※ 引述《fihalon (道)》之铭言： : http://i.imgur.com/UEuKNNK.png : http://i.imgur.com/UE9gNEr.png : http://i.imgur.com/06MB21n.png : http://i.imgur.com/F3wyaLj.png : MDD金额会创高 但DD%已经很久没创高了 实单帐户亦是如此 一点小意见,我不太确定你的标的是甚麽,看实单帐户先推测是台指 2012的报酬率是149% http://i.imgur.com/UE9gNEr.png 2012的报酬率大概有80%集中在10月,看数字10月大概赚了100% 应该不是当冲系统(不然金额累积走势不应该出现喷出的状况) http://i.imgur.com/PWoFsJO.png 台股走势(切周线图) 推测是10月中到10月底的走势,指数跌不到10%,杠杆估计超过10倍 http://tw.stock.yahoo.com/t/idx.php 建议跑回测的时候自己加点压力测试,以台指为例, 就是系统加码到杠杆最大的时候,模拟碰到319或430时的状况, 这种状况下,若我没猜错你会crash... (我不需要知道你实际的系统,只要评估你可能的最大杠杆就能得出这个结论, 如果你有避险或真的是当冲系统应该就没差,除非交易所挂掉 XD) 我以前也有一阵子很热衷跑回测,发现一个很有趣的现象,就是测试区间拉长时, 在某些值绩效会差很大,看逐笔交易就会发现绩效特别好的参数, 碰上系统性风险都刚好做对边...背後的意义大家可以思考一下 最後转篇Ed Seykota的文: System Trade Hi Ed! Today I bought 4 contracts of Cocoa. Entry 890. SL 772. Total Risk: 472 {4*(890-772)} Euro. My system is now up and running. When my SL-risk decreases I will put on new trades so my total SL-risk always is between 40-50% of my total equity. Can you suggest how to test this system more careful. I'm not sure that my money management is optimal. == Hmmm ... you are risking 40-50% of your Equity on one trade. Professional trend traders typically risk around one percent as much as you risk. You can test your system a couple ways. 1. You can back-test it on a computer and notice it goes broke on small whipsaws. 2. You can run it in real-time, as you are doing. When you go broke, your test is complete. -- FB: http://0rz.tw/l3Kcq █◣ ▊ ◥\◣◣\◣▼▼◣◣█◣ ▌ ╬╬ █银ˊ どんだけ── ◥█◣ )) ▲▲██▲╴▲██▲ ▲ ▌ "囧█ ╬╬ █ˊ魂 好文要推── ◥█◣ ◢ ▏ ▼██▼ ▼██▼▏███▌m@▼▄ ╬╬ EVERYBODY SAY ◥█◣█ ▼█▇▇█‥ ▇▇█▏◣ ▌ ▲◇" ╬╬ 市场求生手册─ (( ╲ ╲ ◣▼ ▇▇▇ ▼ ◣ ▌▆▆▆ ╬╬ http://stasistw.blogspot.com/╲█╲█ ◥██████ ◤████▆▅▄▃▂▁ --

※ 发信站: 批踢踢实业坊(ptt.cc) ◆ From: 58.115.27.162