[{"data":1,"prerenderedAt":1151},["ShallowReactive",2],{"\u002Fzh\u002Fglossary\u002Fwhat-is-a-demo-vs-live-account":3,"\u002Fzh\u002Fglossary\u002Fwhat-is-a-demo-vs-live-account-related":288,"\u002Fzh\u002Fglossary\u002Fwhat-is-a-demo-vs-live-account-langs":1147},{"id":4,"title":5,"body":6,"category":270,"date":271,"description":272,"draft":273,"extension":274,"meta":275,"navigation":276,"path":277,"related":278,"seo":283,"stem":284,"term":285,"updated":286,"__hash__":287},"glossary\u002Fzh\u002F5.glossary\u002Fwhat-is-a-demo-vs-live-account.md","什么是模拟账户？模拟账户与真实账户详解",{"type":7,"value":8,"toc":262},"minimark",[9,13,17,20,71,82,85,88,91,99,107,110,128,131,142,146,149,219,222,255],[10,11,12],"p",{},"模拟账户用模拟资金、按经纪商的报价交易；真实账户用真金白银，面对的是真实的流动性。两者的图表一模一样，差距也正因此被低估：变的不是您看到的价格，而是实际成交的价格、被收的费用，以及判断失误的代价。",[14,15,16],"h2",{"id":16},"工作原理",[10,18,19],{},"模拟服务器收到订单后，直接按报价确认成交。没有对手方，订单簿里没有头寸，也没有资金承担风险，所以真实成交中的不少约束，在这里根本不存在。",[21,22,23,37,43,54,65],"ul",{},[24,25,26,30,31,36],"li",{},[27,28,29],"strong",{},"没有深度，没有排队","。不管下多大的单，模拟账户都按报价一次全部成交。真实账户上，订单要一层层吃掉可用深度，剩下的部分",[32,33,35],"a",{"href":34},"\u002Fglossary\u002Fwhat-is-slippage","成交价更差","，或者只能部分成交。",[24,38,39,42],{},[27,40,41],{},"没有重新报价，没有拒单","。模拟服务器很少拒单。真实服务器会，而且往往就在最关键的那几秒。",[24,44,45,48,49,53],{},[27,46,47],{},"理想化的点差","。模拟报价源用的常常是典型",[32,50,52],{"href":51},"\u002Fglossary\u002Fwhat-is-a-spread","点差","，而不是实时点差，所以数据发布前后和每日展期时的点差扩大会被削弱，甚至看不到。",[24,55,56,59,60,64],{},[27,57,58],{},"成本靠模拟，不真收","。模拟账户上的",[32,61,63],{"href":62},"\u002Fglossary\u002Fwhat-is-a-swap-in-forex","库存费","可能沿用过期的费率表，甚至干脆没有。真实账户的库存费按交易品种设定，会调整，展期日还要收三倍。",[24,66,67,70],{},[27,68,69],{},"无限次重置","。模拟账户爆仓了，点一下就能重新充满，曲线上不留痕迹。真实账户爆仓，就原原本本写在曲线上。",[72,73,78],"pre",{"className":74,"code":76,"language":77},[75],"language-text","真实结果 = 模拟结果\n         - 入场和出场的滑点\n         - 被拒绝或因重新报价而错过的入场\n         - 真实的点差扩大和真实的库存费\n         - 因为是自己的钱而没有做的交易\n","text",[79,80,76],"code",{"__ignoreMap":81},"",[10,83,84],{},"最后一行不是玩笑。钱是自己的，确定仓位大小、扛住回撤、照计划执行，就完全是另一回事了。",[14,86,87],{"id":87},"为什么重要",[10,89,90],{},"上面每一项差异都指向同一个方向。模拟账户不是给结果加了随机噪声，而是去掉了摩擦，而且去掉的偏偏是摩擦最贵的那几处：入场、止损和展期。所以模拟账户的业绩记录只能算上限，不能当估计值。",[10,92,93,94,98],{},"交易越频繁，影响越大；目标幅度越大，影响越小，这和",[32,95,97],{"href":96},"\u002Fglossary\u002Fwhat-is-backtesting","回测","背后的算术是一样的。本平台上单笔交易的持仓时间中位数是 2.4 小时，短到执行成本在结果里占了实实在在的份额，而不是可以四舍五入掉的零头。",[10,100,101,102,106],{},"但这并不是说模拟账户没用。检查 EA 能不能跑、",[32,103,105],{"href":104},"\u002Fglossary\u002Fwhat-is-a-magic-number","魔术号","有没有设好、手数算得对不对，模拟账户正合适。它检验的是系统能不能运作，而不是能不能赚钱。",[14,108,109],{"id":109},"数据显示了什么",[10,111,112,113,116,117,116,120,123,124,127],{},"在 ShowMyTrades 上有交易历史的公开账户中（2026 年 8 月），中位数账户的",[27,114,115],{},"时间加权收益率为 +3.2%","，",[27,118,119],{},"最深回撤为 9.7%",[27,121,122],{},"盈利因子中位数为 1.28","。",[27,125,126],{},"63.0%"," 的账户长期收益为正。",[10,129,130],{},"这些数字里包含真实账户，体现的正是大规模的真实摩擦。同一批账户如果没有摩擦，每一项都会更好看，因为造成这些数字的成本，在模拟服务器上都不会被完整收取。",[10,132,133,134,137,138,141],{},"其中最大的几项成本是实打实收取的，不是模拟出来的：这批账户共付了 ",[27,135,136],{},"$4,782,670"," 佣金和 ",[27,139,140],{},"$862,547"," 库存费。模拟服务器可以把这两个数字显示出来，却一分钱也不用收。",[14,143,145],{"id":144},"在-showmytrades-上的位置","在 ShowMyTrades 上的位置",[10,147,148],{},"值得一提的是，这一项不是自己声明的。",[21,150,151,166,182,188],{},[24,152,153,154,157,158,161,162,165],{},"每个已发布账户的账户头部都有一行",[27,155,156],{},"认证","。第一个徽章写着",[27,159,160],{},"真实账户","或",[27,163,164],{},"模拟账户","，依据是经纪商服务器报告的交易模式，而不是所有者自己填的字段。服务器名称里带“demo”的，不管还报告了什么，一律判定为模拟账户。鼠标悬停在徽章上，会说明这个状态对下方的数字意味着什么。",[24,167,168,169,172,173,176,177,181],{},"旁边是",[27,170,171],{},"业绩记录","和",[27,174,175],{},"交易权限","两个徽章，标明历史记录和账户所有权是否经过",[32,178,180],{"href":179},"\u002Fguides\u002Fthird-party-verification-explained","独立佐证","。没有经过佐证的，会显示警告图标。",[24,183,184,187],{},[27,185,186],{},"信息","一行列出平台、经纪商（提示框里有交易服务器名称）、账户货币和杠杆，也就是这份结果所处的交易环境。",[24,189,190,193,194,197,198,201,202,201,205,201,208,201,211,214,215,218],{},[27,191,192],{},"终端","面板（MT5 上为",[27,195,196],{},"终端与已打开的图表","）显示 ",[27,199,200],{},"Ping","、",[27,203,204],{},"重传",[27,206,207],{},"终端版本号",[27,209,210],{},"内存",[27,212,213],{},"自动交易","和 ",[27,216,217],{},"DLL 导入","，也就是这些成交背后的实时连接状况。",[14,220,221],{"id":221},"常见误解",[223,224,225,232,238,244],"ol",{},[24,226,227,228,231],{},"“",[27,229,230],{},"在模拟账户上做了六个月，我已经准备好了。","”这只说明流程跑得通。成交情况，以及真实回撤下您会怎么做，都还没经过检验。",[24,233,227,234,237],{},[27,235,236],{},"模拟账户的价格是假的。","”报价源通常是真的。模拟的是成交，不是报价。",[24,239,227,240,243],{},[27,241,242],{},"小额真实账户和模拟账户没区别。","”比模拟好得多，但极小的账户有自己的失真：最小手数限制、美分账户的缩放，还有相对微薄余额而言高得离谱的佣金。",[24,245,227,246,249,250,254],{},[27,247,248],{},"模拟账户的记录也是记录。","”它记录的只是能跑起来的代码。",[32,251,253],{"href":252},"\u002Fguides\u002Ffinding-real-trading-accounts","任何查看已发布结果的人","，都应该先看账户类型徽章，再看收益。",[10,256,257,258,123],{},"想知道怎样分辨真实的结果和包装过的结果，请阅读",[32,259,261],{"href":260},"\u002Fguides\u002Fspotting-fake-ea-results","识破虚假的 EA 业绩",{"title":81,"searchDepth":263,"depth":263,"links":264},2,[265,266,267,268,269],{"id":16,"depth":263,"text":16},{"id":87,"depth":263,"text":87},{"id":109,"depth":263,"text":109},{"id":144,"depth":263,"text":145},{"id":221,"depth":263,"text":221},"经纪商","2026-08-19T00:00:00.000Z","模拟账户用模拟资金、按真实价格交易。为什么模拟成交会让策略显得更好，ShowMyTrades 又如何从经纪商读取真实或模拟状态。",false,"md",{},true,"\u002Fzh\u002Fglossary\u002Fwhat-is-a-demo-vs-live-account",[279,280,281,282],"what-is-backtesting","what-is-slippage","what-is-order-execution","what-is-a-verified-track-record",{"title":5,"description":272},"zh\u002F5.glossary\u002Fwhat-is-a-demo-vs-live-account","模拟账户与真实账户 (Demo vs Live Account)","2026-09-10T00:00:00.000Z","DPbHp3Na3Gb8PoKnDP_wdf-NxGfl3xNd230bHU5Huws",[289,472,695,941],{"id":290,"title":291,"body":292,"category":460,"date":271,"description":461,"draft":273,"extension":274,"meta":462,"navigation":276,"path":463,"related":464,"seo":468,"stem":469,"term":470,"updated":271,"__hash__":471},"glossary\u002Fzh\u002F5.glossary\u002Fwhat-is-a-verified-track-record.md","什么是经验证的业绩记录？定义与核查",{"type":7,"value":293,"toc":453},[294,297,299,322,328,331,338,341,343,355,358,360,370,386,388,404,415,417,446],[10,295,296],{},"经验证的业绩记录，是通过只读连接直接从经纪商持续传来的交易结果记录，整条链路上没有任何一个环节能让发布者编辑、删除或调整顺序。这正是“说法”和“证据”的区别。这个词里其实藏着两个不同的问题，关于验证的争论，大多源于把两者混为一谈。",[14,298,16],{"id":16},[10,300,301,302,306,307,311,312,316,317,321],{},"第一个问题：数据是不是真的。这由连接本身来回答：通道只能读取账户，永远不能写入，而且不间断地运行，所以记录是持续积累起来的，而不是挑个方便的时候导出来的。在 ShowMyTrades 上，这条通道是以下四种之一：安装在 MetaTrader 4 或 5 中的 EA（智能交易系统）、使用 ",[32,303,305],{"href":304},"\u002Fglossary\u002Fwhat-is-an-mt4-investor-password","MT4 投资者密码","的直接连接、",[32,308,310],{"href":309},"\u002Fctrader-account-tracking","cTrader OAuth2 授权","，或 ",[32,313,315],{"href":314},"\u002Ftradelocker-tracking","TradeLocker REST API","。这四种方式都不能下单、改单或平仓。只读凭据是行业里实现这一点的通用机制，",[32,318,320],{"href":319},"\u002Fglossary\u002Fwhat-is-myfxbook","Myfxbook"," 等第三方追踪工具也是这样连接 MetaTrader 账户的。",[72,323,326],{"className":324,"code":325,"language":77},[75],"经纪商服务器 → 只读连接 → ShowMyTrades → 公开页面\n　　　　　　   整条链路中没有任何可编辑的环节\n",[79,327,325],{"__ignoreMap":81},[10,329,330],{},"第二个问题：发布页面的人是不是账户的主人。只读数据流能证明交易确实存在，却证明不了交易是谁的，因为投资者密码可以交给任何人。要补上这个缺口，得做一件只有拥有交易权限的人才能做的事：以一个不可能成交的价格挂一笔单，并在注释栏里填上发布者的用户 ID。在 cTrader 上，一次 OAuth 授权就能同时回答这两个问题。",[10,332,333],{},[334,335],"img",{"alt":336,"src":337},"“业绩记录已验证”回答的是：数据是否通过一条无法下单、改单或平仓的通道从经纪商传来；“交易权限已验证”回答的是：发布者是否控制着这个账户","\u002Fimages\u002Farticles\u002Ftwo-badges.svg",[10,339,340],{},"一份记录可能过了一关、没过另一关，所以两者分开统计。",[14,342,87],{"id":87},[10,344,345,346,172,350,354],{},"连接之后的一切都只是算术。只要输入能被改动，盈利因子、",[32,347,349],{"href":348},"\u002Fglossary\u002Fwhat-is-drawdown","回撤",[32,351,353],{"href":352},"\u002Fglossary\u002Fwhat-is-an-equity-curve","净值曲线","就全是摆设，页面做得再详细也没用。截图、PDF 对账单和电子表格有着同一个毛病：它们都出自请您信任它们的那个人之手。",[10,356,357],{},"连续性和只读访问同样重要。可以暂停、再恢复的记录，能让一个糟糕的月份凭空消失。数据流连续时，中断本身就会在页面上显示出来，这本身就是信息，而不是信息的缺失。",[14,359,109],{"id":109},[10,361,362,363,116,366,369],{},"验证要自己主动去做，还得花点功夫，数字也说明了这一点。截至 2026 年 8 月，在 ShowMyTrades 上发布的 1 万+ 个账户中，只有 ",[27,364,365],{},"65 个带有“业绩记录已验证”",[27,367,368],{},"264 个带有“交易权限已验证”","。证明所有权的账户比证明数据来源的多，因为挂一笔单，比去经纪商那里改密码容易。",[10,371,372,373,377,378,381,382,385],{},"已验证不等于表现好。在有交易历史的公开账户中，",[32,374,376],{"href":375},"\u002Fglossary\u002Fwhat-is-time-weighted-return","时间加权收益率","的中位数为 ",[27,379,380],{},"+3.2%","，63.0% 为正，最深回撤的中位数为 ",[27,383,384],{},"9.7%","，还有 17.6% 的账户曾经亏掉峰值的一半以上。这个分布来自 15,436,464 笔已同步交易，这才是诚实的记录该有的样子，和广告里的那些完全不同。",[14,387,145],{"id":144},[10,389,390,391,393,394,172,396,398,399,161,401,403],{},"两枚徽章都在每个账户页面头部的",[27,392,156],{},"组里，位于各项数字上方：",[27,395,171],{},[27,397,175],{},"，获得时显示绿色，未获得时显示灰色，每枚都有提示框，说明它在该平台上认证的是什么。同一组里还会显示",[27,400,160],{},[27,402,164],{},"，这是根据经纪商服务器判断的，不是用户自己填的。",[10,405,406,407,410,411,414],{},"验证在账户设置中管理，",[27,408,409],{},"投资者密码（可选）"," 字段就在那里：清空它并保存，业绩记录徽章当即变灰。",[27,412,413],{},"完整控制台","小组件也会显示这两枚徽章，因为它嵌入了账户头部，所以放在别人网站上的小组件，也带着自己的数据来源。",[14,416,221],{"id":221},[21,418,419,428,434,440],{},[24,420,421,123,424,427],{},[27,422,423],{},"指向某个追踪网站的链接，不等于验证",[32,425,426],{"href":179},"任何平台都能托管未经验证的账户","。要看页面上的徽章，而不是网址里的域名。",[24,429,430,433],{},[27,431,432],{},"“业绩记录已验证”证明不了所有权","。它证明的是交易真实存在。转卖别人投资者密码的人，同样能提供真实的数据流，只是那些交易不是他自己的。",[24,435,436,439],{},[27,437,438],{},"已验证不等于盈利","。徽章只认证数据来源，别的一概不管。之后还得把增长率和回撤放在一起看。",[24,441,442,445],{},[27,443,444],{},"经验证的模拟账户，仍然是模拟账户","。偏偏在实盘成交最吃亏的那些时候，模拟成交显得过于乐观。",[10,447,448,449,123],{},"两枚徽章的验证步骤详解：",[32,450,452],{"href":451},"\u002Fguides\u002Fhow-to-verify-your-account","如何验证您的交易账户",{"title":81,"searchDepth":263,"depth":263,"links":454},[455,456,457,458,459],{"id":16,"depth":263,"text":16},{"id":87,"depth":263,"text":87},{"id":109,"depth":263,"text":109},{"id":144,"depth":263,"text":145},{"id":221,"depth":263,"text":221},"验证","经验证的业绩记录，是通过只读连接直接从经纪商持续获取的业绩数据，发布者无法改动其中任何内容。",{},"\u002Fzh\u002Fglossary\u002Fwhat-is-a-verified-track-record",[465,466,467],"what-is-an-mt4-investor-password","what-is-myfxbook","what-is-drawdown",{"title":291,"description":461},"zh\u002F5.glossary\u002Fwhat-is-a-verified-track-record","经验证的业绩记录 (Verified Track Record)","XXw0XywuXXvGTGPkURwZXnd-tURxBRrjPXvj-rr4h2k",{"id":473,"title":474,"body":475,"category":685,"date":271,"description":686,"draft":273,"extension":274,"meta":687,"navigation":276,"path":688,"related":689,"seo":691,"stem":692,"term":693,"updated":271,"__hash__":694},"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":476,"toc":678},[477,480,482,485,491,494,521,529,531,534,537,539,550,573,582,588,591,593,600,627,637,639,670],[10,478,479],{},"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。",[14,481,16],{"id":16},[10,483,484],{},"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：",[72,486,489],{"className":487,"code":488,"language":77},[75],"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",[79,490,488],{"__ignoreMap":81},[10,492,493],{},"输出靠不靠得住，取决于三项设置：",[21,495,496,502,511],{},[24,497,498,501],{},[27,499,500],{},"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。",[24,503,504,507,508,510],{},[27,505,506],{},"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",[32,509,52],{"href":51},"，已经是另一个策略了。",[24,512,513,516,517,520],{},[27,514,515],{},"滑点与成交逻辑","。测试器默认订单都按请求价成交。",[32,518,519],{"href":34},"滑点","、拒单、重新报价和部分成交，在测试器里统统不存在。",[10,522,523,524,528],{},"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",[32,525,527],{"href":526},"\u002Fglossary\u002Fwhat-is-curve-fitting","曲线拟合","，也是测试表现和实盘表现之间最大的一道鸿沟。",[14,530,87],{"id":87},[10,532,533],{},"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。",[10,535,536],{},"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。",[14,538,109],{"id":109},[10,540,541,542,545,546,549],{},"ShowMyTrades 以公开页面发布了 ",[27,543,544],{},"1 万+ 个账户","，背后是 ",[27,547,548],{},"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。",[10,551,552,553,556,557,561,562,116,565,561,569,572],{},"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 ",[27,554,555],{},"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，",[32,558,560],{"href":559},"\u002Fglossary\u002Fwhat-is-a-profit-factor","盈利因子","中位数为 ",[27,563,564],{},"1.28",[32,566,568],{"href":567},"\u002Fglossary\u002Fwhat-is-the-sharpe-ratio","夏普比率",[27,570,571],{},"0.05","。测试报告里的数字，真正要比的就是这些。",[10,574,575,576,578,579,581],{},"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 ",[27,577,136],{},"，库存费 ",[27,580,140],{},"；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。",[10,583,584],{},[334,585],{"alt":586,"src":587},"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",[10,589,590],{},"测试器用固定点差、又不算库存费，三项成本就漏掉了两项。",[14,592,145],{"id":144},[10,594,595,596,599],{},"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",[32,597,171],{"href":598},"\u002Fglossary\u002Fwhat-is-a-track-record","。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。",[10,601,602,603,172,606,609,610,612,613,615,616,618,619,622,623,626],{},"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",[27,604,605],{},"佣金合计",[27,607,608],{},"已付库存费合计","，以账户货币显示策略的真实成本；",[27,611,192],{},"面板，报告与交易服务器之间的 ",[27,614,200],{}," 和",[27,617,204],{},"，也就是您实际成交时所处的条件；还有",[27,620,621],{},"交易历史与余额变化","，它的",[27,624,625],{},"已平仓交易","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。",[10,628,629,632,633,636],{},[27,630,631],{},"业绩记录已验证","徽章（65 个账户持有）和",[27,634,635],{},"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。",[14,638,221],{"id":221},[21,640,641,647,658,664],{},[24,642,227,643,646],{},[27,644,645],{},"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。",[24,648,227,649,652,653,657],{},[27,650,651],{},"它测试了十年。","”",[32,654,656],{"href":655},"\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。",[24,659,227,660,663],{},[27,661,662],{},"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。",[24,665,227,666,669],{},[27,667,668],{},"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。",[10,671,672,673,677],{},"无论是测试结果还是实盘数字，在相信它之前，先按",[32,674,676],{"href":675},"\u002Fguides\u002Fverifying-trading-performance-claims","如何验证交易业绩宣传","逐项核对一遍。",{"title":81,"searchDepth":263,"depth":263,"links":679},[680,681,682,683,684],{"id":16,"depth":263,"text":16},{"id":87,"depth":263,"text":87},{"id":109,"depth":263,"text":109},{"id":144,"depth":263,"text":145},{"id":221,"depth":263,"text":221},"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",[690,282,280],"what-is-curve-fitting",{"title":474,"description":686},"zh\u002F5.glossary\u002Fwhat-is-backtesting","回测 (Backtesting)","JFA3C0OuDmaFAnqzJS-I02fA4QcPMlLuq0a6Gjsakv8",{"id":696,"title":697,"body":698,"category":270,"date":271,"description":929,"draft":273,"extension":274,"meta":930,"navigation":276,"path":931,"related":932,"seo":936,"stem":937,"term":938,"updated":939,"__hash__":940},"glossary\u002Fzh\u002F5.glossary\u002Fwhat-is-order-execution.md","什么是订单执行？执行类型、重新报价与成交",{"type":7,"value":699,"toc":922},[700,708,710,713,722,732,743,746,752,760,762,765,768,770,781,784,799,801,885,887,915],[10,701,702,703,707],{},"订单执行，是经纪商把下单请求变成按某个价格成交的仓位的过程。从终端发出请求，到服务器确认成交，中间的一切都算在内：",[32,704,706],{"href":705},"\u002Fglossary\u002Fwhat-is-a-b-book-broker","路由","、价格校验，以及接受、重新报价还是拒单的决定。两个账户在不同经纪商跑同一套逻辑，策略毫无差别，结果却可能不一样，原因就在这里。",[14,709,16],{"id":16},[10,711,712],{},"MetaTrader 文档列出了四种执行模式：请求执行（Request）、即时执行（Instant）、市价执行（Market）和交易所执行（Exchange），但零售外汇账户只用其中两种。两者的区别在于：请求还在路上时价格变了，会怎样处理。",[10,714,715,718,719,123],{},[27,716,717],{},"市价执行","。订单按到达服务器那一刻的价格成交。价格变动不会导致拒单，所以基本上一定能成交，但可能",[32,720,721],{"href":34},"出现滑点",[10,723,724,727,728,731],{},[27,725,726],{},"即时执行","。您指定一个价格。如果行情变动超出经纪商允许的范围，服务器会返回",[27,729,730],{},"重新报价","：给出一个新价格，由您决定接受还是放弃。不会有滑点，但交易不一定开得成。",[10,733,734,735,738,739,742],{},"另外还有两种结果值得注意。",[27,736,737],{},"部分成交","：您指定的价格上可成交的量小于订单量，于是只成交一部分，剩下的要么以更差的价格成交，要么被取消。",[27,740,741],{},"拒单","：返回一个错误代码，不产生仓位。",[10,744,745],{},"耗时可以清楚地拆成三段：",[72,747,750],{"className":748,"code":749,"language":77},[75],"往返时间 = 终端 -> 交易服务器             （网络路径）\n         + 订单校验与撮合                 （经纪商）\n         + 交易服务器 -> 终端             （确认）\n",[79,751,749],{"__ignoreMap":81},[10,753,754,755,759],{},"只有中间一段才是经纪商的执行速度。第一段和第三段都是网络路径，所以把终端搬到离交易服务器更近的 ",[32,756,758],{"href":757},"\u002Fglossary\u002Fwhat-is-a-vps","VPS"," 上，经纪商什么都没改，测出来的执行速度也会变。",[14,761,87],{"id":87},[10,763,764],{},"到了执行这一步，策略就不再只是算术题了。回测里每笔订单都按请求价成交，从不重新报价，也从不错过入场。实盘中，这些假设每一条都要付出代价；持仓周期越短，优势中由执行而非信号决定的部分就越大。",[10,766,767],{},"拒单是常被忽略的另一半。滑点至少会反映在成交价上，事后还能分析。一笔根本没开出来的交易，哪里都不会留下记录；只看成功下出去的交易来评判策略，看到的是一个带幸存者偏差的版本。",[14,769,109],{"id":109},[10,771,772,773,776,777,780],{},"在 ShowMyTrades 上有交易历史的公开账户中（2026 年 8 月），订单发往 ",[27,774,775],{},"703 个不同的经纪商服务器","，从中读回了 ",[27,778,779],{},"15,436,464 笔交易","。这里没有统一的执行环境：同一条指令和同一个市场之间，隔着七百个各自配置的服务器。",[10,782,783],{},"这 15,436,464 条记录，每一条都是成交，没有一条是尝试：经纪商的历史记录只报告已开立的仓位，所以重新报价、拒单和被取消的剩余部分天然不在统计之内。本平台如此，任何基于经纪商记录的平台也都如此。",[10,785,786,787,116,790,793,794,798],{},"比较仍然做得到，只是要靠重复。",[27,788,789],{},"699 位用户有不止一个账户",[27,791,792],{},"429 位用户在不止一家经纪商开有账户","：",[32,795,797],{"href":796},"\u002Fguides\u002Fsame-strategy-different-brokers","逻辑不变，只换服务器","，是零售交易者最接近受控执行测试的做法。",[14,800,145],{"id":144},[21,802,803,829,858,879],{},[24,804,805,193,807,197,809,814,815,820,821,201,823,201,825,214,827,123],{},[27,806,192],{},[27,808,196],{},[32,810,812],{"href":811},"\u002Fglossary\u002Fwhat-is-terminal-ping",[27,813,200],{},"，即到经纪商交易服务器的往返时间，以及",[32,816,818],{"href":817},"\u002Fglossary\u002Fwhat-is-packet-retransmission",[27,819,204],{},"，即需要重发的网络数据包比例。低于 50 ms 为优秀；高于 200 ms 可能意味着订单执行变慢、滑点增多。同一面板还显示",[27,822,207],{},[27,824,210],{},[27,826,213],{},[27,828,217],{},[24,830,831,833,834,836,837,201,840,201,843,201,846,201,849,201,852,172,855,857],{},[27,832,621],{},"下的",[27,835,625],{},"表格逐笔列出",[27,838,839],{},"开仓价",[27,841,842],{},"平仓价",[27,844,845],{},"止损",[27,847,848],{},"止盈",[27,850,851],{},"持续时间",[27,853,854],{},"佣金",[27,856,63],{},"。出场价原本设在哪里、实际落在哪里，两相对照，是从经纪商记录里能看到的最接近执行质量的信息。",[24,859,860,863,864,867,868,201,874,172,876,123],{},[27,861,862],{},"高级统计","显示",[27,865,866],{},"最差交易（点）","：如果它远远超出账户平常的止损距离，多半是出场时遇到跳空或成交太差留下的痕迹，而不是入场没做好。另外还有",[32,869,871],{"href":870},"\u002Fglossary\u002Fwhat-is-average-trade-length",[27,872,873],{},"平均持仓时间",[27,875,605],{},[27,877,878],{},"总手数",[24,880,881,882,884],{},"账户头部的",[27,883,186],{},"一行有经纪商徽章，提示框里显示交易服务器名称，旁边是平台、货币和杠杆。决定执行环境的是服务器，不是品牌。",[14,886,221],{"id":221},[223,888,889,895,901,909],{},[24,890,227,891,894],{},[27,892,893],{},"市价执行不如即时执行。","”两者没有高下之分。一种用成交的确定性换价格的确定性，另一种正好相反。哪种让您吃亏，要看策略。",[24,896,227,897,900],{},[27,898,899],{},"ECN、STP 和 NDD 说的是执行方式。","”它们是营销标签，不是受监管的定义，也没有人审计。不如直接衡量账户。",[24,902,227,903,906,907,123],{},[27,904,905],{},"成交慢都怪经纪商。","”很多时候问题出在网络路径上。对服务器下结论之前，先看 ",[27,908,200],{},[24,910,227,911,914],{},[27,912,913],{},"执行快就是执行好。","”速度快而价格差，毫无意义。一笔瞬间成交却差了三个点的订单，照样差了三个点。",[10,916,917,918,123],{},"想比较两家经纪商的执行质量和成本，方法见",[32,919,921],{"href":920},"\u002Fguides\u002Fbroker-cost-audit","经纪商成本核算指南",{"title":81,"searchDepth":263,"depth":263,"links":923},[924,925,926,927,928],{"id":16,"depth":263,"text":16},{"id":87,"depth":263,"text":87},{"id":109,"depth":263,"text":109},{"id":144,"depth":263,"text":145},{"id":221,"depth":263,"text":221},"订单执行，就是经纪商把您的下单请求变成实际成交的过程。市价执行与即时执行、重新报价、部分成交，以及如何衡量执行质量。",{},"\u002Fzh\u002Fglossary\u002Fwhat-is-order-execution",[280,933,934,935],"what-is-server-latency","what-is-a-b-book-broker","what-is-a-spread",{"title":697,"description":929},"zh\u002F5.glossary\u002Fwhat-is-order-execution","订单执行 (Order Execution)","2026-08-29T00:00:00.000Z","rLQV3ID6e807Wa7mw_LLLhn1OhUiAbc-Q1FgIK2m9-Y",{"id":942,"title":943,"body":944,"category":1136,"date":271,"description":1137,"draft":273,"extension":274,"meta":1138,"navigation":276,"path":1139,"related":1140,"seo":1143,"stem":1144,"term":1145,"updated":939,"__hash__":1146},"glossary\u002Fzh\u002F5.glossary\u002Fwhat-is-slippage.md","什么是滑点（Slippage）？定义、成因与真实数据",{"type":7,"value":945,"toc":1129},[946,949,951,957,960,966,972,982,989,991,1001,1007,1009,1025,1036,1038,1041,1093,1095,1121],[10,947,948],{},"滑点，是您下单时预期的价格与实际成交价格之间的差额。从您的终端发出请求，到经纪商服务器完成撮合，中间这段时间市场还在动，滑点就是这样产生的。滑点可能对您不利，也可能对您有利，但实际上并不对称：最容易出现大滑点的时候，恰恰也是最需要精确成交的时候。",[14,950,16],{"id":16},[72,952,955],{"className":953,"code":954,"language":77},[75],"滑点（点） = (成交价 − 请求价) \u002F 点的大小\n             卖出订单取相反符号\n",[79,956,954],{"__ignoreMap":81},[10,958,959],{},"滑点有三个来源。",[10,961,962,965],{},[27,963,964],{},"延迟","。从请求到执行，每一毫秒价格都可能变。延迟来自通往交易服务器的网络路径、终端自身的处理，以及经纪商的撮合队列。",[10,967,968,971],{},[27,969,970],{},"流动性","。订单要在订单簿上成交。如果报价上可成交的数量比您的订单小，剩下的部分就会在下一档价格成交。所以仓位越大，滑点越大；市场越深，滑点越小。",[10,973,974,977,978,981],{},[27,975,976],{},"订单类型","。市价单按当时能拿到的价格成交，所以会滑点。限价单从不滑点，但可能成交不了，这本身也是一种成本。Stop 类订单（包括",[32,979,845],{"href":980},"\u002Fglossary\u002Fwhat-is-a-stop-loss","）一触发就变成市价单，这是最重要的一点：止损保证的是您能离场，而不是离场的价格。",[10,983,984,985,988],{},"最糟糕的时段是可以预见的：预定的经济数据公布、每日展期（rollover）前后流动性稀薄的时段、周日开盘跳空，以及一大批止损同时触发、把订单簿抽空的那一刻。",[27,986,987],{},"正滑点","也存在（成交价比请求价更好）。经纪商如果让您承担负滑点，也把正滑点照样给您，做法就是对称的；如果好的成交自己留着，差的推给您，那就不对称。",[14,990,87],{"id":87},[10,992,993,994,996,997,1000],{},"滑点就是纸面上的策略和服务器上的同一策略之间的差距。用历史卖出价（bid）、按零滑点跑出来的",[32,995,97],{"href":96},"，假设了一种根本不存在的执行方式；",[32,998,999],{"href":870},"平均每笔交易","越短，这个假设凭空造出来的优势就占得越多。",[10,1002,1003,1004,1006],{},"目标越小，这笔账越无情。半个点的不利滑点，是 5 点目标的十分之一，却只是 200 点目标的四百分之一；而且它是在",[32,1005,52],{"href":51},"和佣金之外另算的，不是替代它们。",[14,1008,109],{"id":109},[10,1010,1011,1012,116,1015,1018,1019,793,1021,1024],{},"不等合适时机就出手的系统，最看重执行质量，而平台上的大多数账户正是这样。截至 2026 年 8 月，ShowMyTrades 上有交易历史的公开账户中，",[27,1013,1014],{},"自动交易占比的中位数为 99%",[27,1016,1017],{},"53.9% 的账户自动化程度超过 90%","（42.2% 低于 10%）。这些账户连接了 ",[27,1020,775],{},[32,1022,1023],{"href":796},"同一个策略，发往七百个不同的撮合引擎","，面对七百种不同的延迟情况。",[10,1026,1027,1028,1031,1032,1035],{},"持仓时间又放大了这种风险。这些账户的",[27,1029,1030],{},"持仓时间中位数为 2.4 小时","，中位数账户已平仓 ",[27,1033,1034],{},"171 笔交易","：执行成本付得频繁，每笔交易又没有多少时间让价格波动把它消化掉。",[14,1037,145],{"id":144},[10,1039,1040],{},"我们不记录您的终端请求的价格，所以不公布逐笔滑点这个指标；只靠经纪商数据，任何诚实的平台也都公布不了。我们公布的是能衡量的那一半：造成滑点的延迟。",[21,1042,1043,1068,1085],{},[24,1044,1045,193,1047,1049,1050,1054,1055,1059,1060,201,1062,201,1064,214,1066,123],{},[27,1046,192],{},[27,1048,196],{},"）会显示 ",[32,1051,1052],{"href":811},[27,1053,200],{},"，即终端与经纪商交易服务器之间的往返时间，以及",[32,1056,1057],{"href":817},[27,1058,204],{},"，即需要重新发送的网络数据包比例。Ping 低于 50 ms 属于优秀；高于 200 ms 可能意味着执行更慢、滑点更多。重传率一直偏高，说明网络不稳定，或者交易服务器离得太远。同一面板还会显示",[27,1061,207],{},[27,1063,210],{},[27,1065,213],{},[27,1067,217],{},[24,1069,1070,833,1072,1074,1075,172,1077,1079,1080,172,1082,1084],{},[27,1071,621],{},[27,1073,625],{},"表格，每个订单都有",[27,1076,839],{},[27,1078,842],{},"，旁边是",[27,1081,845],{},[27,1083,848],{},"列。把离场价位的设定值和实际成交的位置一比，就是从经纪商记录里能读到的最接近滑点的数据。",[24,1086,1087,1089,1090,1092],{},[27,1088,862],{},"中的",[27,1091,866],{}," 可以用来快速检查。如果最差交易远远超出账户原本设定的止损距离，那就是跳空或滑点离场留下的痕迹，而不是入场没做好。",[14,1094,221],{"id":221},[223,1096,1097,1103,1109,1115],{},[24,1098,1099,1102],{},[27,1100,1101],{},"止损保证的是离场，不是价格","。遇到跳空，它会按第一个能成交的价格平仓，可能远远超出您设定的价位。",[24,1104,1105,1108],{},[27,1106,1107],{},"市价单上的“零滑点”，是谁都兑现不了的承诺","。保证止损确实有，但那是另一种产品，要另外付费。",[24,1110,1111,1114],{},[27,1112,1113],{},"滑点不一定是亏损","。看看您的成交价有没有比请求价更好的时候；如果从来没有，这本身就说明了这家经纪商的情况。",[24,1116,1117,1120],{},[27,1118,1119],{},"模拟账户的结果偏乐观","。模拟服务器不模拟排队位置和订单簿深度，所以在模拟账户上完美无缺的策略，到了实盘可能只是勉强能用。",[10,1122,1123,1124,1128],{},"想了解 Ping、重传与页面上的其他内容如何结合起来看，请阅读",[32,1125,1127],{"href":1126},"\u002Fguides\u002Freading-a-trading-account-dashboard","如何读懂交易账户控制台","指南。",{"title":81,"searchDepth":263,"depth":263,"links":1130},[1131,1132,1133,1134,1135],{"id":16,"depth":263,"text":16},{"id":87,"depth":263,"text":87},{"id":109,"depth":263,"text":109},{"id":144,"depth":263,"text":145},{"id":221,"depth":263,"text":221},"成本","滑点是您预期的价格与实际成交价之间的差额。本文解释它为什么总在最要命的时候最严重，以及延迟如何让它变得可以衡量。",{},"\u002Fzh\u002Fglossary\u002Fwhat-is-slippage",[935,1141,1142,279],"what-is-a-stop-loss","what-is-a-vps",{"title":943,"description":1137},"zh\u002F5.glossary\u002Fwhat-is-slippage","滑点 (Slippage)","_usMaUZBN8x6rtKI1x17K10-_LxzY0duV2FLyHX73gE",[1148,1150],{"path":1149},"\u002Fglossary\u002Fwhat-is-a-demo-vs-live-account",{"path":277},1793898060484]