[{"data":1,"prerenderedAt":864},["ShallowReactive",2],{"\u002Fzh\u002Fglossary\u002Fwhat-is-a-vps":3,"\u002Fzh\u002Fglossary\u002Fwhat-is-a-vps-related":223,"\u002Fzh\u002Fglossary\u002Fwhat-is-a-vps-langs":860},{"id":4,"title":5,"body":6,"category":206,"date":207,"description":208,"draft":209,"extension":210,"meta":211,"navigation":212,"path":213,"related":214,"seo":218,"stem":219,"term":220,"updated":221,"__hash__":222},"glossary\u002Fzh\u002F5.glossary\u002Fwhat-is-a-vps.md","交易中的 VPS 是什么？为什么正常运行时间决定结果",{"type":7,"value":8,"toc":198},"minimark",[9,19,23,26,37,40,43,51,54,62,65,68,84,87,91,136,151,159,162,190],[10,11,12,13,18],"p",{},"VPS（虚拟专用服务器，virtual private server）是从数据中心某台物理机器上分出来的一部分，租下后可以像远程电脑一样使用。在交易中，它只做一件事：让您的 ",[14,15,17],"a",{"href":16},"\u002Fglossary\u002Fwhat-is-metatrader","MetaTrader"," 终端不间断地运行，网络不掉线，机房还离经纪商的交易服务器很近。对自动化策略来说，VPS 算不上优化手段，它就是策略真正运行的环境。",[20,21,22],"h2",{"id":22},"工作原理",[10,24,25],{},"通过远程桌面连上 VPS，装好终端，登录，然后让它一直开着。之后您的笔记本合上、更新系统，甚至扔掉，策略都不受影响。花钱买的是两样东西：正常运行时间，以及离交易服务器足够近的物理距离；后者可以用往返延迟来衡量。",[27,28,33],"pre",{"className":29,"code":31,"language":32},[30],"language-text","Ping = 终端 → 交易服务器 → 终端，一次往返\n","text",[34,35,31],"code",{"__ignoreMap":36},"",[10,38,39],{},"多少算好，没有官方标准，而且要看相对值：同样一个数字，对持仓数天的策略无关紧要，对靠点差吃饭的策略却根本不能接受。方向则毫无争议：往返时间越长，成交价离策略想要的价格就越远。",[10,41,42],{},"人们最容易误解的，是持仓期间终端宕机会怎样。在 MetaTrader 中，挂在持仓上的止损和止盈保存在经纪商的服务器上，终端关了也照样触发。其他一切都在客户端运行，终端一停就全部停摆：新开仓、追踪止损、部分平仓、移至保本、EA（智能交易系统）存在内存里的虚拟止损、网格和回本逻辑。马丁格尔系统如果在加仓序列中途失去终端，并不会暂停：那一篮子仓位会一直无人管理，直到终端恢复；原本用来平衡整体仓位的那一单，根本不会开出来。",[10,44,45,46,50],{},"延迟和丢包还有另一层影响：它们会拉大 EA 决定的价格和实际成交价之间的差距。这个差距就是",[14,47,49],{"href":48},"\u002Fglossary\u002Fwhat-is-slippage","滑点","，策略参数怎么调都绕不开。",[20,52,53],{"id":53},"为什么重要",[10,55,56,57,61],{},"自动化策略的实盘结果，等于策略本身加上执行环境。同一套代码在两台机器上跑，两个账户的曲线会不一样，差别出在 ping、正常运行时间和成交质量上，而不在逻辑上。这也是",[14,58,60],{"href":59},"\u002Fglossary\u002Fwhat-is-backtesting","回测","结果往往好于实盘的原因之一：回测从不断线，从不漏掉一根 K 线，而且总按模型价格成交。",[10,63,64],{},"看别人的业绩记录时也要想到这一点。在 ping 为 40 ms 的 VPS 上跑出来的记录，换到 ping 300 ms、每晚还要装 Windows 更新的家用笔记本上，就算 EA 和经纪商都一样，也复制不出来。",[20,66,67],{"id":67},"数据显示了什么",[10,69,70,71,75,76,79,80,83],{},"在我们的平台上，自动化是常态，而不是例外。截至 2026 年 8 月，在 ShowMyTrades 上有交易历史的公开账户中，",[72,73,74],"strong",{},"自动交易占比的中位数为 99%","，",[72,77,78],{},"53.9% 的账户自动化程度超过 90%","。分布呈哑铃形，两头多、中间少：另有 ",[72,81,82],{},"42.2% 的账户自动化程度低于 10%","，介于两者之间的很少。交易者要么完全交给程序，要么完全手动。",[10,85,86],{},"也就是说，在这里公开业绩的账户中，超过一半要靠一个始终运行、保持连接、并且允许交易的终端。这些账户分布在 703 个不同的经纪商服务器上，已同步的交易共 15,436,464 笔。",[20,88,90],{"id":89},"在-showmytrades-上的位置","在 ShowMyTrades 上的位置",[10,92,93,94,97,98,104,105,108,109,112,113,116,117,120,121,124,125,104,128,131,132,135],{},"账户页面上的",[72,95,96],{},"终端与已打开的图表","模块，显示运行策略那台机器的实时状态：到交易服务器的 ",[14,99,101],{"href":100},"\u002Fglossary\u002Fwhat-is-terminal-ping",[72,102,103],{},"Ping","、",[72,106,107],{},"重传","（需要重新发送的网络数据包占比）、",[72,110,111],{},"终端版本号","、主机",[72,114,115],{},"内存","，以及是否允许",[72,118,119],{},"自动交易","和 ",[72,122,123],{},"DLL 导入","。连接徽章会显示",[72,126,127],{},"已连接",[72,129,130],{},"未连接交易服务器","或",[72,133,134],{},"离线","，旁边注明终端上次报告是多久以前；标题栏则统计您有几个终端在线。",[10,137,138,139,145,146,150],{},"下方的已打开图表表格，逐个列出每张图表的品种、时间周期、",[14,140,142],{"href":141},"\u002Fglossary\u002Fwhat-is-an-expert-advisor",[72,143,144],{},"智能交易系统","和指标，实际加载了什么一看便知，不必听信对方的说法。账户头部还有一个徽章，显示根据交易本身算出的自动化比例，例如 ",[147,148,149],"em",{},"99% autotrading","（99% 自动交易）。",[10,152,153,154,158],{},"一份如实交代自身条件的",[14,155,157],{"href":156},"\u002Fglossary\u002Fwhat-is-a-verified-track-record","经验证的业绩记录","，就该是这个样子：数字，连同产生这些数字的环境，放在同一个页面上。",[20,160,161],{"id":161},"常见误解",[163,164,165,172,178,184],"ul",{},[166,167,168,171],"li",{},[72,169,170],{},"VPS 不会让策略变得盈利","。它只是避免漏单和停机。如果策略本来就没有优势，机器越快，亏钱只会亏得越稳定。",[166,173,174,177],{},[72,175,176],{},"到网站的 ping 不等于到交易服务器的 ping","。要看的是到经纪商服务器的往返时间，也就是终端报告的那个数。VPS 选错了城市，可能还不如您桌上的电脑快。",[166,179,180,183],{},[72,181,182],{},"低延迟不能消除滑点","。点差、流动性和经纪商的执行模式照样起作用。VPS 去掉的是您这一侧的延迟，去不掉经纪商那一侧的。",[166,185,186,189],{},[72,187,188],{},"问题不在于“我的电脑从不死机”","。Windows 更新、休眠设置、网络运营商故障、停电，都会让终端断线，而且发生的时机与您的持仓毫无关联，完全随机。",[10,191,192,193,197],{},"想用同样审慎的眼光看完账户页面的其余部分，请读：",[14,194,196],{"href":195},"\u002Fguides\u002Freading-a-trading-account-dashboard","如何读懂交易账户控制台","。",{"title":36,"searchDepth":199,"depth":199,"links":200},2,[201,202,203,204,205],{"id":22,"depth":199,"text":22},{"id":53,"depth":199,"text":53},{"id":67,"depth":199,"text":67},{"id":89,"depth":199,"text":90},{"id":161,"depth":199,"text":161},"平台","2026-08-19T00:00:00.000Z","VPS 是一台租来的、全天候在线的服务器。您的电脑休眠、重启或断网时，交易终端照样在上面运行、保持连接。",false,"md",{},true,"\u002Fzh\u002Fglossary\u002Fwhat-is-a-vps",[215,216,217],"what-is-a-verified-track-record","what-is-slippage","what-is-backtesting",{"title":5,"description":208},"zh\u002F5.glossary\u002Fwhat-is-a-vps","VPS (虚拟专用服务器)","2026-09-04T00:00:00.000Z","DL1NAToJhz4AbS1tLr3efSqyW6iByPVGN45GYUmUlic",[224,414,642],{"id":225,"title":226,"body":227,"category":402,"date":207,"description":403,"draft":209,"extension":210,"meta":404,"navigation":212,"path":405,"related":406,"seo":410,"stem":411,"term":412,"updated":207,"__hash__":413},"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":228,"toc":395},[229,232,234,257,263,266,273,276,278,291,294,296,306,322,324,345,356,358,388],[10,230,231],{},"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。",[20,233,22],{"id":22},[10,235,236,237,241,242,246,247,251,252,256],{},"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 ",[14,238,240],{"href":239},"\u002Fglossary\u002Fwhat-is-an-mt4-investor-password","MT4 投资者密码","的直接连接、",[14,243,245],{"href":244},"\u002Fctrader-account-tracking","cTrader OAuth2 授权","，或 ",[14,248,250],{"href":249},"\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，",[14,253,255],{"href":254},"\u002Fglossary\u002Fwhat-is-myfxbook","Myfxbook"," 等第三方追踪工具也是这样连接 MetaTrader 账户的。",[27,258,261],{"className":259,"code":260,"language":32},[30],"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",[34,262,260],{"__ignoreMap":36},[10,264,265],{},"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。",[10,267,268],{},[269,270],"img",{"alt":271,"src":272},"“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",[10,274,275],{},"一份记录可能过了一关、没过另一关，所以两者分开统计。",[20,277,53],{"id":53},[10,279,280,281,285,286,290],{},"连接之后的一切都只是算术。只要输入能被改动，盈利因子、",[14,282,284],{"href":283},"\u002Fglossary\u002Fwhat-is-drawdown","回撤","和",[14,287,289],{"href":288},"\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。",[10,292,293],{},"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。",[20,295,67],{"id":67},[10,297,298,299,75,302,305],{},"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 ",[72,300,301],{},"65 个带有“业绩记录已验证”",[72,303,304],{},"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。",[10,307,308,309,313,314,317,318,321],{},"已验证不等于表现好。在有交易历史的公开账户中，",[14,310,312],{"href":311},"\u002Fglossary\u002Fwhat-is-time-weighted-return","时间加权收益率","的中位数为 ",[72,315,316],{},"+3.2%","，63.0% 为正，最深回撤的中位数为 ",[72,319,320],{},"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。",[20,323,90],{"id":89},[10,325,326,327,330,331,285,334,337,338,131,341,344],{},"两枚徽章都在每个账户页面头部的",[72,328,329],{},"认证","组里，位于各项数字上方：",[72,332,333],{},"业绩记录",[72,335,336],{},"交易权限","，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",[72,339,340],{},"真实账户",[72,342,343],{},"模拟账户","，这是根据经纪商服务器判断的，不是用户自己填的。",[10,346,347,348,351,352,355],{},"验证在账户设置中管理，",[72,349,350],{},"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。",[72,353,354],{},"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。",[20,357,161],{"id":161},[163,359,360,370,376,382],{},[166,361,362,197,365,369],{},[72,363,364],{},"指向某个追踪网站的链接，不等于验证",[14,366,368],{"href":367},"\u002Fguides\u002Fthird-party-verification-explained","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。",[166,371,372,375],{},[72,373,374],{},"“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。",[166,377,378,381],{},[72,379,380],{},"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。",[166,383,384,387],{},[72,385,386],{},"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。",[10,389,390,391,197],{},"两枚徽章的验证步骤详解：",[14,392,394],{"href":393},"\u002Fguides\u002Fhow-to-verify-your-account","如何验证您的交易账户",{"title":36,"searchDepth":199,"depth":199,"links":396},[397,398,399,400,401],{"id":22,"depth":199,"text":22},{"id":53,"depth":199,"text":53},{"id":67,"depth":199,"text":67},{"id":89,"depth":199,"text":90},{"id":161,"depth":199,"text":161},"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",[407,408,409],"what-is-an-mt4-investor-password","what-is-myfxbook","what-is-drawdown",{"title":226,"description":403},"zh\u002F5.glossary\u002Fwhat-is-a-verified-track-record","经验证的业绩记录 (Verified Track Record)","XXw0XywuXXvGTGPkURwZXnd-tURxBRrjPXvj-rr4h2k",{"id":415,"title":416,"body":417,"category":632,"date":207,"description":633,"draft":209,"extension":210,"meta":634,"navigation":212,"path":635,"related":636,"seo":638,"stem":639,"term":640,"updated":207,"__hash__":641},"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":418,"toc":625},[419,422,424,427,433,436,464,472,474,477,480,482,493,516,527,533,536,538,545,573,583,585,617],[10,420,421],{},"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。",[20,423,22],{"id":22},[10,425,426],{},"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：",[27,428,431],{"className":429,"code":430,"language":32},[30],"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",[34,432,430],{"__ignoreMap":36},[10,434,435],{},"输出靠不靠得住，取决于三项设置：",[163,437,438,444,455],{},[166,439,440,443],{},[72,441,442],{},"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。",[166,445,446,449,450,454],{},[72,447,448],{},"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",[14,451,453],{"href":452},"\u002Fglossary\u002Fwhat-is-a-spread","点差","，已经是另一个策略了。",[166,456,457,460,461,463],{},[72,458,459],{},"滑点与成交逻辑","。测试器默认订单都按请求价成交。",[14,462,49],{"href":48},"、拒单、重新报价和部分成交，在测试器里统统不存在。",[10,465,466,467,471],{},"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",[14,468,470],{"href":469},"\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。",[20,473,53],{"id":53},[10,475,476],{},"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。",[10,478,479],{},"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。",[20,481,67],{"id":67},[10,483,484,485,488,489,492],{},"ShowMyTrades 以公开页面发布了 ",[72,486,487],{},"1 万+ 个账户","，背后是 ",[72,490,491],{},"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。",[10,494,495,496,499,500,504,505,75,508,504,512,515],{},"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 ",[72,497,498],{},"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，",[14,501,503],{"href":502},"\u002Fglossary\u002Fwhat-is-a-profit-factor","盈利因子","中位数为 ",[72,506,507],{},"1.28",[14,509,511],{"href":510},"\u002Fglossary\u002Fwhat-is-the-sharpe-ratio","夏普比率",[72,513,514],{},"0.05","。测试报告里的数字，真正要比的就是这些。",[10,517,518,519,522,523,526],{},"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 ",[72,520,521],{},"$4,782,670","，库存费 ",[72,524,525],{},"$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。",[10,528,529],{},[269,530],{"alt":531,"src":532},"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",[10,534,535],{},"测试器用固定点差、又不算库存费，三项成本就漏掉了两项。",[20,537,90],{"id":89},[10,539,540,541,544],{},"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",[14,542,333],{"href":543},"\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。",[10,546,547,548,285,551,554,555,558,559,561,562,564,565,568,569,572],{},"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",[72,549,550],{},"佣金合计",[72,552,553],{},"已付库存费合计","，以账户货币显示策略的真实成本；",[72,556,557],{},"终端","面板，报告与交易服务器之间的 ",[72,560,103],{}," 和",[72,563,107],{},"，也就是您实际成交时所处的条件；还有",[72,566,567],{},"交易历史与余额变化","，它的",[72,570,571],{},"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。",[10,574,575,578,579,582],{},[72,576,577],{},"业绩记录已验证","徽章（65 个账户持有）和",[72,580,581],{},"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。",[20,584,161],{"id":161},[163,586,587,594,605,611],{},[166,588,589,590,593],{},"“",[72,591,592],{},"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。",[166,595,589,596,599,600,604],{},[72,597,598],{},"它测试了十年。","”",[14,601,603],{"href":602},"\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。",[166,606,589,607,610],{},[72,608,609],{},"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。",[166,612,589,613,616],{},[72,614,615],{},"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。",[10,618,619,620,624],{},"无论是测试结果还是实盘数字，在相信它之前，先按",[14,621,623],{"href":622},"\u002Fguides\u002Fverifying-trading-performance-claims","如何验证交易业绩宣传","逐项核对一遍。",{"title":36,"searchDepth":199,"depth":199,"links":626},[627,628,629,630,631],{"id":22,"depth":199,"text":22},{"id":53,"depth":199,"text":53},{"id":67,"depth":199,"text":67},{"id":89,"depth":199,"text":90},{"id":161,"depth":199,"text":161},"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",[637,215,216],"what-is-curve-fitting",{"title":416,"description":633},"zh\u002F5.glossary\u002Fwhat-is-backtesting","回测 (Backtesting)","JFA3C0OuDmaFAnqzJS-I02fA4QcPMlLuq0a6Gjsakv8",{"id":643,"title":644,"body":645,"category":847,"date":207,"description":848,"draft":209,"extension":210,"meta":849,"navigation":212,"path":850,"related":851,"seo":855,"stem":856,"term":857,"updated":858,"__hash__":859},"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":646,"toc":840},[647,650,652,658,661,667,673,684,691,693,704,710,712,729,740,742,745,805,807,834],[10,648,649],{},"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。",[20,651,22],{"id":22},[27,653,656],{"className":654,"code":655,"language":32},[30],"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",[34,657,655],{"__ignoreMap":36},[10,659,660],{},"滑点有三个来源。",[10,662,663,666],{},[72,664,665],{},"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。",[10,668,669,672],{},[72,670,671],{},"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。",[10,674,675,678,679,683],{},[72,676,677],{},"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",[14,680,682],{"href":681},"\u002Fglossary\u002Fwhat-is-a-stop-loss","止损","）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。",[10,685,686,687,690],{},"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。",[72,688,689],{},"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。",[20,692,53],{"id":53},[10,694,695,696,698,699,703],{},"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",[14,697,60],{"href":59},"，假设了一种根本不存在的执行方式；",[14,700,702],{"href":701},"\u002Fglossary\u002Fwhat-is-average-trade-length","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。",[10,705,706,707,709],{},"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",[14,708,453],{"href":452},"和佣金之外另算的，不是替代它们。",[20,711,67],{"id":67},[10,713,714,715,75,717,719,720,723,724,728],{},"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，",[72,716,74],{},[72,718,78],{},"（42.2% 低于 10%）。这些账户连接了 ",[72,721,722],{},"703 个不同的经纪商服务器","：",[14,725,727],{"href":726},"\u002Fguides\u002Fsame-strategy-different-brokers","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。",[10,730,731,732,735,736,739],{},"持仓时间又放大了这种风险。这些账户的",[72,733,734],{},"持仓时间中位数为 2.4 小时","，中位数账户已平仓 ",[72,737,738],{},"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。",[20,741,90],{"id":89},[10,743,744],{},"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。",[163,746,747,774,795],{},[166,748,749,751,752,754,755,759,760,765,766,104,768,104,770,120,772,197],{},[72,750,557],{},"面板（MT5 上为",[72,753,96],{},"）会显示 ",[14,756,757],{"href":100},[72,758,103],{},"，即终端与经纪商交易服务器之间的往返时间，以及",[14,761,763],{"href":762},"\u002Fglossary\u002Fwhat-is-packet-retransmission",[72,764,107],{},"，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",[72,767,111],{},[72,769,115],{},[72,771,119],{},[72,773,123],{},[166,775,776,778,779,781,782,285,785,788,789,285,791,794],{},[72,777,567],{},"下的",[72,780,571],{},"表格，每个订单都有",[72,783,784],{},"开仓价",[72,786,787],{},"平仓价","，旁边是",[72,790,682],{},[72,792,793],{},"止盈","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。",[166,796,797,800,801,804],{},[72,798,799],{},"高级统计","中的",[72,802,803],{},"最差交易（点）"," 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。",[20,806,161],{"id":161},[808,809,810,816,822,828],"ol",{},[166,811,812,815],{},[72,813,814],{},"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。",[166,817,818,821],{},[72,819,820],{},"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。",[166,823,824,827],{},[72,825,826],{},"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。",[166,829,830,833],{},[72,831,832],{},"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。",[10,835,836,837,839],{},"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",[14,838,196],{"href":195},"指南。",{"title":36,"searchDepth":199,"depth":199,"links":841},[842,843,844,845,846],{"id":22,"depth":199,"text":22},{"id":53,"depth":199,"text":53},{"id":67,"depth":199,"text":67},{"id":89,"depth":199,"text":90},{"id":161,"depth":199,"text":161},"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",[852,853,854,217],"what-is-a-spread","what-is-a-stop-loss","what-is-a-vps",{"title":644,"description":848},"zh\u002F5.glossary\u002Fwhat-is-slippage","滑点 (Slippage)","2026-08-29T00:00:00.000Z","_usMaUZBN8x6rtKI1x17K10-_LxzY0duV2FLyHX73gE",[861,863],{"path":862},"\u002Fglossary\u002Fwhat-is-a-vps",{"path":213},1793898070185]