[{"data":1,"prerenderedAt":1098},["ShallowReactive",2],{"\u002Fzh\u002Fglossary\u002Fwhat-is-order-execution":3,"\u002Fzh\u002Fglossary\u002Fwhat-is-order-execution-related":297,"\u002Fzh\u002Fglossary\u002Fwhat-is-order-execution-langs":1095},{"id":4,"title":5,"body":6,"category":279,"date":280,"description":281,"draft":282,"extension":283,"meta":284,"navigation":285,"path":286,"related":287,"seo":292,"stem":293,"term":294,"updated":295,"__hash__":296},"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":8,"toc":271},"minimark",[9,19,23,26,38,48,59,62,73,81,84,87,90,93,104,107,123,127,231,234,264],[10,11,12,13,18],"p",{},"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：",[14,15,17],"a",{"href":16},"\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。",[20,21,22],"h2",{"id":22},"工作原理",[10,24,25],{},"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。",[10,27,28,32,33,37],{},[29,30,31],"strong",{},"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",[14,34,36],{"href":35},"\u002Fglossary\u002Fwhat-is-slippage","出现滑点","。",[10,39,40,43,44,47],{},[29,41,42],{},"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",[29,45,46],{},"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。",[10,49,50,51,54,55,58],{},"另外还有两种结果值得注意。",[29,52,53],{},"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。",[29,56,57],{},"拒单","：返回一个错误代码，不产生仓位。",[10,60,61],{},"耗时可以清楚地拆成三段：",[63,64,69],"pre",{"className":65,"code":67,"language":68},[66],"language-text","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","text",[70,71,67],"code",{"__ignoreMap":72},"",[10,74,75,76,80],{},"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 ",[14,77,79],{"href":78},"\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。",[20,82,83],{"id":83},"为什么重要",[10,85,86],{},"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。",[10,88,89],{},"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。",[20,91,92],{"id":92},"数据显示了什么",[10,94,95,96,99,100,103],{},"在 ShowMyTrades 上有交易历史的公开账户中（2026 年 8 月），订单发往 ",[29,97,98],{},"703 个不同的经纪商服务器","，从中读回了 ",[29,101,102],{},"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。",[10,105,106],{},"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。",[10,108,109,110,113,114,117,118,122],{},"比较仍然做得到，只是要靠重复。",[29,111,112],{},"699 位用户有不止一个账户","，",[29,115,116],{},"429 位用户在不止一家经纪商开有账户","：",[14,119,121],{"href":120},"\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。",[20,124,126],{"id":125},"在-showmytrades-上的位置","在 ShowMyTrades 上的位置",[128,129,130,169,202,224],"ul",{},[131,132,133,136,137,140,141,147,148,154,155,158,159,158,162,165,166,37],"li",{},[29,134,135],{},"终端","面板（MT5 上为",[29,138,139],{},"终端与已打开的图表","）显示 ",[14,142,144],{"href":143},"\u002Fglossary\u002Fwhat-is-terminal-ping",[29,145,146],{},"Ping","，即到经纪商交易服务器的往返时间，以及",[14,149,151],{"href":150},"\u002Fglossary\u002Fwhat-is-packet-retransmission",[29,152,153],{},"重传","，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",[29,156,157],{},"终端版本号","、",[29,160,161],{},"内存",[29,163,164],{},"自动交易","和 ",[29,167,168],{},"DLL 导入",[131,170,171,174,175,178,179,158,182,158,185,158,188,158,191,158,194,197,198,201],{},[29,172,173],{},"交易历史与余额变化","下的",[29,176,177],{},"已平仓交易","表格逐笔列出",[29,180,181],{},"开仓价",[29,183,184],{},"平仓价",[29,186,187],{},"止损",[29,189,190],{},"止盈",[29,192,193],{},"持续时间",[29,195,196],{},"佣金","和",[29,199,200],{},"库存费","。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。",[131,203,204,207,208,211,212,158,218,197,221,37],{},[29,205,206],{},"高级统计","显示",[29,209,210],{},"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",[14,213,215],{"href":214},"\u002Fglossary\u002Fwhat-is-average-trade-length",[29,216,217],{},"平均持仓时间",[29,219,220],{},"佣金合计",[29,222,223],{},"总手数",[131,225,226,227,230],{},"账户头部的",[29,228,229],{},"信息","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。",[20,232,233],{"id":233},"常见误解",[235,236,237,244,250,258],"ol",{},[131,238,239,240,243],{},"“",[29,241,242],{},"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。",[131,245,239,246,249],{},[29,247,248],{},"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。",[131,251,239,252,255,256,37],{},[29,253,254],{},"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 ",[29,257,146],{},[131,259,239,260,263],{},[29,261,262],{},"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。",[10,265,266,267,37],{},"想比较两家经纪商的执行质量和成本，方法见",[14,268,270],{"href":269},"\u002Fguides\u002Fbroker-cost-audit","经纪商成本核算指南",{"title":72,"searchDepth":272,"depth":272,"links":273},2,[274,275,276,277,278],{"id":22,"depth":272,"text":22},{"id":83,"depth":272,"text":83},{"id":92,"depth":272,"text":92},{"id":125,"depth":272,"text":126},{"id":233,"depth":272,"text":233},"经纪商","2026-08-19T00:00:00.000Z","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。",false,"md",{},true,"\u002Fzh\u002Fglossary\u002Fwhat-is-order-execution",[288,289,290,291],"what-is-slippage","what-is-server-latency","what-is-a-b-book-broker","what-is-a-spread",{"title":5,"description":281},"zh\u002F5.glossary\u002Fwhat-is-order-execution","订单执行 (Order Execution)","2026-08-29T00:00:00.000Z","rLQV3ID6e807Wa7mw_LLLhn1OhUiAbc-Q1FgIK2m9-Y",[298,502,706,891],{"id":299,"title":300,"body":301,"category":279,"date":280,"description":492,"draft":282,"extension":283,"meta":493,"navigation":285,"path":494,"related":495,"seo":498,"stem":499,"term":500,"updated":295,"__hash__":501},"glossary\u002Fzh\u002F5.glossary\u002Fwhat-is-a-b-book-broker.md","什么是 B-book 经纪商？A-book 与 B-book 详解",{"type":7,"value":302,"toc":485},[303,306,308,311,317,323,329,335,338,340,348,351,359,361,375,382,384,387,451,453,479],[10,304,305],{},"B-book 经纪商不把客户的订单送往外部流动性提供商，而是自己在账上接下对手盘。A-book 经纪商则把订单转出去，靠这部分订单流的加价或佣金赚钱。两种都是合法、常见的商业模式。多数零售经纪商两者并用，按客户、按交易品种决定哪些订单流走哪条路。",[20,307,22],{"id":22},[10,309,310],{},"订单到了经纪商那里，只有两种处理方式。",[10,312,313,316],{},[29,314,315],{},"A-book","。经纪商在流动性提供商那里开一笔对应的仓位。您赚的钱由流动性提供商支付；经纪商的收入来自点差加价和佣金，您盈也好亏也好，它都赚一样多。",[10,318,319,322],{},[29,320,321],{},"B-book","。经纪商直接做您的对手方，仓位留在内部，不会产生外部订单。您亏的钱就是它的收入，您赚的钱就是它的成本。",[10,324,325,328],{},[29,326,327],{},"混合路由","才是常态，而不是例外。经纪商会给订单流分类（按客户、规模、交易品种、历史盈利情况），一部分内部化，其余的对冲出去。内部化的订单流还会互相轧差：一位客户做多一手，另一位做空一手，两笔正好抵消，经纪商没有任何市场敞口，两边的点差却都照收。",[63,330,333],{"className":331,"code":332,"language":68},[66],"A-book 收入 = 加价 + 佣金\nB-book 收入 = 加价 + 佣金 + 客户净亏损 - 客户净盈利\n",[70,334,332],{"__ignoreMap":72},[10,336,337],{},"这里说的是订单路由，不是经营操守，两者不能混为一谈。",[20,339,83],{"id":83},[10,341,342,343,347],{},"B-book 模式有一个结构性的利益冲突：客户账户越差，公司收入越好。这是激励机制层面的事实，决定把钱放在哪里之前值得弄明白，但单凭这一点，证明不了任何一家经纪商的具体行为。受监管的公司要遵守最佳执行义务、客户资金隔离和报告要求，这些规定正是因为存在这种激励才设立的。所以",[14,344,346],{"href":345},"\u002Fglossary\u002Fwhat-is-a-regulated-broker","监管机构","比路由模式更要紧。",[10,349,350],{},"反过来，这种取舍也有另一面。正因为有内部化，才有微型手，也不会因为下单量大而被拒单。纯 A-book 接入的是真实流动性，也就意味着行情清淡时会真的遇到重新报价、部分成交和最小下单量限制。两种模式谈不上谁优谁劣。",[10,352,353,354,358],{},"麻烦在于，路由从外部看不到。经纪商不必公布每笔订单怎么路由，第三方也无法从交易记录里还原。所以该问的不是您在哪本账簿上，而是您自己的",[14,355,357],{"href":356},"\u002Fglossary\u002Fwhat-is-order-execution","订单执行","实际表现如何。",[20,360,92],{"id":92},[10,362,363,364,366,367,370,371,374],{},"在 ShowMyTrades 上有交易历史的公开账户中（2026 年 8 月），账户分布在 ",[29,365,98],{},"上，账户所有者在 ",[29,368,369],{},"1,724,575 手","交易上一共付了 ",[29,372,373],{},"$4,782,670 的佣金","。这类成本之所以能衡量，是因为它单独列成一项收费。路由根本不收费，所以本页（以及任何根据经纪商记录汇总的页面）上没有一个数字能说明订单去了哪里。A-book 和 B-book 两边，都有公司采用佣金制报价。",[10,376,377,378,381],{},"能比较的，是一个交易环境和另一个交易环境。",[29,379,380],{},"这里有 429 位用户在不止一家经纪商开有账户","，用同样的规则、在同一时期跑两个账户，就能把成本和成交质量从策略本身中剥离出来。路由标签本来想回答的问题，这样的比较就能回答，而且靠的是数字，不是推测。",[20,383,126],{"id":125},[10,385,386],{},"我们不给经纪商贴 A-book 或 B-book 的标签，以后也不会：要诚实地贴这个标签，所需的信息我们拿不到，公司以外的任何人也拿不到。账户页面给您的，是能衡量的那一层。",[128,388,389,398,412,435],{},[131,390,226,391,393,394,397],{},[29,392,229],{},"一行有经纪商徽章，提示框里显示",[29,395,396],{},"交易服务器名称","。决定交易环境的是服务器，不是品牌：同一家经纪商如果有几个服务器组，各组的定价和执行往往不一样。",[131,399,400,136,402,140,404,408,409,411],{},[29,401,135],{},[29,403,139],{},[14,405,406],{"href":143},[29,407,146],{}," 和",[29,410,153],{},"，也就是每笔成交背后的连接质量。",[131,413,414,207,416,158,418,158,420,197,423,426,427,430,431,37],{},[29,415,206],{},[29,417,220],{},[29,419,223],{},[29,421,422],{},"总点数",[29,424,425],{},"已付库存费合计","。佣金除以手数，就是",[14,428,429],{"href":269},"账户每手的真实成本","；点数是正的、资金却没涨，说明",[14,432,434],{"href":433},"\u002Fglossary\u002Fwhat-is-cost-percentage","执行成本正在吃掉优势",[131,436,437,174,439,178,441,197,443,445,446,158,448,450],{},[29,438,173],{},[29,440,177],{},[29,442,181],{},[29,444,184],{},"，可以和",[29,447,187],{},[29,449,190],{},"对照。",[20,452,233],{"id":233},[235,454,455,461,467,473],{},[131,456,239,457,460],{},[29,458,459],{},"B-book 就是经纪商专门在和我对赌。","”内部化的订单流大部分在客户之间轧差掉了。公司持有的是轧差后剩下的部分，并不是您仓位的镜像。",[131,462,239,463,466],{},[29,464,465],{},"A-book 就没有利益冲突。","”A-book 经纪商靠成交量赚钱，它的动力在于交易量，这同样不会天然地与您的生存利益一致。",[131,468,239,469,472],{},[29,470,471],{},"名字里有 ECN、STP 或 NDD，就一定是 A-book。","”这些词是不受监管的营销标签，也没有人审计。",[131,474,239,475,478],{},[29,476,477],{},"连亏一大段，就证明有人操纵。","”通常证明的只是方差。要检验，就在另一家经纪商用同样的规则跑第二个账户，而不是去论坛发帖。",[10,480,481,482,37],{},"想有条理地做这种比较，请阅读",[14,483,484],{"href":120},"在不同经纪商运行同一策略",{"title":72,"searchDepth":272,"depth":272,"links":486},[487,488,489,490,491],{"id":22,"depth":272,"text":22},{"id":83,"depth":272,"text":83},{"id":92,"depth":272,"text":92},{"id":125,"depth":272,"text":126},{"id":233,"depth":272,"text":233},"B-book 经纪商不把您的订单送到市场，而是自己做您的对手方。A-book 与 B-book 的订单路由有何不同，各自又让您付出什么成本。",{},"\u002Fzh\u002Fglossary\u002Fwhat-is-a-b-book-broker",[496,497,288,291],"what-is-order-execution","what-is-a-regulated-broker",{"title":300,"description":492},"zh\u002F5.glossary\u002Fwhat-is-a-b-book-broker","B-book 经纪商 (B-Book Broker)","sI_MiwbkIuDlGwV9jblaPEpZ18mS40cCTp7Vl3dz2q0",{"id":503,"title":504,"body":505,"category":695,"date":280,"description":696,"draft":282,"extension":283,"meta":697,"navigation":285,"path":698,"related":699,"seo":702,"stem":703,"term":704,"updated":280,"__hash__":705},"glossary\u002Fzh\u002F5.glossary\u002Fwhat-is-a-spread.md","外汇中的点差（Spread）是什么？定义与真实成本",{"type":7,"value":506,"toc":688},[507,510,512,515,521,529,540,555,557,564,567,569,575,582,585,591,597,599,647,649,681],[10,508,509],{},"点差是卖出价（bid，您卖出时的价格）和买入价（ask，您买入时的价格）之间的差额。这是经纪商为您成交订单收取的费用，不管这笔交易是赚是亏，每一笔都要付。在大多数零售账户上，点差是最大的单项交易成本，也是唯一一项从不在对账单上单独列出的成本。",[20,511,22],{"id":22},[10,513,514],{},"假设 EUR\u002FUSD 报价 1.08432 \u002F 1.08442，点差就是 1.0 点。买入按买入价（ask），卖出按卖出价（bid），所以仓位一开出来，就带着一笔与点差相等的浮亏。价格得先走完这段距离，交易才能回到不赚不亏。",[63,516,519],{"className":517,"code":518,"language":68},[66],"点差（点）   = (买入价 − 卖出价) \u002F 点的大小\n每笔交易成本 = 点差（点） × 点值 × 手数\n",[70,520,518],{"__ignoreMap":72},[10,522,523,524,528],{},"EUR\u002FUSD 一个标准手的",[14,525,527],{"href":526},"\u002Fglossary\u002Fwhat-is-a-pip","点值","是 $10，所以 1.0 点的点差，每手成本就是 $10，一开一平只收一次。",[10,530,531,532,535,536,539],{},"经纪商的定价方式有两种。",[29,533,534],{},"纯点差","账户把报价拉宽，此外不收任何费用，成本就藏在您的成交价里。",[29,537,538],{},"原始点差加佣金","账户的报价贴近银行间价格，差额单独收取。两种方式的总成本可能完全一样，但只有第二种让成本看得见。",[10,541,542,543,546,547,550,551,554],{},"点差分",[29,544,545],{},"固定点差","（不变，通常较宽，由做市商报价）和",[29,548,549],{},"浮动点差","（随真实流动性变化）。伦敦和纽约时段重叠时，主要货币对的浮动点差接近于零；而在经济数据公布时、每日展期（rollover）前后流动性稀薄的那一小时，以及周日开盘时，点差会急剧扩大。偏偏这些时候，也是您的",[14,552,187],{"href":553},"\u002Fglossary\u002Fwhat-is-a-stop-loss","最容易被打到的时候。",[20,556,83],{"id":83},[10,558,559,560,563],{},"点差随交易频率增长，跟交易水平无关。每笔目标 5 点的系统，1 点的点差就吃掉了",[14,561,562],{"href":433},"毛优势的 20%","；目标 200 点的系统，只被吃掉 0.5%。同一套逻辑的策略，在一家经纪商活得下来，在另一家却活不下去，往往就取决于这一个比例。",[10,565,566],{},"而且它在看不见的地方不断累积。ShowMyTrades 上已发布账户的已平仓交易中位数是 171 笔。171 个标准手，每手多 1 点点差，就是 $1,710。这笔钱不会出现在任何地方，只会让净值曲线稍微差一点。",[20,568,92],{"id":92},[10,570,571,572,574],{},"截至 2026 年 8 月，ShowMyTrades 上有交易历史的公开账户共支付了 ",[29,573,373],{},"。这只是执行成本中看得见的那一半；点差没法用同样的方式汇总，因为它包含在成交价里，而不是单独收取。",[10,576,577],{},[578,579],"img",{"alt":580,"src":581},"经纪商的三项成本：点差在每次开仓和平仓时收取，在盈亏列中看不见；佣金逐笔列出；库存费按持仓时间收取：ShowMyTrades 公开账户共支付 $4,782,670 佣金和 $862,547 库存费","\u002Fimages\u002Farticles\u002Fwhere-costs-hide.svg",[10,583,584],{},"三项成本里，只有一项会带着数字出现。",[10,586,587,588,590],{},"看不见的这一半有多大，不难估算。这些账户共交易了 ",[29,589,369],{},"。在每点 $10 的品种上，这么大的交易量，平均点差哪怕只差 0.2 点，也大约是 340 万美元，超过平台上有记录以来全部佣金的三分之二。",[10,592,593,594,596],{},"而且各家的交易条件确实不一样：这些账户连接了 ",[29,595,98],{},"，还有 429 位用户特意在多家经纪商开户，就是为了相互比较。",[20,598,126],{"id":125},[128,600,601,616,626,641],{},[131,602,603,207,605,158,607,197,609,611,612,615],{},[29,604,206],{},[29,606,220],{},[29,608,223],{},[29,610,422],{},"。用第一项除以第二项，就得到",[14,613,614],{"href":269},"该账户每手的真实佣金","，经纪商很少用可比较的形式公布这个数字。",[131,617,618,619,621,622,37],{},"最有用的检查，是把",[29,620,422],{},"和净利润放在一起看。账户按点数是赚的，按金额却持平甚至亏损，说明执行成本正在吃掉优势。",[14,623,625],{"href":624},"\u002Fguides\u002Ftrading-cost-percentage","这个差距就是点差和佣金的账单",[131,627,628,174,630,632,633,635,636,197,638,640],{},[29,629,173],{},[29,631,177],{},"表格，每个订单都有一列",[29,634,196],{},"，旁边是",[29,637,181],{},[29,639,184],{},"，每一笔成交实际花了多少钱，一看便知。",[131,642,643,646],{},[29,644,645],{},"账户头部","显示经纪商标识（提示框里有交易服务器名称），旁边是账户货币和杠杆。正因为有这些信息，运行同一策略的两个账户才谈得上比较。",[20,648,233],{"id":233},[235,650,651,657,663,669],{},[131,652,653,656],{},[29,654,655],{},"“零点差”不等于免费","。零点差或原始点差账户只是把成本挪到了佣金上。要比总成本，别看宣传口号。",[131,658,659,662],{},[29,660,661],{},"广告上的点差是最理想的情况","。经纪商报的是流动性充足时段测得的典型点差或最低点差。利率决议期间您拿到的成交价，完全是另一回事。",[131,664,665,668],{},[29,666,667],{},"点差不在对账单上，于是人们以为自己没付","。其实它在开仓时就扣掉了，那时这笔交易还没成为表格里的一行。",[131,670,671,37,674,197,677,680],{},[29,672,673],{},"光看点差，排不出经纪商的高下",[14,675,200],{"href":676},"\u002Fglossary\u002Fwhat-is-a-swap-in-forex",[14,678,679],{"href":35},"滑点","也要放进同一个比较里；点差窄但执行差，反而更贵。",[10,682,683,684,37],{},"想用您自己的品种和手数算出真实数字，请参阅",[14,685,687],{"href":686},"\u002Fguides\u002Fforex-calculators-guide","外汇计算器指南",{"title":72,"searchDepth":272,"depth":272,"links":689},[690,691,692,693,694],{"id":22,"depth":272,"text":22},{"id":83,"depth":272,"text":83},{"id":92,"depth":272,"text":92},{"id":125,"depth":272,"text":126},{"id":233,"depth":272,"text":233},"成本","点差是买入价与卖出价之间的差距，也就是您开仓要付的成本。本文介绍点差的原理、每手成本，以及各家经纪商为何不同。",{},"\u002Fzh\u002Fglossary\u002Fwhat-is-a-spread",[700,288,497,701],"what-is-a-swap-in-forex","what-is-a-profit-factor",{"title":504,"description":696},"zh\u002F5.glossary\u002Fwhat-is-a-spread","点差 (Spread)","pmILRTUEL-hRknwSO9zZZssKW8sK6YY1FwEoFE5VfCc",{"id":707,"title":708,"body":709,"category":878,"date":280,"description":879,"draft":282,"extension":283,"meta":880,"navigation":285,"path":881,"related":882,"seo":886,"stem":887,"term":888,"updated":889,"__hash__":890},"glossary\u002Fzh\u002F5.glossary\u002Fwhat-is-server-latency.md","什么是服务器延迟？Ping、重传与滑点",{"type":7,"value":710,"toc":871},[711,720,722,728,733,741,748,750,756,759,762,764,785,799,801,812,827,833,835,864],[10,712,713,714,719],{},"服务器延迟，是一条消息从您的交易终端到经纪商的交易服务器、再返回所用的时间。MetaTrader 把它显示为 ",[14,715,716],{"href":143},[29,717,718],{},"ping","，单位是毫秒。它取决于两台具体机器之间的那条路径，跟您办的宽带套餐无关，也不单是经纪商的事。",[20,721,22],{"id":22},[63,723,726],{"className":724,"code":725,"language":68},[66],"Ping (ms) = 终端 → 交易服务器 → 终端，一次往返\n",[70,727,725],{"__ignoreMap":72},[10,729,730,731,37],{},"决定延迟的主要是两样东西：物理距离，以及中间经过多少个网络节点（跳数）。带宽几乎不起作用。光在光纤里每毫秒大约走 200 km，一来一回时间翻倍，每台路由器还会加上自己的排队延迟。悉尼一条 500 Mbit 的家庭宽带连伦敦的服务器，怎么也比不过和交易服务器在同一个数据中心的 10 Mbit ",[14,732,79],{"href":78},[10,734,735,736,740],{},"每笔交易都要在两个方向上各付一次延迟的代价。价格传到终端时已经晚了，",[14,737,739],{"href":738},"\u002Fglossary\u002Fwhat-is-an-expert-advisor","EA（智能交易系统）","据以决策的报价其实已经过时；订单再原路返回，等服务器收到时，行情又变了。",[10,742,743,747],{},[14,744,745],{"href":150},[29,746,153],{},"衡量的是另一回事，数值高时问题更严重。TCP 会把丢失或没有得到确认的数据包重新发送，以保证送达，重传数值就是需要重发的数据包占比。丢一个包，代价不是多走一跳，而是一次超时，所以重传不会平稳地抬高平均延迟，而是制造尖峰。ping 40 ms 但有明显重传的账户，线路反而不如稳定在 90 ms 的账户可预测；而且这些尖峰出现的时机与您的持仓毫无关联，完全随机。",[20,749,83],{"id":83},[10,751,752,753,755],{},"延迟是",[14,754,679],{"href":35},"中您真正能测量的那一半。从做出决定到成交，每一毫秒价格都可能变动；而在趋势行情或新闻冲击下，变动的方向并不是随机的。",[10,757,758],{},"风险随交易频率放大，与赚多少无关。持仓三周的单子，察觉不到 200 ms 的差别；每天开好几笔交易的系统，每次开仓、每次平仓都要为这段延迟付出代价。",[10,760,761],{},"最伤人的是这种不对称：让行情剧烈波动的那股负载，同样会拖慢线路，所以最慢的往返和变得最快的价格，往往是一起来的，而不是各自独立发生。",[20,763,92],{"id":92},[10,765,766,767,773,774,113,777,784],{},"延迟按订单收费，所以关键在于账户交易有多频繁、利润空间有多薄。截至 2026 年 8 月，在 ShowMyTrades 上有交易历史的公开账户中，",[29,768,769,772],{},[14,770,771],{"href":214},"持仓时间中位数","为 2.4 小时","，每个账户已平仓交易的中位数为 ",[29,775,776],{},"171 笔",[29,778,779,783],{},[14,780,782],{"href":781},"\u002Fglossary\u002Fwhat-is-a-profit-factor","盈利因子","中位数为 1.28","。利润空间这么薄，执行质量就不是结果里可以忽略的零头，而是结果本身的一部分。",[10,786,787,788,790,791,794,795,798],{},"这些账户连接着 ",[29,789,98],{},"，已同步的交易共 ",[29,792,793],{},"15,436,464 笔","：七百个撮合引擎，分处七百栋楼里，与终端所在位置的距离各不相同。而且 ",[29,796,797],{},"53.9% 的账户自动化程度超过 90%","，也就是说，这里的大多数订单都是条件一满足就立即发出的，发单的程序不可能自己决定等行情平静些再出手。",[20,800,126],{"id":125},[10,802,803,804,806,807,408,809,811],{},"公开的业绩记录，几乎从不交代产生它的那台机器。使用 EA 连接时，我们的业绩记录会交代：",[29,805,139],{},"模块公开到交易服务器的 ",[29,808,146],{},[29,810,153],{},"，旁边是终端的版本号、内存和权限，还有连接徽章、终端上次报告距今多久，以及您有几个终端在线。",[10,813,814,815,818,819,822,823,826],{},"这两个数字都是终端对自己实际所走路径的测量，不是根据您所在位置的估算；判读它们的阈值写在提示框里，而不是全凭某个人的经验：ping 低于 ",[29,816,817],{},"50 ms"," 为优秀，高于 ",[29,820,821],{},"200 ms"," 可能意味着执行更慢、滑点更多；重传接近 ",[29,824,825],{},"0%"," 表示连接稳定，持续偏高则说明网络不可靠，或者交易服务器距离较远。",[10,828,829,830,832],{},"有两点要说明。直接连接从不接触您的机器，所以这些数据一项都没有。另外，在 MT4 上这个模块的标题只是",[29,831,135],{},"，因为 MT4 的 EA 不发送图表列表。",[20,834,233],{"id":233},[128,836,837,843,849,855],{},[131,838,239,839,842],{},[29,840,841],{},"Ping 就是我的执行速度。","”Ping 只衡量网络往返，其他一概不管。经纪商自己的处理时间、排队，以及任何交易员干预，都叠加在这个数字之上，从数字里根本看不出来。",[131,844,239,845,848],{},[29,846,847],{},"网速快，延迟就低。","”一个订单只有几百字节，管道粗细从来不是瓶颈，管道长短才是。所以离交易服务器近，永远胜过带宽大。",[131,850,239,851,854],{},[29,852,853],{},"平均值看着没问题，线路就没问题。","”平均值会掩盖尖峰，而重传数值报告的恰恰就是尖峰。订单不会挑平均的时刻发出，而是扎堆出现在行情波动的时候。",[131,856,239,857,860,861,37],{},[29,858,859],{},"显示已连接，说明终端还在运行。","”这个标志只是终端最后一次报告的值，一台几周前就关了机的机器，也照样显示已连接。关键看数据新不新：所以数据过时后，徽章会退回",[29,862,863],{},"离线",[10,865,866,867,37],{},"延迟和账户页面上的其他数字该怎么对照着看，见：",[14,868,870],{"href":869},"\u002Fguides\u002Freading-a-trading-account-dashboard","如何读懂交易账户控制台",{"title":72,"searchDepth":272,"depth":272,"links":872},[873,874,875,876,877],{"id":22,"depth":272,"text":22},{"id":83,"depth":272,"text":83},{"id":92,"depth":272,"text":92},{"id":125,"depth":272,"text":126},{"id":233,"depth":272,"text":233},"平台","服务器延迟是您的终端与经纪商交易服务器之间的往返时间。本文讲它为什么会变成滑点，以及重传怎样暴露线路的问题。",{},"\u002Fzh\u002Fglossary\u002Fwhat-is-server-latency",[288,883,884,885],"what-is-a-vps","what-is-metatrader","what-is-an-expert-advisor",{"title":708,"description":879},"zh\u002F5.glossary\u002Fwhat-is-server-latency","服务器延迟 (Server Latency)","2026-09-17T00:00:00.000Z","k8FyXCWak4kY1xagURaKqgzzCLA7HOEoVOhF3orePQI",{"id":892,"title":893,"body":894,"category":695,"date":280,"description":1085,"draft":282,"extension":283,"meta":1086,"navigation":285,"path":1087,"related":1088,"seo":1091,"stem":1092,"term":1093,"updated":295,"__hash__":1094},"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":895,"toc":1078},[896,899,901,907,910,916,922,931,938,940,952,960,962,977,988,990,993,1044,1046,1072],[10,897,898],{},"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。",[20,900,22],{"id":22},[63,902,905],{"className":903,"code":904,"language":68},[66],"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",[70,906,904],{"__ignoreMap":72},[10,908,909],{},"滑点有三个来源。",[10,911,912,915],{},[29,913,914],{},"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。",[10,917,918,921],{},[29,919,920],{},"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。",[10,923,924,927,928,930],{},[29,925,926],{},"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",[14,929,187],{"href":553},"）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。",[10,932,933,934,937],{},"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。",[29,935,936],{},"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。",[20,939,83],{"id":83},[10,941,942,943,947,948,951],{},"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",[14,944,946],{"href":945},"\u002Fglossary\u002Fwhat-is-backtesting","回测","，假设了一种根本不存在的执行方式；",[14,949,950],{"href":214},"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。",[10,953,954,955,959],{},"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",[14,956,958],{"href":957},"\u002Fglossary\u002Fwhat-is-a-spread","点差","和佣金之外另算的，不是替代它们。",[20,961,92],{"id":92},[10,963,964,965,113,968,970,971,117,973,976],{},"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，",[29,966,967],{},"自动交易占比的中位数为 99%",[29,969,797],{},"（42.2% 低于 10%）。这些账户连接了 ",[29,972,98],{},[14,974,975],{"href":120},"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。",[10,978,979,980,983,984,987],{},"持仓时间又放大了这种风险。这些账户的",[29,981,982],{},"持仓时间中位数为 2.4 小时","，中位数账户已平仓 ",[29,985,986],{},"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。",[20,989,126],{"id":125},[10,991,992],{},"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。",[128,994,995,1020,1036],{},[131,996,997,136,999,1001,1002,1006,1007,1011,1012,158,1014,158,1016,165,1018,37],{},[29,998,135],{},[29,1000,139],{},"）会显示 ",[14,1003,1004],{"href":143},[29,1005,146],{},"，即终端与经纪商交易服务器之间的往返时间，以及",[14,1008,1009],{"href":150},[29,1010,153],{},"，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",[29,1013,157],{},[29,1015,161],{},[29,1017,164],{},[29,1019,168],{},[131,1021,1022,174,1024,1026,1027,197,1029,635,1031,197,1033,1035],{},[29,1023,173],{},[29,1025,177],{},"表格，每个订单都有",[29,1028,181],{},[29,1030,184],{},[29,1032,187],{},[29,1034,190],{},"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。",[131,1037,1038,1040,1041,1043],{},[29,1039,206],{},"中的",[29,1042,210],{}," 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。",[20,1045,233],{"id":233},[235,1047,1048,1054,1060,1066],{},[131,1049,1050,1053],{},[29,1051,1052],{},"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。",[131,1055,1056,1059],{},[29,1057,1058],{},"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。",[131,1061,1062,1065],{},[29,1063,1064],{},"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。",[131,1067,1068,1071],{},[29,1069,1070],{},"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。",[10,1073,1074,1075,1077],{},"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",[14,1076,870],{"href":869},"指南。",{"title":72,"searchDepth":272,"depth":272,"links":1079},[1080,1081,1082,1083,1084],{"id":22,"depth":272,"text":22},{"id":83,"depth":272,"text":83},{"id":92,"depth":272,"text":92},{"id":125,"depth":272,"text":126},{"id":233,"depth":272,"text":233},"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",[291,1089,883,1090],"what-is-a-stop-loss","what-is-backtesting",{"title":893,"description":1085},"zh\u002F5.glossary\u002Fwhat-is-slippage","滑点 (Slippage)","_usMaUZBN8x6rtKI1x17K10-_LxzY0duV2FLyHX73gE",[1096,1097],{"path":356},{"path":286},1793898063109]