海龟交易法则

Way of the Turtle

柯蒂斯·费思 · Curtis Faith · 2007

系统系统化·技术与量化S1 · E07 · 第 7 集难度 ●●○○○290 页通读约 7 小时P14

规则两周教完，费思头一个月赚7.8万美元，同学同堂听课却空仓，书讲的是那个差距。Same two-week class, same rules: one Turtle made $78,000 in month one, the rest sat out. The book is about that gap.

本集角度 · 普通人照着系统做，能不能变成交易员

📖 微信读书 · 打开书页本季全部 10 集

ABOUT这本书

1983年，芝加哥期货交易员理查德·丹尼斯登整版广告招人，选了十几个素人，培训两周，然后给真钱。费思当年19岁，班里最小，账户最大，四年给丹尼斯赚了三千多万美元。前半段是回忆录：1984年1月取暖油，他一个人满仓12张合约，其他人要么没进要么早跑，头一个月收益是同学的三倍。后半段把规则摊开：20日突破进场、按ATR算头寸、最多四个单位、单笔风险2%左右，附原版规则全文，还有一章讲回测怎么骗人。数据来自他自己的软件Trading Blox，28个美国期货市场十年。

In 1983 Chicago futures trader Richard Dennis ran a full-page newspaper ad, picked a dozen novices, trained them for two weeks and handed them real money. Faith was 19, the youngest in the room, and ended up running the largest account, making Dennis over $31 million in four years. The first half is memoir: February 1984 heating oil, Faith alone with a full 12-contract position while classmates hesitated, tripling their month-one returns. The second half lays out the rules, 20-day Donchian breakouts, ATR-based units, a four-unit cap, roughly 2% risk per trade, plus the original Turtle rules in full and a chapter on how backtests lie.

POINTS要点

优势是一个统计量：正期望的规则，加上足够多次的重复。An edge is a statistic: positive-expectancy rules, repeated enough times.
头寸大小由波动率决定，ATR说了算，信心说了不算。Position size comes from volatility. ATR decides, conviction does not.
1987年那天说明，单笔风险3%的系统会一夜归零。One day in October 1987 shows a 3%-per-trade system going to zero overnight.
同一套规则，赢的和输的差在执行，差不在知识。Same rules, different outcomes: the gap is execution, not knowledge.

EXCERPTS原文摘句 · 5 条

交易里的优势，是一种可以被利用的统计优势，来自大概率会在未来重复出现的市场行为。交易里最好的优势，来自心理偏差造成的市场行为。

An edge in trading is an exploitable statistical advantage based on market behavior that is likely to recur in the future. In trading, the best edges come from the market behaviors caused by cognitive biases.

Ch.5 · 印刷页 p.63–64 · PDF p.90–91

全书的定义句，把「优势」从玄学拉回统计

我是唯一一个满仓的海龟。其他每一个海龟，出于我搞不懂的理由，都决定不照Rich和Bill教的系统做。

I was the only Turtle with a full position. Every other single Turtle had decided for some unfathomable reason not to follow the system Rich and Bill had outlined.

Ch.3 · 印刷页 p.41 · PDF p.68

1984年1月取暖油，全书最有画面的一幕

收益的差别跟知识没有一点关系，全部来自情绪和心理因素。

The difference in return had nothing to do with knowledge and everything to do with emotional and psychological factors.

Ch.3 · 印刷页 p.44 · PDF p.71

同班同学收益差三倍，费思自己的结论

所以，如果你交易得很激进，每笔拿3%的资金去冒险，你会在一夜之间破产，因为这次回撤完全来自利率市场急剧反转的那一天。

This means that if you were trading aggressively and risking 3 percent of your trading capital on each trade, you would have gone bust overnight because this drawdown is due entirely to that single day when the interest-rate markets reversed precipitously.

Ch.8 · 印刷页 p.111 · PDF p.138

1987年那一天，做利率的人读到会停一下

自我想要正确，想要预测，想要知道秘密。

The ego wants to be right, it wants to predict, and it wants to know secrets.

Ch.14 · 印刷页 p.225 · PDF p.252

为什么简单规则反而难执行，一句说透

GODZILLA哥斯拉怎么读

从东京的利率台子上看

管日本国债和funding的人没谁跑过一天趋势系统，读这本书是拿它当照妖镜。第八章那句该记住：1987年10月利率市场反转那天，单笔风险3%的系统一夜归零。一本DV01过一百万美元的JGB账，2026年这个秋天美债10Y在5%附近晃，油价站上100，一天动20个bp就是两千万美元的账，规则再简单，扛不扛得住的从来是人。第三章取暖油更扎心，全班听一样的课，只有他敢满仓，做利率的人都认得这个错，方向对了，仓位只放三分之一，被抢跑，等于没做。过时的是1984年的滑点假设和E-ratio的数字，今天算法抢突破，要打折看。「有优势就重复足够多次」，二十年没变。

Nobody running JGBs and funding trades a trend system; read this as a mirror, not a manual. Chapter 8 has the line to keep: on the October 1987 day rate markets reversed, 3% risk per trade would have wiped you out overnight. A JGB book carrying over $1 million of DV01: with US 10-year yields near 5% and oil above $100 this autumn, a 20bp day is a $20 million line. Rules are simple; surviving them is the job. The heating-oil chapter cuts deeper: same class, only Faith went full size. Every rates desk knows it: right direction, one-third size, front-run. The 1984 slippage assumptions and E-ratio figures deserve a haircut; a real edge repeated enough times has not.

FOR适合谁

想第一次看清一套完整系统长什么样的人，读回忆录半段加第八章。已经跑系统的人直接翻第十一章回测骗局。想找入场秘诀的人别读，书里明说没有。

For anyone who wants to see a complete trading system end to end: the memoir half plus Chapter 8. Systematic traders can jump to Chapter 11 on backtest lies. Not for readers hunting an entry secret; there is none.