[{"data":1,"prerenderedAt":760},["ShallowReactive",2],{"\u002Fzh\u002Fguides\u002Fsame-strategy-different-brokers":3,"\u002Fzh\u002Fguides\u002Fsame-strategy-different-brokers-surround":745,"\u002Fzh\u002Fguides\u002Fsame-strategy-different-brokers-langs":756},{"id":4,"title":5,"authors":6,"badge":12,"body":14,"date":732,"description":733,"draft":734,"extension":735,"image":736,"meta":738,"navigation":739,"path":740,"seo":741,"stem":742,"updated":743,"__hash__":744},"posts\u002Fzh\u002F3.guides\u002F28.same-strategy-different-brokers.md","同一策略、两家经纪商、不同结果：如何做对比测试（{YEAR} 年版）",[7],{"name":8,"to":9,"avatar":10},"ShowMyTrades Team","https:\u002F\u002Fshowmytrades.com",{"src":11},"\u002Ficon-512.png",{"label":13},"经纪商",{"type":15,"value":16,"toc":719},"minimark",[17,21,24,27,32,35,124,127,138,142,145,195,202,206,209,304,307,343,357,361,364,371,386,399,406,413,419,423,431,438,490,497,508,514,518,531,534,540,546,569,575,593,597,629,633,639,650,656,662,668,672,699,702,712],[18,19,20],"p",{},"同一套策略，您放在两个账户上跑：逻辑一样，参数一样，品种也一样。结果一条曲线节节上涨，另一条半天爬不动，您想知道问题是不是出在经纪商身上。",[18,22,23],{},"通常的做法是看一眼两边的余额，凭感觉认定哪家经纪商更差，然后把资金转走。结果新账户照样跑输，同样的事又来一遍。",[18,25,26],{},"另一种做法是做一次受控实验：两个账户同时运行，两边用同样的筛选条件，再把差异拆开，一项一项找原因。本指南讲的就是这个方法。它不会告诉您该用哪家经纪商，这个问题什么都替您回答不了，净值曲线也不能。它告诉您的是：钱到底去了哪里。",[28,29,31],"h2",{"id":30},"换了经纪商究竟变了什么","🧩 换了经纪商，究竟变了什么",[18,33,34],{},"策略一模一样，环境却变了。会变的变量有五个，而有记录可查的只有两个。",[36,37,38,54],"table",{},[39,40,41],"thead",{},[42,43,44,48,51],"tr",{},[45,46,47],"th",{},"变的是什么",[45,49,50],{},"体现在哪里",[45,52,53],{},"看不看得见",[55,56,57,72,85,98,111],"tbody",{},[42,58,59,66,69],{},[60,61,62],"td",{},[63,64,65],"strong",{},"点差",[60,67,68],{},"开仓价和平仓价，每笔交易少赚几个点",[60,70,71],{},"不单独列出，藏在毛利润里",[42,73,74,79,82],{},[60,75,76],{},[63,77,78],{},"佣金",[60,80,81],{},"开仓和平仓时按手数收取",[60,83,84],{},"交易表格里有专门一列",[42,86,87,92,95],{},[60,88,89],{},[63,90,91],{},"库存费",[60,93,94],{},"每持仓过夜一晚扣除或入账",[60,96,97],{},"有专门一列，但很容易被忽略",[42,99,100,105,108],{},[60,101,102],{},[63,103,104],{},"执行与滑点",[60,106,107],{},"成交价比请求价差",[60,109,110],{},"看不见，只体现为少掉的点数",[42,112,113,118,121],{},[60,114,115],{},[63,116,117],{},"品种规格与服务器时间",[60,119,120],{},"合约规模、交易时段边界、日 K 线收盘时间",[60,122,123],{},"连哪些交易会被触发都会变",[18,125,126],{},"五项里有两项印在您的交易表格上，另外三项没有，差异也正因为这样常常被算到错误的原因头上。一套策略每笔交易被点差和滑点吃掉 0.4 点，却在哪里都看不到点差费用，只是不声不响地少赚了。",[18,128,129,130,133,134,137],{},"还有第六个变量常被人忘记：",[63,131,132],{},"服务器","。连接到 ShowMyTrades 的 1 万+ 个账户分布在 ",[63,135,136],{},"703 个不同的经纪商服务器","上，大型经纪商通常不止一个服务器。同一品牌下的两个账户，可能在不同的服务器上，交易条件也不一样。账户页面上的经纪商徽章，提示框里会显示交易服务器名称：先核对一下，再假定两边可以比较。",[28,139,141],{"id":140},"设计一个经得起推敲的对比","🧪 设计一个经得起推敲的对比",[18,143,144],{},"受控对比，说白了就是提前把各种借口都堵死。一共六项决定，都要在第一笔交易之前做好：",[146,147,148,155,167,173,179,189],"ol",{},[149,150,151,154],"li",{},[63,152,153],{},"同时运行，不能先后进行","。3 月和 4 月是两个不同的市场。如果账户 A 在一个月交易、账户 B 在另一个月交易，您测到的其实是波动率，却把它当成了经纪商的差别。",[149,156,157,160,161,166],{},[63,158,159],{},"要么都用真实账户，要么都用模拟账户","。模拟账户的成交偏偏在真实成交最吃亏的行情里显得过于理想，这也是",[162,163,165],"a",{"href":164},"\u002Fguides\u002Fverified-vs-unverified-track-records","经纪商同步的记录比自行报告的更可靠","的原因之一。拿真实账户和模拟账户比，算不上测试。",[149,168,169,172],{},[63,170,171],{},"品种相同，时段相同","。如果某个品种只有一家经纪商有，就把它从两边一起剔除，别让它混进总数。",[149,174,175,178],{},[63,176,177],{},"仓位按净值比例设定，不用固定手数","。同样的手数放在两个不同的余额上，承担的风险就不一样，而风险对结果的影响比成本还大。",[149,180,181,184,185,188],{},[63,182,183],{},"样本量事先定好","。看结果之前就定好交易笔数，免得答案一偏向您中意的经纪商就立刻收手。作为参考：ShowMyTrades 上公开账户的中位数，整个历史共有 ",[63,186,187],{},"171 笔已平仓交易","。",[149,190,191,194],{},[63,192,193],{},"先把问题写下来","。“这套策略在哪家经纪商每手更便宜”是能回答的问题，“哪家经纪商更好”不是。",[18,196,197,198,201],{},"同时运行两个账户并不稀奇：699 位 ShowMyTrades 用户不止有一个账户，",[63,199,200],{},"429 位在不止一家经纪商开了账户","。但很少有人把这种安排变成一次真正的测量。",[28,203,205],{"id":204},"把差距拆成几个部分","🧮 把差距拆成几个部分",[18,207,208],{},"假设这个季度结束时，账户 A 领先账户 B。这个差距由六部分加总而成，要做的就是把它们逐一拆开，直到剩下的最后一项小到可以忽略。",[36,210,211,224],{},[39,212,213],{},[42,214,215,218,221],{},[45,216,217],{},"差距的组成",[45,219,220],{},"怎么测",[45,222,223],{},"怎么排除影响",[55,225,226,239,252,265,278,291],{},[42,227,228,233,236],{},[60,229,230],{},[63,231,232],{},"仓位大小",[60,234,235],{},"两边每笔交易的手数与净值之比",[60,237,238],{},"用百分比控制风险，不用固定手数",[42,240,241,246,249],{},[60,242,243],{},[63,244,245],{},"样本噪声",[60,247,248],{},"去掉最好和最差的一笔后，结论会不会反转",[60,250,251],{},"增加交易笔数，时间窗口不变",[42,253,254,259,262],{},[60,255,256],{},[63,257,258],{},"已披露的成本",[60,260,261],{},"佣金合计与已付库存费合计，除以总手数",[60,263,264],{},"无需排除：直接读数即可",[42,266,267,272,275],{},[60,268,269],{},[63,270,271],{},"报价质量",[60,273,274],{},"同一品种、同一时间窗口内每笔交易的平均点数",[60,276,277],{},"一次只看一个品种",[42,279,280,285,288],{},[60,281,282],{},[63,283,284],{},"基础设施",[60,286,287],{},"每个终端的 Ping 和重传",[60,289,290],{},"先解决托管问题，再怪报价",[42,292,293,298,301],{},[60,294,295],{},[63,296,297],{},"交易品种规格",[60,299,300],{},"合约规模、品种能否交易、服务器时区",[60,302,303],{},"把对不上的品种从两边一起剔除",[18,305,306],{},"按顺序一项项往下排查。每排除一项，下一项就看得更清楚；排到最后还归不了因的部分，就是实实在在的残差。",[18,308,309,310,313,314,317,318,321,322,326,327,331,332,335,336,339,340,188],{},"已披露成本这一项最简单，因为系统已经替您加好了。",[63,311,312],{},"佣金合计","、",[63,315,316],{},"已付库存费合计","和",[63,319,320],{},"总手数","都在每个账户页面的",[162,323,325],{"href":324},"\u002Fguides\u002Freading-a-trading-account-dashboard","高级统计模块","里，",[162,328,330],{"href":329},"\u002Fguides\u002Fbroker-cost-audit","每手成本","只需做一次除法。这笔钱可不少：在 ShowMyTrades 的公开账户中，这两项加起来是 ",[63,333,334],{},"$4,782,670 的佣金","和 ",[63,337,338],{},"$862,547 的库存费","，对应 ",[63,341,342],{},"1,724,575 手",[18,344,345,348,349,352,353,356],{},[162,346,91],{"href":347},"\u002Fguides\u002Fswap-and-carry-trade","值得多看一眼：同样的仓位，库存费是各家经纪商之间差别最大的一项，因为它由经纪商自己定，而不是由市场定。在有库存费记录的 ShowMyTrades 账户中，",[63,350,351],{},"86.3% 的账户净支付库存费","。如果您的策略会隔夜持仓，就把同一组交易跑两遍，一次打开、一次关闭",[63,354,355],{},"计算盈利时包含库存费","开关：两次的差就是差距中的融资部分，和执行毫无关系。",[28,358,360],{"id":359},"每笔交易的点数点差和滑点藏在哪里","🎯 每笔交易的点数：点差和滑点藏在哪里",[18,362,363],{},"已披露的成本可以直接加总。点差和成交质量却从来不以费用的形式收取：它们早已含在您拿到的价格里，只会表现为该到手却没到手的点数。",[18,365,366,367,370],{},"高级统计里有一行",[63,368,369],{},"期望值","，并排显示两个数字：每笔交易的平均点数和平均金额。跨经纪商比较要看点数这一半，因为它不受仓位大小和账户货币影响。请在同一品种、同一时间窗口下读取：",[372,373,374,380],"ul",{},[149,375,376,377],{},"账户 A：",[63,378,379],{},"每笔交易 +2.1 点",[149,381,382,383],{},"账户 B：",[63,384,385],{},"每笔交易 +1.4 点",[18,387,388,389,393,394,398],{},"这 0.7 点的差距既不是技术，也不是行情，因为两个账户是在同一时间交易同样的信号。它就是点差加滑点。在 USD 账户上交易 90 手 EURUSD，每手每点约 $10，这个差距大约就是 $630，却从来没有作为一笔费用出现过。用",[162,390,392],{"href":391},"\u002Ftools\u002Fpip-calculator","点值计算器","可以按您自己的品种和手数换算，其余的计算都在",[162,395,397],{"href":396},"\u002Ftools","交易计算器","里。",[18,400,401,402,405],{},"有两点要注意。黄金的一点不等于 EURUSD 的一点，所以只能逐个品种比较。另外样本要够：ShowMyTrades 公开账户的持仓时间中位数是 ",[63,403,404],{},"2.4 小时","，这里的大多数系统一两个月内就能攒够可用的样本，节奏慢的波段系统则要久得多。",[18,407,408],{},[409,410],"img",{"alt":411,"src":412},"四档历史长度对比：三个月算样本，六个月算迹象，十二个月算论据，二十四个月及以上才算业绩记录；对照公开账户的中位数（171 笔已平仓交易，平均每笔 2.4 小时）","\u002Fimages\u002Farticles\u002Fsample-size.svg",[18,414,415],{},[416,417,418],"em",{},"先数清两个账户各有多少笔交易，再去谈它们之间的差距说明了什么。",[28,420,422],{"id":421},"️-执行证据终端报告了什么","🖥️ 执行证据：终端报告了什么",[18,424,425,426,430],{},"成本只能解释一部分差距，另一部分要靠",[162,427,429],{"href":428},"\u002Fglossary\u002Fwhat-is-server-latency","延迟","来解释，而这部分最常被算到错误的一方头上。",[18,432,433,434,437],{},"对于通过我们的 EA（智能交易系统）连接的 MT4 和 MT5 账户，您控制台账户页面上的",[63,435,436],{},"终端与已打开的图表","模块会显示终端自己发来的数据：",[372,439,440,454,463,472,478,484],{},[149,441,442,448,449,453],{},[63,443,444],{},[162,445,447],{"href":446},"\u002Fglossary\u002Fwhat-is-terminal-ping","Ping","：终端到经纪商交易服务器的往返时间。低于 50 ms 为优秀；高于 200 ms 可能意味着",[162,450,452],{"href":451},"\u002Fglossary\u002Fwhat-is-order-execution","订单执行","变慢、滑点增多。",[149,455,456,462],{},[63,457,458],{},[162,459,461],{"href":460},"\u002Fglossary\u002Fwhat-is-packet-retransmission","重传","：需要重新发送的网络数据包所占比例。接近 0% 说明连接稳定；长期偏高，则说明网络不稳定或交易服务器离得太远。",[149,464,465,317,468,471],{},[63,466,467],{},"终端版本号",[63,469,470],{},"内存","：策略实际运行的环境。",[149,473,474,477],{},[63,475,476],{},"自动交易：允许或禁止","：一旦禁止，EA 仍在运行，但无法开仓、修改或平仓。",[149,479,480,483],{},[63,481,482],{},"DLL 导入","：终端是否允许调用外部库。",[149,485,486,489],{},[63,487,488],{},"连接","：已连接、未连接交易服务器或离线，并附最近一次更新的时间。",[18,491,492,493,496],{},"这样，一句含糊的抱怨就能拆成三种诊断。一个账户 Ping 220 ms、重传明显，另一个只有 18 ms：这是托管问题，该换一台离得更近的 VPS，而不是换经纪商。两边 Ping 一样，点数却有差距：这是报价和执行的问题。自动交易显示",[63,494,495],{},"禁止","：缺掉的交易就直接有了解释，跟经纪商没有关系。",[18,498,499,500,503,504,507],{},"有四个限制，直说在前。这个模块",[63,501,502],{},"仅所有者可见","：只在您的控制台里，从不出现在公开页面上。它需要 ",[63,505,506],{},"EA"," 连接，只用投资者密码同步的账户，没有任何终端为它上报数据。cTrader 和 TradeLocker 是纯 REST 集成，没有终端，所以在这两个平台上只能比较成本和点数。另外，MT4 从不发送已打开图表的列表，所以那张表只有 MT5 才有。",[18,509,510,513],{},[63,511,512],{},"有一条界线始终不变","。终端上报的软件信息，一律保密。我们从不推断、估算或公开某个账户跑的是哪款商业 EA，平台上任何地方都没有按 EA 划分的页面或汇总数据。魔术号、标签和交易注释都归账户所有者。",[28,515,517],{"id":516},"在-showmytrades-里操作","🧰 在 ShowMyTrades 里操作",[18,519,520,523,524,527,528,188],{},[63,521,522],{},"两个账户分别连接","。MT4 和 MT5 可以通过我们的 EA 或只读的投资者密码连接，cTrader 通过 OAuth 授权，TradeLocker 用邮箱、密码和服务器连接。连好后，每个账户头部都会显示平台、经纪商（提示框里有交易服务器）、货币、杠杆，以及交易模式徽章，比如 ",[416,525,526],{},"99% autotrading","。99% 正是这里自动交易占比的中位数，",[63,529,530],{},"53.9% 的账户自动化程度超过 90%",[18,532,533],{},"先比这个徽章：如果一边完全自动化，另一边大多手动，那跑的就不是同一套策略，后面的比较全都作废。",[18,535,536,539],{},[63,537,538],{},"保持两个独立页面，不要合并成投资组合","。投资组合功能确实有，目前有 182 个在用，但它会把多个账户合并成一条曲线，和对比需要的恰好相反。",[18,541,542,545],{},[63,543,544],{},"建一个自定义分析预设，两个账户都用它","。面板可以按日期范围、品种、魔术号（按您自己的标签列出）、交易类型、星期、交易时段、手数范围、盈亏范围和持续时间范围筛选，另有一个包含库存费的开关。在账户 A 上保存预设，再在账户 B 上设置完全相同的筛选条件。时间窗口相同，品种相同，交易子集相同：这就是您的对照组。",[18,547,548,551,552,313,555,313,557,313,559,313,562,317,565,568],{},[63,549,550],{},"在交易表格里打开成本相关的列","。净盈亏一直显示；通过列选择器还能加上",[63,553,554],{},"盈利（毛额）",[63,556,91],{},[63,558,78],{},[63,560,561],{},"点数",[63,563,564],{},"持续时间",[63,566,567],{},"魔术号","。毛额和净额特意分开显示，因为两者之间的差，正是您要找的答案。",[18,570,571,574],{},[63,572,573],{},"然后按这个顺序看四个模块","：高级统计，看手数、佣金、库存费和期望值；货币对统计，确认差异不是集中在某一个品种上；终端与已打开的图表，看延迟；最后是月度收益表，看差距是一直存在，还是只是某一周表现差。",[18,576,577,580,581,584,585,588,589,188],{},[63,578,579],{},"答案值得分享，就公开出来","。验证按账户进行（这里有 ",[63,582,583],{},"65 个账户","获得“业绩记录已验证”，",[63,586,587],{},"264 个","获得“交易权限已验证”），在已验证账户上公开的对比，读者可以自行核实。",[162,590,592],{"href":591},"\u002Fguides\u002Fhow-to-verify-your-account","如何拿到这两个徽章",[28,594,596],{"id":595},"️-这种对比容易在哪里出错","⚠️ 这种对比容易在哪里出错",[146,598,599,605,611,617,623],{},[149,600,601,604],{},[63,602,603],{},"两个账户先后运行","。最常见，也最致命。不同的月份就是不同的市场，这个测试从此就救不回来了。",[149,606,607,610],{},[63,608,609],{},"测试中途改策略","。只要一边改了一个参数，此后的每个数字都属于另一个实验。要么把两边的配置都冻结，要么重新开始。",[149,612,613,616],{},[63,614,615],{},"混进了对不上的品种","。只有一家经纪商提供、或合约规模不同的品种，会不声不响地扛下全部差异。在自定义分析里把它从两边一起剔除。",[149,618,619,622],{},[63,620,621],{},"把自己的托管问题怪到经纪商头上","。Ping 200 ms 加上明显的重传，可能是您 VPS 的位置问题，而不是经纪商的报价。终端与已打开的图表会告诉您是哪一种；何况换台服务器，比转移资金便宜得多。",[149,624,625,628],{},[63,626,627],{},"凭寥寥几笔交易下结论","。这里公开账户的已平仓交易中位数是 171 笔，40 笔的样本只是噪声；靠一笔异常盈利单定输赢的对比也一样。",[28,630,632],{"id":631},"常见问题","❓ 常见问题",[18,634,635,638],{},[63,636,637],{},"两个账户要跑多久才能下结论？","\n跑到一笔交易无法让结论反转为止。有个实用的检验办法：两边各去掉最好的一笔和最差的一笔，再重算对比。如果结论变了，说明样本太小，不管日历上已经过了多久。",[18,640,641,644,645,649],{},[63,642,643],{},"两个账户的余额需要一样吗？","\n不需要，但风险逻辑要一样。仓位按净值百分比设定，不要用固定手数，否则比较的是仓位管理，而不是经纪商。用",[162,646,648],{"href":647},"\u002Ftools\u002Fposition-size-calculator","仓位计算器","可以让两边保持一致。",[18,651,652,655],{},[63,653,654],{},"我是主观交易，不用 EA，这个方法还适用吗？","\n部分适用。手动交易没法保证入场完全一致，所以仓位大小和样本这两项的噪声会一直存在。仍然有效的有两样：每手成本，不管交易由什么触发都成立；还有两边在同一时段都交易过的品种上的点数期望值。",[18,657,658,661],{},[63,659,660],{},"cTrader 或 TradeLocker 账户能和 MT5 账户比吗？","\n成本、点数和结果可以，因为每个平台都是按真实成交、用同样的方法计算的。终端与已打开的图表不行，它只存在于运行我们 EA 的 MetaTrader 账户上。",[18,663,664,667],{},[63,665,666],{},"ShowMyTrades 会告诉我哪家经纪商更好吗？","\n不会，也不该由我们来说。我们展示的是您的账户花了多少成本、执行得怎么样。“更好”取决于您的持仓周期、交易品种和仓位大小。这些数字能帮您做决定，但不会替您做决定。",[28,669,671],{"id":670},"相关指南","🔗 相关指南",[372,673,674,680,686,692],{},[149,675,676,679],{},[162,677,678],{"href":324},"如何读懂交易账户控制台","：这次对比用到的所有模块",[149,681,682,685],{},[162,683,684],{"href":164},"经验证与未经验证的业绩记录","：为什么只有经纪商同步的数据才有可比性",[149,687,688,691],{},[162,689,690],{"href":591},"如何验证您的交易账户","：让您公开的账户拿到这两个徽章",[149,693,694,698],{},[162,695,697],{"href":696},"\u002Fguides\u002Fforex-calculators-guide","外汇计算器指南","：上面这些数字背后按手、按点的计算方法",[700,701],"hr",{},[18,703,704,188,707,711],{},[63,705,706],{},"看看真实的成本是什么样子",[162,708,710],{"href":709},"\u002Fexplore","探索","里链接的每个账户都与经纪商同步且公开，佣金和库存费一项不少。",[18,713,714,718],{},[162,715,717],{"href":716},"\u002Fsignup","免费连接您的两个账户","，用自己的数据做一次对比。",{"title":720,"searchDepth":721,"depth":721,"links":722},"",2,[723,724,725,726,727,728,729,730,731],{"id":30,"depth":721,"text":31},{"id":140,"depth":721,"text":141},{"id":204,"depth":721,"text":205},{"id":359,"depth":721,"text":360},{"id":421,"depth":721,"text":422},{"id":516,"depth":721,"text":517},{"id":595,"depth":721,"text":596},{"id":631,"depth":721,"text":632},{"id":670,"depth":721,"text":671},"2026-08-19T00:00:00.000Z","同一套策略，两家经纪商，跑出两条不同的曲线。本文讲如何让两个账户并行运行，并把差距拆解到仓位大小、成本、报价质量和执行延迟上。",false,"md",{"src":737},"\u002Fimages\u002Fblog\u002Fsame-strategy-different-brokers-hero.svg",{},true,"\u002Fzh\u002Fguides\u002Fsame-strategy-different-brokers",{"title":5,"description":733},"zh\u002F3.guides\u002F28.same-strategy-different-brokers","2026-09-10T00:00:00.000Z","r-DlygDkBAMrX3kGimFVoJ5BzsFeofCfG0On30hiq1U",[746,751],{"title":747,"path":748,"stem":749,"description":750,"children":-1},"美分账户：数字为什么大得惊人，实际又代表什么（{YEAR} 年版）","\u002Fzh\u002Fguides\u002Fcent-accounts-explained","zh\u002F3.guides\u002F27.cent-accounts-explained","交易者只入金 $100，美分账户却显示 10,000。本文讲清计价单位怎么回事、手数为什么跟着缩放，以及哪些指标完全不受它影响。",{"title":752,"path":753,"stem":754,"description":755,"children":-1},"买 EA 之前，如何识破虚假业绩（{YEAR} 年版）","\u002Fzh\u002Fguides\u002Fspotting-fake-ea-results","zh\u002F3.guides\u002F29.spotting-fake-ea-results","EA 销售页面给您看的是回测，不是业绩记录。本文教您看懂两者之间的落差：曲线拟合、tick 质量、手数递增、账户重置，外加一份买家十分钟自查清单。",[757,759],{"path":758},"\u002Fguides\u002Fsame-strategy-different-brokers",{"path":740},1793898037849]