手动点击:  magic = 0\n","text",[39,40,36],"code",{"__ignoreMap":41},"",[10,43,44],{},"由此带来三个特点，每一个都容易让人踩坑。",[46,47,48,56,62],"ul",{},[49,50,51,55],"li",{},[52,53,54],"strong",{},"没有人校验它","。没有任何环节检查这个值是否唯一、有没有意义。两个 EA 设了同一个整数，在记录里就永远混在一起，而且各自都会把对方的仓位当成自己的来管理。",[49,57,58,61],{},[52,59,60],{},"零是一个实实在在的值，不是空值","。手动点击下单带的是零，魔术号参数一直没改默认值的 EA 带的也是零，两者从此就分不开了。",[49,63,64,67],{},[52,65,66],{},"事后补不上","。请求离开终端的那一刻，这个值就定死了，并随交易一起保存；当初是零的历史，永远都是零。",[10,69,70],{},"其他平台用别的字段表达同样的意思：cTrader 上的 cBot 写的是文本标签，TradeLocker 在订单上附带策略 ID，其中数字形式的 ID 在本站按魔术号读取。实际上，按策略拆分主要还是 MT4 和 MT5 的事。",[25,72,73],{"id":73},"为什么重要",[10,75,76,77,81],{},"一个账户跑三个策略，页面上每个核心数字都是三者混在一起的结果。",[14,78,80],{"href":79},"\u002Fglossary\u002Fwhat-is-a-profit-factor","盈利因子","、胜率、回撤和净值曲线都按合并后的数据计算，所以一个稳定亏钱的策略可以在盈利的总账里藏上好几个月，任何图表上都看不到它单独的那条线。账户看起来很健康，因为从总体上看，它确实健康。",[10,83,84],{},"合并之后还能把策略区分开的，只有魔术号这一个字段。设得好，历史一进来就已经按策略分成了一份份独立的业绩记录，可以互相比较，分别调整仓位，或者一个个停掉。设得马虎，事后再怎么分析也还原不出来。",[25,86,87],{"id":87},"数据显示了什么",[10,89,90,91,94,95,98],{},"这个字段在本站的分量，比大多数账户所有者想象的更重，因为每个账户页面上的自动化比例就是靠它算出来的：一笔交易只要带着非零魔术号，或者在用策略标签代替魔术号的平台上带着策略标签，就算作自动交易。在 ShowMyTrades 上有交易历史的公开账户中（2026 年 8 月），",[52,92,93],{},"自动交易占比中位数为 99%","，",[52,96,97],{},"53.9% 的账户自动化程度超过 90%","。换句话说，这些账户里几乎每一笔订单进来时都已经带好了标记。",[10,100,101,102,105,106,110,111,115,116,94,119,122],{},"之所以要尽早打标记，是因为样本量。",[52,103,104],{},"账户的已平仓交易数中位数为 171 笔","。分到三个策略上，每个只有五十七笔左右，还不足以让",[14,107,109],{"href":108},"\u002Fguides\u002Fwin-rate-is-not-an-edge","胜率","或盈利因子说明什么。从第一笔订单就打标记，每个策略才能在您必须对它做决定之前，攒够一个看得懂的样本。另一种做法是",[14,112,114],{"href":113},"\u002Fguides\u002Ftracking-multiple-accounts-portfolio","一个策略一个账户","，在本站也很常见：",[52,117,118],{},"699 位用户有不止一个账户",[52,120,121],{},"429 位在不止一家经纪商交易","。",[25,124,126],{"id":125},"在-showmytrades-上的位置","在 ShowMyTrades 上的位置",[10,128,129,132,133,136,137,140,141,144,145,148,149,152,153,156],{},[52,130,131],{},"分类统计","模块（账户设置里叫",[52,134,135],{},"货币对统计","）在",[52,138,139],{},"按品种","旁边有一个",[52,142,143],{},"按魔术号","标签页，只要账户有可以拆分的已平仓交易就会出现。每一行显示 ",[52,146,147],{},"EA \u002F 策略","、原始",[52,150,151],{},"魔术号","、交易数、盈利，以及盈利单和亏损单的数量与百分比。魔术号为 0 的那一行标为",[52,154,155],{},"手动","，也是唯一不能改名的一行：它不是一个策略，而是所有没带标记的交易。",[10,158,159,161,162,165,166,169,170,172,173,179,180,182,183,186,187,190],{},[52,160,147],{},"单元格可以直接编辑，55566 就这样变成一个半年后您还认得出的名字。这些名字之后会跟着数据走：",[52,163,164],{},"交易历史与余额变化","下的",[52,167,168],{},"已平仓交易","表格有一个",[52,171,151],{},"列，逐笔显示原始整数；",[14,174,176],{"href":175},"\u002Fguides\u002Fcustom-analysis-and-filter-presets",[52,177,178],{},"自定义分析","侧边面板有一个",[52,181,151],{},"筛选器，标签显示为",[52,184,185],{},"手动 (0)"," 或",[52,188,189],{},"标签 - 55566","。选中其中一个，页面上的所有统计都会只按这个策略重新计算。",[25,192,193],{"id":193},"常见误解",[46,195,196,203,209,215],{},[49,197,198,199,202],{},"“",[52,200,201],{},"魔术号以后再设。","”没有以后。这个字段在下单时写入，过去的交易没法重新打标记。",[49,204,198,205,208],{},[52,206,207],{},"我的 EA 会自动搞定。","”前提是参数改过。默认值为 0 的情况很常见，会悄无声息地把 EA 和您的手动交易混在一起。",[49,210,198,211,214],{},[52,212,213],{},"标为手动，就是我手动下的单。","”手动这一行装的是所有没带标记的交易：手动下的单，加上所有魔术号停在零的 EA。",[49,216,198,217,220],{},[52,218,219],{},"魔术号能告诉您这是哪个机器人。","”什么也说明不了。55566 不过是作者填进去的数字。",[10,222,223,224,122],{},"从给策略命名，到揪出那个悄悄亏钱的策略，完整流程见",[14,225,227],{"href":226},"\u002Fguides\u002Fmagic-numbers-and-strategy-labels","魔术号与策略标签",{"title":41,"searchDepth":229,"depth":229,"links":230},2,[231,232,233,234,235],{"id":27,"depth":229,"text":27},{"id":73,"depth":229,"text":73},{"id":87,"depth":229,"text":87},{"id":125,"depth":229,"text":126},{"id":193,"depth":229,"text":193},"自动化","2026-08-19T00:00:00.000Z","魔术号是 EA 给自己开的每笔订单打上的整数编号。有了它，一个账户里跑多个策略，也能逐个分开来看。",false,"md",{},true,"\u002Fzh\u002Fglossary\u002Fwhat-is-a-magic-number",[245,246,247,248],"what-is-an-expert-advisor","what-is-metatrader","what-is-a-profit-factor","what-is-backtesting",{"title":5,"description":238},"zh\u002F5.glossary\u002Fwhat-is-a-magic-number","魔术号 (Magic Number)","2026-09-04T00:00:00.000Z","6MCBx5dRlfBGfCWb6cUQEPJaeMGLLDoA4TP0bWC1lng",[255,538,733,956],{"id":256,"title":257,"body":258,"category":526,"date":237,"description":527,"draft":239,"extension":240,"meta":528,"navigation":242,"path":529,"related":530,"seo":534,"stem":535,"term":536,"updated":237,"__hash__":537},"glossary\u002Fzh\u002F5.glossary\u002Fwhat-is-a-profit-factor.md","什么是盈利因子？公式、合理数值与真实数据",{"type":7,"value":259,"toc":519},[260,263,265,271,274,277,283,286,353,355,358,366,373,375,388,395,398,405,416,418,448,474,476,512],[10,261,262],{},"盈利因子就是总盈利除以总亏损：所有盈利交易赚的钱加起来，除以所有亏损交易亏的钱。盈利因子 1.50，意思是账户每亏 $1.00，就赚回 $1.50。低于 1.00，不管胜率多高，账户从结构上就注定亏钱。",[25,264,27],{"id":27},[32,266,269],{"className":267,"code":268,"language":37},[35],"盈利因子 = 总盈利 \u002F |总亏损|\n",[39,270,268],{"__ignoreMap":41},[10,272,273],{},"注意这个公式没有考虑什么：账户交易了多久、用了多少资金、中途跌得有多深。",[10,275,276],{},"它也可以用胜率、平均盈利和平均亏损来写，这样写就能拿来诊断问题：",[32,278,281],{"className":279,"code":280,"language":37},[35],"盈利因子 = (胜率 × 平均盈利) \u002F ((1 − 胜率) × 平均亏损)\n",[39,282,280],{"__ignoreMap":41},[10,284,285],{},"两个截然不同的系统，可能算出同一个数值。胜率 30%、盈利单平均是亏损单的四倍，结果是 1.71。胜率 80%、盈利单只有亏损单的一半，结果是 2.00。两者都能赚钱，但出问题的方式不同，做起来的感受也完全两样。",[287,288,289,301],"table",{},[290,291,292],"thead",{},[293,294,295,298],"tr",{},[296,297,80],"th",{},[296,299,300],{},"解读",[302,303,304,313,321,329,337,345],"tbody",{},[293,305,306,310],{},[307,308,309],"td",{},"低于 1.00",[307,311,312],{},"亏钱，而且是结构性的，不是差一点",[293,314,315,318],{},[307,316,317],{},"1.00 – 1.10",[307,319,320],{},"还在噪声范围内，一周行情不好就没了",[293,322,323,326],{},[307,324,325],{},"1.10 – 1.30",[307,327,328],{},"优势很薄但真实存在，前提是样本够大",[293,330,331,334],{},[307,332,333],{},"1.30 – 2.00",[307,335,336],{},"策略有效",[293,338,339,342],{},[307,340,341],{},"2.00 – 3.00",[307,343,344],{},"很强。请核对交易笔数，以及成本是否算全",[293,346,347,350],{},[307,348,349],{},"高于 3.00",[307,351,352],{},"在长期历史里很少见：通常是样本太短、靠几笔超大盈利撑着，或者还有亏损单没平仓",[25,354,73],{"id":73},[10,356,357],{},"要回答“这个系统到底赚不赚钱”，盈利因子是最干净的单一指标。它也比胜率难包装得多，因为每一笔亏损都按全额计入分母。",[10,359,360,361,365],{},"但它不是风险指标。盈利因子不管顺序：一个账户一个月赚够全部利润、另外十一个月一直失血，和一个每周稳步上涨的账户，数值可以一模一样。所以要和",[14,362,364],{"href":363},"\u002Fglossary\u002Fwhat-is-maximum-drawdown","最大回撤","一起看，否则只看到了终点，没看到过程。",[10,367,368,372],{},[14,369,371],{"href":370},"\u002Fglossary\u002Fwhat-is-backtesting","回测","给出的盈利因子同样要小心：优化器追求的就是把这个比值做到最大，所以回测里的高数值，往往反映的是拟合做得多好，而不是策略本身多好。",[25,374,87],{"id":87},[10,376,377,378,94,381,387],{},"以下数字只描述在 ShowMyTrades 上发布的账户，不代表全体交易者。截至 2026 年 8 月，有交易历史的已发布账户中，",[52,379,380],{},"盈利因子中位数为 1.28",[52,382,383,386],{},[14,384,385],{"href":108},"胜率中位数","为 68.8%","。两个数字都是扣除成本后的：每笔交易先计入库存费和佣金，再判定是盈利单还是亏损单，所以 1.28 是净值，不是毛值。",[10,389,390],{},[391,392],"img",{"alt":393,"src":394},"ShowMyTrades 公开账户的中位数：68.8% 的胜率与 1.28 的盈利因子并列","\u002Fimages\u002Farticles\u002Fwin-rate-vs-payoff.svg",[10,396,397],{},"同一组数据的两个中位数：胜率看着很高，背后的优势却很薄。",[10,399,400,401,404],{},"要看的正是这两个数放在一起。中位数账户里，将近七成的交易以盈利平仓，可整体优势也只是每亏 $1.00 赚 $1.28。拿一个两项都正好处在中位数的账户，把公式倒推回去：它的平均盈利只有平均亏损的约 ",[52,402,403],{},"0.58 倍","。赢得多和赚得多是两回事，前者要容易制造得多：亏损单扛着不平，盈利单早早落袋，胜率就往上走，盈利因子却往下掉。",[10,406,407,408,411,412,415],{},"同一组账户的夏普比率中位数是 ",[52,409,410],{},"0.05","，中位数账户有 ",[52,413,414],{},"171 笔已平仓交易","。样本这么小，盈利因子高于 3.00 并不能证明系统更强，只能说明样本还太少，最差的那笔交易还没出现。",[25,417,126],{"id":125},[10,419,420,421,424,425,427,428,430,431,430,434,430,437,430,440,443,444,447],{},"每个已发布账户页面的",[52,422,423],{},"高级统计","模块里都有",[52,426,80],{},"，旁边是能解释它的几个数字：",[52,429,109],{},"、",[52,432,433],{},"平均盈利",[52,435,436],{},"平均亏损",[52,438,439],{},"期望值",[52,441,442],{},"总交易数","和",[52,445,446],{},"夏普比率","。这六个数字合起来，能说清单一比值说不清的事。",[10,449,450,451,461,462,443,465,468,469,473],{},"有两个产品细节值得注意。第一，已经算进比值的成本会另外逐项列出：交易表格有",[52,452,453,454,460],{},"盈利（毛额）",[52,455,456,457,443],{},"列，",[52,458,459],{},"库存费","佣金","各占一列；高级统计还汇总了",[52,463,464],{},"佣金合计",[52,466,467],{},"已付库存费合计","，可以看出净值数字里扣掉了多少成本。在所有已发布账户中，这两项合计分别是佣金 $4,782,670、库存费 $862,547。第二，",[14,470,471],{"href":175},[52,472,178],{},"可以先筛选（按日期范围、品种、魔术号、方向或手数），再对子集重算整个统计区块。用它可以检验：去掉表现最好的那个季度，或者去掉最赚钱的那个品种，盈利因子是否依然站得住。",[25,475,193],{"id":193},[46,477,478,489,500,506],{},[49,479,198,480,483,484,488],{},[52,481,482],{},"盈利因子高于 1，就说明我在赚钱。","”前提是成本已经算进比值。如果没有，一个毛值 1.05 的高频系统，",[14,485,487],{"href":486},"\u002Fglossary\u002Fwhat-is-cost-percentage","计入佣金和库存费之后","可能持平甚至亏损。",[49,490,198,491,494,495,499],{},[52,492,493],{},"盈利因子高，风险就低。","”它反映不了回撤、仓位大小和盈亏顺序。马丁格尔式逐级加仓的",[14,496,498],{"href":497},"\u002Fglossary\u002Fwhat-is-position-sizing","仓位管理","方案，可以一直显示 4.00，直到哪天突然失灵。",[49,501,198,502,505],{},[52,503,504],{},"不同时间周期之间可以直接比。","”不能，它没有年化。剥头皮交易者 20,000 笔交易得出的 1.20，和波段交易者 60 笔交易得出的 1.20，说的不是一回事。",[49,507,198,508,511],{},[52,509,510],{},"未平仓的交易不影响它。","”会影响，恰恰因为它们不在里面。一直不平的浮动亏损完全不计入总亏损，在平仓之前会一直把比值抬高。",[10,513,514,515,122],{},"该长期追踪哪些指标、按什么顺序看，请参阅",[14,516,518],{"href":517},"\u002Fguides\u002Ftracking-trading-performance","交易表现追踪完全指南",{"title":41,"searchDepth":229,"depth":229,"links":520},[521,522,523,524,525],{"id":27,"depth":229,"text":27},{"id":73,"depth":229,"text":73},{"id":87,"depth":229,"text":87},{"id":125,"depth":229,"text":126},{"id":193,"depth":229,"text":193},"指标","盈利因子是总盈利除以总亏损，低于 1.0 的账户注定亏损。本文给出各数值区间的解读，以及数千个账户的中位数 1.28。",{},"\u002Fzh\u002Fglossary\u002Fwhat-is-a-profit-factor",[531,532,533,248],"what-is-drawdown","what-is-maximum-drawdown","what-is-position-sizing",{"title":257,"description":527},"zh\u002F5.glossary\u002Fwhat-is-a-profit-factor","盈利因子 (Profit Factor)","OLz2UdKZBQ-s3NM0RQISYNe6Jrhtmy0tM84z39S0Thk",{"id":539,"title":540,"body":541,"category":236,"date":237,"description":723,"draft":239,"extension":240,"meta":724,"navigation":242,"path":725,"related":726,"seo":729,"stem":730,"term":731,"updated":252,"__hash__":732},"glossary\u002Fzh\u002F5.glossary\u002Fwhat-is-an-expert-advisor.md","什么是 EA（智能交易系统）？MetaTrader 中的 EA 详解",{"type":7,"value":542,"toc":716},[543,549,551,554,560,563,575,582,584,587,598,600,611,627,630,632,640,667,681,683,709],[10,544,545,546,548],{},"EA（智能交易系统，Expert Advisor）是用 MetaQuotes 的 MQL 语言写成的程序，运行在 ",[14,547,22],{"href":21}," 终端里，没人守在键盘前也能开仓、改单和平仓。它挂在一张图表上，随这张图表收到的每个价格跳动做出反应，按编译进去的规则行事。EA 只是执行策略的一种手段，并不意味着策略本身管用。",[25,550,27],{"id":27},[10,552,553],{},"代码编译好，拖到图表上，之后行情每动一下，终端就调用它一次。",[32,555,558],{"className":556,"code":557,"language":37},[35],"价格跳动（tick）到达图表\n  → EA 的 tick 处理函数运行\n  → 根据价格、指标和持仓评估规则\n  → 发送订单，并打上魔术号\n  → 经纪商的交易服务器接受、拒绝或重新排队处理\n",[39,559,557],{"__ignoreMap":41},[10,561,562],{},"三个条件缺一不可：终端在运行且已登录，允许自动交易，经纪商接受请求。少了任何一个，EA 照样悄悄运行，却一单也下不出去。",[10,564,565,566,569,570,574],{},"关键在于，EA 运行在",[52,567,568],{},"客户端","：它寄居在一个程序里，而这个程序跑在您自己的机器上。经纪商的服务器上，除了已经挂在持仓上的止损或止盈，没有它的任何东西。终端一关，代码也就不在了。所以无人值守的策略应该放在 ",[14,571,573],{"href":572},"\u002Fglossary\u002Fwhat-is-a-vps","VPS"," 上，而不是一台会休眠的笔记本电脑上。",[10,576,577,578,581],{},"EA 发出的每笔订单都带有",[14,579,151],{"href":580},"\u002Fglossary\u002Fwhat-is-a-magic-number","。有了这个整数，多个 EA 可以共用一个账户，事后照样分得清。",[25,583,73],{"id":73},[10,585,586],{},"EA 能消除犹豫、疲劳和凌晨 3 点的难题，但也消除了本该把它关掉的那份判断。它不知道自己的优势早已失效；一套坏掉的规则，它执行起来和好规则一样一丝不苟。",[10,588,589,590,594,595,597],{},"EA 能给的好处，是执行始终如一。可一致不等于盈利，大部分自动化交易的钱，就亏在这两者之间。在测试器里完美无瑕的策略，通常是对测试样本做了拟合，参见",[14,591,593],{"href":592},"\u002Fglossary\u002Fwhat-is-curve-fitting","曲线拟合","。所以与经纪商同步的账户，比任何",[14,596,371],{"href":370},"报告都更有说服力。测试器展示的是代码在历史数据上做了什么；账户展示的是代码、经纪商和机器一起，对真金白银做了什么。",[25,599,87],{"id":87},[10,601,602,603,94,605,94,607,610],{},"在本站，自动化是常态，不是小众。在 ShowMyTrades 上有交易历史的公开账户中（2026 年 8 月），",[52,604,93],{},[52,606,97],{},[52,608,609],{},"42.2% 低于 10%","。账户分成两堆，中间几乎是空的：要么全靠代码，要么全靠手动。",[10,612,613,614,94,619,622,623,626],{},"自动化并没有让结果变得整齐。同一批账户中，",[52,615,616,618],{},[14,617,80],{"href":79},"中位数为 1.28",[52,620,621],{},"最深回撤中位数为 9.7%","，左尾还拖得很长：",[52,624,625],{},"38.2% 的账户曾回撤超过 20%，17.6% 超过 50%","。用代码跑策略，改变的是谁来按按钮，而不是策略本身值多少。",[10,628,629],{},"我们守着一条界线：从不推测某个账户上跑的是哪款商业 EA。同一款产品，每个买家装上去用的魔术号都不一样，经纪商数据里没有任何信息能认出是哪个软件；把猜测当成事实发布，还不如不回答。",[25,631,126],{"id":125},[10,633,634,635,639],{},"账户头部有一枚显示自动化比例的徽章，例如 ",[636,637,638],"em",{},"99% autotrading","。这不是自己声明的：一笔交易只要带着非零魔术号进来，就算作自动交易，所以这枚徽章统计的是 EA 在订单流上留下的指纹。",[10,641,642,643,646,647,650,651,654,655,658,659,662,663,666],{},"在 EA 连接下，",[52,644,645],{},"终端与已打开的图表","模块（MT4 上叫",[52,648,649],{},"终端","，因为 MT4 的 EA 不上报图表列表）会显示代码所在的运行环境，其中",[52,652,653],{},"自动交易","一项显示",[52,656,657],{},"允许","或",[52,660,661],{},"禁止","：正是这个开关，会让一个本来好好的 EA 悄无声息地一单也下不了。卡片下方的已打开图表表格，会列出每张图表上加载的",[52,664,665],{},"智能交易系统","，以及它的品种、时间周期和指标：看到的是实际在运行的东西，而不是别人声称的东西。仅用密码的直接连接从不接触您的机器，所以不会显示这些信息。",[10,668,669,670,672,673,675,676,680],{},"按策略的结果在",[52,671,131],{},"模块的",[52,674,143],{},"标签页里，每个 EA 单独一行，显示交易数、盈利以及盈利单和亏损单的拆分。",[14,677,678],{"href":175},[52,679,178],{},"侧边面板可以把页面上的所有统计缩小到单个魔术号，单独看一个 EA。",[25,682,193],{"id":193},[46,684,685,691,697,703],{},[49,686,198,687,690],{},[52,688,689],{},"EA 就是策略。","”EA 只是执行者。两个 EA 实现同一套规则，就会做出同样的交易；真正有价值的是规则。",[49,692,198,693,696],{},[52,694,695],{},"回测漂亮，EA 就好。","”这只说明代码和历史数据对得上。这是最起码的要求，不是证据。",[49,698,198,699,702],{},[52,700,701],{},"自动交易开着，它就在交易。","”允许不等于在做：自动交易（AutoTrading）按钮关着时，EA 照样加载运行，只是发不出订单。不少账户就是这样一连几周没有动静。",[49,704,198,705,708],{},[52,706,707],{},"电脑关了它也在跑。","”不会。终端停了，只有保存在服务器上的止损和止盈还在。",[10,710,711,712,122],{},"真实的自动化记录和人为做出来的记录，有哪些具体的分辨迹象，请看",[14,713,715],{"href":714},"\u002Fguides\u002Fspotting-fake-ea-results","如何识破虚假的 EA 业绩",{"title":41,"searchDepth":229,"depth":229,"links":717},[718,719,720,721,722],{"id":27,"depth":229,"text":27},{"id":73,"depth":229,"text":73},{"id":87,"depth":229,"text":87},{"id":125,"depth":229,"text":126},{"id":193,"depth":229,"text":193},"EA（智能交易系统）是在 MetaTrader 里替您交易的程序。它能做什么、做不到什么，本站发布的账户自动化程度又有多高。",{},"\u002Fzh\u002Fglossary\u002Fwhat-is-an-expert-advisor",[246,727,728,248],"what-is-a-magic-number","what-is-a-vps",{"title":540,"description":723},"zh\u002F5.glossary\u002Fwhat-is-an-expert-advisor","智能交易系统 (Expert Advisor)","0dEs_0iks9JdASudwXZIijWJ-6M8sh5aGgFY31uFZRo",{"id":734,"title":735,"body":736,"category":944,"date":237,"description":945,"draft":239,"extension":240,"meta":946,"navigation":242,"path":947,"related":948,"seo":952,"stem":953,"term":954,"updated":237,"__hash__":955},"glossary\u002Fzh\u002F5.glossary\u002Fwhat-is-backtesting.md","什么是回测？定义、局限与实盘数据",{"type":7,"value":737,"toc":937},[738,741,743,746,752,755,785,791,793,796,799,801,812,831,842,848,851,853,861,886,896,898,929],[10,739,740],{},"回测是用历史价格数据模拟交易策略，估算它在过去会有怎样的表现。一张订单都不用发到经纪商，就能得出净值曲线、盈利因子和回撤。这些结果只是对策略的一种假设，并不是策略的真实记录。",[25,742,27],{"id":27},[10,744,745],{},"测试器逐根回放历史 K 线或 tick，在每一步套用策略的入场和离场规则，记下一次模拟成交。模拟和现实之间的所有差距，都藏在成本模型和成交模型里：",[32,747,750],{"className":748,"code":749,"language":37},[35],"每笔交易净结果 = 毛盈亏 − 点差 − 佣金 − 库存费 − 滑点\n\n点差成本 = 点差（价格单位）× 合约规模 × 手数\n库存费   = 库存费率 × 手数 × 持仓过夜数\n",[39,751,749],{"__ignoreMap":41},[10,753,754],{},"输出靠不靠得住，取决于三项设置：",[46,756,757,763,774],{},[49,758,759,762],{},[52,760,761],{},"建模质量","。MT4 和 MT5 用百分比表示。如果不提供真实 tick 数据，“每个 tick”（Every tick）模式其实是用 1 分钟 K 线插值出来的；“控制点”（control points）和“仅开盘价”（open prices only）都是近似算法，对依赖 K 线内部走势的策略美化得很厉害。",[49,764,765,768,769,773],{},[52,766,767],{},"点差假设","。多数测试器默认用固定点差。实盘点差会在每日展期（rollover）、新闻公布和时段开盘时扩大，而很多系统恰恰在这些时候交易。按 0.8 点建模、实际却要付 4.0 点的",[14,770,772],{"href":771},"\u002Fglossary\u002Fwhat-is-a-spread","点差","，已经是另一个策略了。",[49,775,776,779,780,784],{},[52,777,778],{},"滑点与成交逻辑","。测试器默认订单都按请求价成交。",[14,781,783],{"href":782},"\u002Fglossary\u002Fwhat-is-slippage","滑点","、拒单、重新报价和部分成交，在测试器里统统不存在。",[10,786,787,788,790],{},"还有第四个问题，软件不会告诉您：这组参数本身就是从一轮搜索里筛出来的。如果您跑了 400 个变体，只留下最好的那个，眼前的结果就是事后挑出来的。这就是",[14,789,593],{"href":592},"，也是测试表现和实盘表现之间最大的一道鸿沟。",[25,792,73],{"id":73},[10,794,795],{},"要否决一个坏主意，回测是最便宜的办法；要确认一个好主意，它却是最昂贵的办法。它能说明一套规则在机制上前后一致，扛得住粗略的成本假设，回撤幅度也在您忍受得了的范围内。但它证明不了优势真的存在，因为它用到的每一项输入都是重建出来的，每一个参数都是看过数据之后才定下来的。",[10,797,798],{},"决定资金投向的人，应该把回测当成一个有一票否决权的筛选工具，而永远不要把它当作证据。",[25,800,87],{"id":87},[10,802,803,804,807,808,811],{},"ShowMyTrades 以公开页面发布了 ",[52,805,806],{},"1 万+ 个账户","，背后是 ",[52,809,810],{},"15,436,464 笔已同步的交易","（2026 年 8 月）。没有一笔是模拟的：每一笔都由经纪商成交，再从经纪商那里读取回来。",[10,813,814,815,818,819,821,822,94,825,821,828,830],{},"这些已发布账户绝大多数是自动交易：自动交易占比的中位数为 ",[52,816,817],{},"99%","，53.9% 的账户自动化程度超过 90%，42.2% 低于 10%。自动交易的那部分，主要是投入资金之前做过回测的策略。在全部已发布账户中，",[14,820,80],{"href":79},"中位数为 ",[52,823,824],{},"1.28",[14,826,446],{"href":827},"\u002Fglossary\u002Fwhat-is-the-sharpe-ratio",[52,829,410],{},"。测试报告里的数字，真正要比的就是这些。",[10,832,833,834,837,838,841],{},"模拟最容易低估的是成本，而这里的成本是实际扣掉的，不是假设出来的：佣金 ",[52,835,836],{},"$4,782,670","，库存费 ",[52,839,840],{},"$862,547","；在产生过库存费的账户中，86.3% 的净库存费为负。策略要隔夜持仓，回测却把库存费设为零，这不是稍微乐观了一点，而是漏掉了一笔反复发生的费用。",[10,843,844],{},[391,845],{"alt":846,"src":847},"点差、佣金和库存费是三项独立的成本：ShowMyTrades 公开账户共被收取 $4,782,670 的佣金和 $862,547 的库存费","\u002Fimages\u002Farticles\u002Fwhere-costs-hide.svg",[10,849,850],{},"测试器用固定点差、又不算库存费，三项成本就漏掉了两项。",[25,852,126],{"id":125},[10,854,855,856,860],{},"回测有自己单独的区域，从不和账户混在一起。您可以上传 MetaTrader 4、MetaTrader 5 或 cTrader 的策略测试器报告，发布成标有 BACKTEST 的页面。页面会检查有没有篡改迹象，并单独列在您交易者主页的一个标签页里。回测页面从不带验证徽章，也不会出现在小组件、探索、投资组合和 Developer API 中，因为模拟结果不是",[14,857,859],{"href":858},"\u002Fglossary\u002Fwhat-is-a-track-record","业绩记录","。账户页面则不同，每一个都是从经纪商只读同步来的，所以上面的数字描述的是真实执行的交易，而不是模拟出来的交易。",[10,862,863,864,443,866,868,869,871,872,875,876,879,880,882,883,885],{},"回测藏起来的东西，这几个模块能揭出来：高级统计模块里的",[52,865,464],{},[52,867,467],{},"，以账户货币显示策略的真实成本；",[52,870,649],{},"面板，报告与交易服务器之间的 ",[52,873,874],{},"Ping"," 和",[52,877,878],{},"重传","，也就是您实际成交时所处的条件；还有",[52,881,164],{},"，它的",[52,884,168],{},"标签页按真实成交价列出每一笔已执行的订单，而不是测试器会给的价格。",[10,887,888,891,892,895],{},[52,889,890],{},"业绩记录已验证","徽章（65 个账户持有）和",[52,893,894],{},"交易权限已验证","徽章（264 个账户持有）标出的，是从经纪商到页面这条链路确实经过核查、而不只是自己声称的记录。",[25,897,193],{"id":193},[46,899,900,906,917,923],{},[49,901,198,902,905],{},[52,903,904],{},"它用了每个 tick，所以很准确。","”tick 建模改善的是价格路径，不是成交。经纪商的点差、执行和拒单，仍然没有算进去。",[49,907,198,908,911,912,916],{},[52,909,910],{},"它测试了十年。","”",[14,913,915],{"href":914},"\u002Fguides\u002Fhow-much-history-a-track-record-needs","时长不等于独立性","。如果这十年里只有一种占主导的市场状态，那它只是一次观察，而不是 2,500 次。",[49,918,198,919,922],{},[52,920,921],{},"只优化了三个参数。","”三个参数各取二十个值，就是 8,000 次实验。8,000 次实验里的赢家，从构造上就注定有一部分靠的是运气。",[49,924,198,925,928],{},[52,926,927],{},"模拟账户的前向测试已经证实了它。","”模拟服务器的成交偏乐观，也从不拒单。模拟账户的结果只是第二次模拟，算不上第一次实盘测试。",[10,930,931,932,936],{},"无论是测试结果还是实盘数字，在相信它之前，先按",[14,933,935],{"href":934},"\u002Fguides\u002Fverifying-trading-performance-claims","如何验证交易业绩宣传","逐项核对一遍。",{"title":41,"searchDepth":229,"depth":229,"links":938},[939,940,941,942,943],{"id":27,"depth":229,"text":27},{"id":73,"depth":229,"text":73},{"id":87,"depth":229,"text":87},{"id":125,"depth":229,"text":126},{"id":193,"depth":229,"text":193},"策略","回测就是拿历史价格数据模拟交易策略。它能证明什么、证明不了什么，以及为什么与经纪商同步的实盘结果几乎总会和它不一样。",{},"\u002Fzh\u002Fglossary\u002Fwhat-is-backtesting",[949,950,951],"what-is-curve-fitting","what-is-a-verified-track-record","what-is-slippage",{"title":735,"description":945},"zh\u002F5.glossary\u002Fwhat-is-backtesting","回测 (Backtesting)","JFA3C0OuDmaFAnqzJS-I02fA4QcPMlLuq0a6Gjsakv8",{"id":957,"title":958,"body":959,"category":1227,"date":237,"description":1228,"draft":239,"extension":240,"meta":1229,"navigation":242,"path":1230,"related":1231,"seo":1234,"stem":1235,"term":1236,"updated":252,"__hash__":1237},"glossary\u002Fzh\u002F5.glossary\u002Fwhat-is-metatrader.md","什么是 MetaTrader？MT4 与 MT5 的区别详解",{"type":7,"value":960,"toc":1220},[961,964,966,969,975,983,986,1057,1063,1065,1068,1074,1076,1104,1111,1113,1128,1146,1155,1185,1187,1213],[10,962,963],{},"MetaTrader 是 MetaQuotes 开发的交易平台，授权给经纪商使用：经纪商自己运行服务器，再把终端交给客户安装。目前仍在广泛使用的有两个版本：2005 年发布的 MetaTrader 4 和 2010 年发布的 MetaTrader 5。MetaTrader 不是经纪商，也不保管资金，它只是一层软件，一头是您，另一头是给您开立登录号的经纪商。",[25,965,27],{"id":27},[10,967,968],{},"任何一套 MetaTrader 都由三部分组成。大多数客服问题，都出在把这三部分搞混了。",[32,970,973],{"className":971,"code":972,"language":37},[35],"终端（您的电脑或 VPS） ← 图表、EA、客户端逻辑\n    ↕ 往返时间即 ping\n交易服务器（您的经纪商） ← 余额、订单撮合、保存在服务器端的止损\u002F止盈\nMQL（MQL4 或 MQL5） ← 编写 EA 和指标所用的语言\n",[39,974,972],{"__ignoreMap":41},[10,976,977,978,982],{},"登录需要账号、服务器名称和密码。同一个账户有两个密码：主密码可以交易；",[14,979,981],{"href":980},"\u002Fglossary\u002Fwhat-is-an-mt4-investor-password","投资者密码","只读，是第三方验证的标准凭据。",[10,984,985],{},"两个版本的实际区别如下：",[287,987,988,1000],{},[290,989,990],{},[293,991,992,994,997],{},[296,993],{},[296,995,996],{},"MT4",[296,998,999],{},"MT5",[302,1001,1002,1013,1024,1035,1046],{},[293,1003,1004,1007,1010],{},[307,1005,1006],{},"编程语言",[307,1008,1009],{},"MQL4",[307,1011,1012],{},"MQL5",[293,1014,1015,1018,1021],{},[307,1016,1017],{},"记账方式",[307,1019,1020],{},"基于订单：一个订单就是一个仓位",[307,1022,1023],{},"基于成交和仓位，轧差或对冲模式",[293,1025,1026,1029,1032],{},[307,1027,1028],{},"时间周期",[307,1030,1031],{},"9",[307,1033,1034],{},"21",[293,1036,1037,1040,1043],{},[307,1038,1039],{},"市场",[307,1041,1042],{},"外汇和差价合约（CFD）",[307,1044,1045],{},"增加了交易所交易品种和市场深度",[293,1047,1048,1051,1054],{},[307,1049,1050],{},"策略测试器",[307,1052,1053],{},"单品种、单线程",[307,1055,1056],{},"多品种、多线程、真实 tick 数据",[10,1058,1059,1060,1062],{},"从实际使用看，它们根本不是同一产品的两个版本。MQL4 代码不能当作 MQL5 编译，",[14,1061,17],{"href":16},"不能互相移植，账户历史也搬不过去。要迁移，只能新开账户。",[25,1064,73],{"id":73},[10,1066,1067],{},"不管用哪个版本，终端都只是一扇窗。成交、点差、库存费、止损触发，都由经纪商的交易服务器决定；同一版本号的 MT5，换一家经纪商，结果就不一样。",[10,1069,1070,1071,1073],{},"MetaTrader 真正给您的，也是它二十年来一直是默认选择的原因，是一份机器可读的记录，第三方可以直接从经纪商那里读取。公开",[14,1072,859],{"href":858},"的基础就在这里：只读凭据、开放的数据通道，中间没有任何交易者能动手编辑的文件。",[25,1075,87],{"id":87},[10,1077,1078,1079,1082,1083,1086,1087,1091,1092,1096,1097,1100,1101,122],{},"我们看到的市场，几乎全是 MetaTrader。截至 2026 年 8 月，在",[52,1080,1081],{},"连接到 ShowMyTrades 的 1 万+ 个账户","中，绝大多数用的是 ",[52,1084,1085],{},"MT4 或 MT5","，其余为 ",[14,1088,1090],{"href":1089},"\u002Fglossary\u002Fwhat-is-ctrader","cTrader"," 和 ",[14,1093,1095],{"href":1094},"\u002Fglossary\u002Fwhat-is-tradelocker","TradeLocker","。这些账户分布在 ",[52,1098,1099],{},"703 个不同的经纪商服务器","上，累计同步了 ",[52,1102,1103],{},"15,436,464 笔交易",[10,1105,1106,1107,1110],{},"在这里，MT5 账户是 MT4 的四倍多。这反映的是 2026 年来到本平台的账户构成，并不是在评判哪个版本更好；不少长期运行的 MT4 记录，放在页面上依然是更好的业绩记录。在 ShowMyTrades 上有交易历史的公开账户中（2026 年 8 月），",[52,1108,1109],{},"自动交易占比的中位数为 99%","。能做到这一点，是因为两个版本都原生运行代码，不需要桥接，不需要 API 密钥，也不需要任何第三方中转。",[25,1112,126],{"id":125},[10,1114,1115,1116,430,1119,430,1122,1124,1125,1127],{},"账户头部用徽章显示平台：",[52,1117,1118],{},"MetaTrader 4",[52,1120,1121],{},"MetaTrader 5",[52,1123,1090],{}," 或 ",[52,1126,1095],{},"，旁边是经纪商、账户币种、杠杆和自动化占比。",[10,1129,1130,1131,1091,1134,1137,1138,1141,1142,1145],{},"添加账户时，连接方式选择器直接点明了区别：",[52,1132,1133],{},"MetaTrader 4 + EA",[52,1135,1136],{},"MetaTrader 5 + EA"," 标为推荐，因为更新频率高得多；",[52,1139,1140],{},"MetaTrader 4 直接连接","和 ",[52,1143,1144],{},"MetaTrader 5 直接连接","则用投资者密码从经纪商同步，终端里什么都不用装。",[10,1147,1148,1149,1151,1152,1154],{},"这个选择决定了页面能显示什么，两个版本都一样。只有 EA 连接才会生成",[52,1150,645],{},"模块：终端在这里报告所在主机和网络连接的情况，并列出当前加载的图表和 EA。在 MT4 上，这个模块的标题只是",[52,1153,649],{},"，因为 MT4 的 EA 根本不报告图表列表：与其声称没有打开图表，我们宁可什么都不显示。",[10,1156,1157,1158,1163,1164,1167,1168,94,1171,122,1174,1177,1178,1181,1182,1184],{},"验证与版本无关。账户头部有一枚",[14,1159,1161],{"href":1160},"\u002Fglossary\u002Fwhat-is-a-verified-track-record",[52,1162,859],{},"徽章，存入只读投资者密码后变绿；还有一枚",[52,1165,1166],{},"交易权限","徽章，在您下一笔挂单、并在注释字段填入您的 ShowMyTrades 用户 ID 后变绿。在所有已连接的账户中，",[52,1169,1170],{},"65 个获得业绩记录已验证",[52,1172,1173],{},"264 个获得交易权限已验证",[52,1175,1176],{},"分类统计 → 按魔术号","里的",[14,1179,1180],{"href":226},"按策略报告","只有 MT4 和 MT5 才有，因为",[14,1183,151],{"href":580},"字段本来就是 MetaTrader 的概念。",[25,1186,193],{"id":193},[46,1188,1189,1195,1201,1207],{},[49,1190,198,1191,1194],{},[52,1192,1193],{},"MetaTrader 就是我的经纪商。","”它只是经纪商获得授权使用的软件。入金、出金、点差和执行都归经纪商管，投诉也要找他们。",[49,1196,198,1197,1200],{},[52,1198,1199],{},"MT5 就是功能更多的 MT4。","”MT5 是重写的。EA、指标和脚本都得重新开发，所以很多运行良好的系统一直没有迁移。",[49,1202,198,1203,1206],{},[52,1204,1205],{},"MT5 不能对冲。","”它既支持轧差，也支持对冲；用哪一种由经纪商在开户时设定，之后不能更改。",[49,1208,198,1209,1212],{},[52,1210,1211],{},"成交由平台决定。","”决定成交的是交易服务器。同一版本号的两个终端，接在两家经纪商上，同一笔订单的成交不会一样。",[10,1214,1215,1216,122],{},"想把 MetaTrader 登录号变成别人能核查的记录，请看：",[14,1217,1219],{"href":1218},"\u002Fguides\u002Fhow-to-verify-your-account","如何验证您的交易账户",{"title":41,"searchDepth":229,"depth":229,"links":1221},[1222,1223,1224,1225,1226],{"id":27,"depth":229,"text":27},{"id":73,"depth":229,"text":73},{"id":87,"depth":229,"text":87},{"id":125,"depth":229,"text":126},{"id":193,"depth":229,"text":193},"平台","大多数零售外汇账户都在 MetaTrader 上运行。本文讲清 MT4 与 MT5 的区别、投资者密码的作用，以及我们看到的真实平台分布。",{},"\u002Fzh\u002Fglossary\u002Fwhat-is-metatrader",[1232,245,727,728,1233],"what-is-an-mt4-investor-password","what-is-ctrader",{"title":958,"description":1228},"zh\u002F5.glossary\u002Fwhat-is-metatrader","MetaTrader (MT4\u002FMT5 交易平台)","1x5iUHpZOMxeiYe_TqtlAnu0zIzsRu9Fpa3t-E3DgzQ",[1239,1240],{"path":580},{"path":243},1793898059384]