傳奇人物理查德·丹尼斯和威廉·?？斯亟M建了海龜，他還建立了這個(gè)系統(tǒng)，全世界商品交易市場恐怕沒有人像他那么有學(xué)問。當(dāng)?shù)つ崴挂庾R到任何和他一起工作的人都可以把交易原則濃縮并定義出來時(shí)，他發(fā)明了這個(gè)令人吃驚的成功的機(jī)械交易系統(tǒng)。現(xiàn)在，他刻意回避投資界，把自己關(guān)起來，潛心研究自己用的機(jī)械交易思想。過去25年來，他和一些程序員的合作穩(wěn)定有序。

為了解除他的神秘性，《股票和商品》的撰稿人阿特·柯林斯2004年10月在芝加哥的耀西餐廳和理查德·丹尼斯有過交談。

問：現(xiàn)在的交易環(huán)境和你過去賺錢的時(shí)代比，有什么不同？

丹尼斯：現(xiàn)在比過去難10倍。應(yīng)該如此。市場工作就是讓系統(tǒng)交易者出軌。有些人可以賺錢，但不會長久。

問：為什么？

丹尼斯：因?yàn)楹?5年或20年前比，現(xiàn)在市場變化太劇烈了。我想可能是因?yàn)楝F(xiàn)在的趨勢跟蹤者比過去多。在這個(gè)游戲中，總是有人追著你。

在機(jī)械交易的世界，期望不起什么作用，是嗎？人們說是系統(tǒng)在反應(yīng)不是預(yù)測在反應(yīng)，在一定程度上交易者根本不用做事。

關(guān)于“我在觀察數(shù)字”這個(gè)說法，有必要解釋一下。多年以前，我就知道如何脫離系統(tǒng)。你付了錢，你就要抓住自己的機(jī)會。如果你認(rèn)為市場和兩年前一樣，這些思想會幫助你的。

我還是舉個(gè)例子吧。似乎波動的行情就是好機(jī)會，波動小的最好做趨勢跟蹤。也許你想把包絡(luò)線移開遠(yuǎn)點(diǎn)，這樣你可以抓住波動的25%，它至少波動50%。這要根據(jù)數(shù)據(jù)來說，但是如果你認(rèn)為波動小是好事，那么其它的就是壞事。

未來可能有所不同。然而，不會每個(gè)人都賺錢的。

問：這些變化有什么原因嗎？

丹尼斯：對沖基金和商品交易顧問的再現(xiàn)應(yīng)該讓人們感到吃驚。過去對沖基金有很多錢，表現(xiàn)就好，這也是為什么現(xiàn)在不行的原因。錢太多不好操作。我想是因?yàn)閷_基金在拼命研究趨勢跟蹤。很多對沖基金只根據(jù)內(nèi)部消息交易。他們不是交易。他們是在桶里釣魚。

想想20或25年前我們有什么。那就是未知的信息滲透進(jìn)市場并慢慢地推動價(jià)格。

你到外匯市場，基本面是一樣的，但是也有很多內(nèi)部消息。它不是未知的內(nèi)部消息，它是官僚主義。德國銀行的官僚主義者告訴對沖基金他們要怎么做?，F(xiàn)在趨勢開始了，它突然爆發(fā)，這樣趨勢跟蹤者無法及時(shí)跟進(jìn)，然后又結(jié)束了。它不是慢，它不是間接。它是突然的。

系統(tǒng)跟蹤者在市場中看見了信號，他們以為這和大豆的信號一樣，但不一樣。趨勢的結(jié)構(gòu)在變。讓我們面對它，如果你在管理某種基金，90%是利率和外匯。10——15年的數(shù)據(jù)都沒用，因?yàn)樗煌恕?/p>

問：所以你之前的趨勢跟蹤系統(tǒng)沒用了？

丹尼斯：20年前有用的東西，現(xiàn)在十分之九都沒用了。去年和過去相比，只有一小部分的突破是有用的。我總是看N天的軌線——10天、20天、40天，一直到280天。20天的軌線有時(shí)候有用，但是從15年前開始就走下坡路了。如果你用更長的，比如一年的，它們一開始比20天的差，然后直到下一年都很好。這就是它的強(qiáng)項(xiàng)。

問：既然你不認(rèn)為任何系統(tǒng)都會永遠(yuǎn)有用，那么在你換系統(tǒng)前，你會準(zhǔn)備用幾年呢？

丹尼斯：事實(shí)上很難知道。我現(xiàn)在做的事有點(diǎn)主觀。我想知道在去年市場不好時(shí)有用的東西，5年前市場很平淡時(shí)也有用，對15或20年前也有用。這并不是說為每個(gè)階段找到最好的參數(shù)，只是找到對于過去，前不久，和很久以前很糟糕的時(shí)候都通用的東西。

問：你是不是說你現(xiàn)在不是趨勢跟蹤了？

丹尼斯：我想做點(diǎn)逆勢的——也許是現(xiàn)在有用，但10年前不好的東西——和趨勢跟蹤系統(tǒng)一起配對用。我想讓交易互相依靠，如果你用了一種，你同時(shí)也用了另外一種。這事不好辦。部分解決方案就是平衡。

問：你的方法還是長線嗎？

丹尼斯：目前，我在使用一個(gè)系統(tǒng)，他有6個(gè)不同的因素，最長的是30天。它沒起作用。我也許要使用一個(gè)大的逆勢系統(tǒng)，也許是60天，它僅僅在去年賺錢。當(dāng)趨勢跟蹤系統(tǒng)不能賺錢時(shí)，他就可以賺錢。

問：有句古老的說法，系統(tǒng)的核心最好是簡單的，你是否同意？

丹尼斯：我不同意。如果你談的是調(diào)整參數(shù)的回報(bào)，那例外情況是重要的。最好的使用系統(tǒng)的方法是放棄。如果這筆交易是趨勢跟蹤的，而市場卻是振蕩的，放棄。如果你的交易時(shí)間太長，放棄。

問：我聽說你不會編程。你能明白不同參數(shù)導(dǎo)致的結(jié)果嗎，比如標(biāo)準(zhǔn)背離，夏普比？

丹尼斯：我能明白。我做了24年了。那些懂?dāng)?shù)學(xué)的人會幫助我明白的。我不會打字，更不會編程，但我知道結(jié)果是什么。

我大部分時(shí)間就是和程序員談話，然后等結(jié)果。我和他們一起工作了20年了，所以我不急著想什么就要什么。我告訴他們我的思想，他們編程。這是一段冒險(xiǎn)，因?yàn)榧词故亲詈玫某绦騿T，一開始也會犯錯(cuò)。

問：他們的對優(yōu)化做出的決定，你相信他們嗎？

丹尼斯：我想說所有參數(shù)都是我選擇的。“我們有7個(gè)參數(shù)，這是范圍。如果某個(gè)參數(shù)有用，我們會測試10次，并盡量找到最好的參數(shù)”。這都是我的決定。他們永遠(yuǎn)不會說：“哦，讓我們再試試其它的數(shù)字?！蔽业呐d趣就是找到最好的參數(shù)。

問：我聽別人說，你說你可以把你的方法印在報(bào)紙上，還沒人能學(xué)會。

丹尼斯：趨勢跟蹤者不希望別人嘲笑他，詆毀他。像看持倉量報(bào)告這樣的事，任何人都會。市場和持倉量之間并沒有直接的關(guān)系。但是了解持倉量報(bào)告可以讓人們在關(guān)鍵的改變趨勢的時(shí)刻抓住機(jī)會。

不管現(xiàn)在原油的價(jià)格是多少——也不管5或10年后它怎么走——這都和投機(jī)無關(guān)。使用過去模式的人都會在市場中碰到人為操縱，發(fā)現(xiàn)他們的系統(tǒng)過去有用，但現(xiàn)在沒用了。

所以，短期而言，心理對人有影響。

現(xiàn)在更飄渺不定。在大行情之后有些合約要被平掉。如果你的基本面認(rèn)為會有所不同，你就占了很大的優(yōu)勢。如果原油突破了關(guān)鍵水平，每個(gè)人都要在那里賣出。如果基本面能讓下跌實(shí)現(xiàn)，那么逆勢的基金也可以。

我說的話有些人不愛聽，我稱之為“特殊情況”。我在研究時(shí)把逆勢放在心里。這里有個(gè)例子。我們過去根據(jù)N天突破交易。如果我現(xiàn)在發(fā)明了新的系統(tǒng)，我會使用一個(gè)過濾器，以判斷N天的突破是賺錢的或虧錢的。如果是虧錢的，我就會交易。比如你想在100天最高點(diǎn)買入，我就測試系統(tǒng)，看看如果在20天最高點(diǎn)買入，在20天最低點(diǎn)賣出會如何。如果20天的測試失敗了，我就會使用100的最高點(diǎn)。

問：為什么這么做？

丹尼斯：因?yàn)椋紫?，測試時(shí)失敗的交易保證了很多人在虧損時(shí)要平倉。更說明了當(dāng)你進(jìn)場時(shí)，市場是在整固，不是有趨勢的。如果上次20天的突破有利潤，那么市場就不是在整固。所以，如果市場在振蕩，20天突破虧損了，你就用100天突破，你知道市場在整固，很多人正在被洗來洗去。你知道，如果人們被洗的太多了，他們就不會跟蹤這個(gè)趨勢。這個(gè)行為會讓交易成功。

問：有趣啊。你是不是現(xiàn)在正在干？

丹尼斯：事實(shí)上我把它當(dāng)作過濾器。另外一個(gè)過濾器是振蕩——和最近1，2年比，現(xiàn)在的振蕩范圍有多大。如果范圍大，那么交易的方向是逆勢。如果你有逆勢的方法，那么就去尋找和過去相比有很大振蕩的市場。如果是趨勢跟蹤，就去尋找振蕩不厲害的。

問：你真的在做嗎？人們說你早就退休了。

丹尼斯：我可以6個(gè)月不交易，但我不會6天不做研究。有些人早上起來就在研究誰是美國總統(tǒng)。我早上起來就想搞清楚到底哪個(gè)系統(tǒng)對過去有用，對現(xiàn)在有用，對未來也有用。

問：以前你和威廉·埃克哈特爭吵才誕生了海龜——是不是受電影《交易所》啟發(fā)的？

丹尼斯：哦，上帝啊，不是！我想電影是后出來的。至少我希望如此！我做海龜?shù)氖率且驗(yàn)槊總€(gè)人都相信直覺，包括比爾，比爾很理性。我想直覺和交易不搭配。我想應(yīng)該是原則。

問：傳說中說整件事只是好玩的一時(shí)興起。你事實(shí)上是不是真的想辦海龜班？你的嘗試讓你賺錢了嗎？

丹尼斯：賺了很多的錢。他們的毛利潤是1.5億美元，我們賺了1.1億美元。我們一開始給了他們10%。為什么不給更多？這是我們的錢，我們在承受風(fēng)險(xiǎn)。

問：你們的團(tuán)隊(duì)當(dāng)時(shí)在做什么交易？

丹尼斯：我們用系統(tǒng)交易了5，6年。我知道有用的是原則。沒用的是判斷。

如果你不用原則，你就會陷入日本神秘主義，你會受靈感的啟發(fā)而行動——你本能地做事。你不能在早上醒來說：“我有一個(gè)市場直覺”。你必須做很多判斷。

問：人們說你用機(jī)械交易是因?yàn)槟阒黄谕嵰欢ū嚷实腻X。看起來非系統(tǒng)的因素也起了很大的作用。

丹尼斯：一開始，我的系統(tǒng)叫“無論如何都不能有想法”。我賺了4年的錢。每天，我都告訴自己我沒有想法。如果有人給我一個(gè)消息，叫我去買燕麥合約，我也不相信他什么都知道。我僅僅知道我有個(gè)消息，只是小優(yōu)勢，那天結(jié)束時(shí)，我的優(yōu)勢也許等于我的利潤。我就像是一個(gè)賭場。

這是中美商品交易所，很難搞，因?yàn)闆]有任何對沖。你應(yīng)該買小麥，賣活牛。這并不能減少風(fēng)險(xiǎn)：這是給了你優(yōu)勢。如果你的交易很多，你也會喪失優(yōu)勢的。系統(tǒng)也是如此。

問：你的直覺是什么？

丹尼斯：我沒那么天真——交易所的人總是在交易。但是中美商品交易所是一個(gè)創(chuàng)新，因?yàn)闆]人知道你可以依靠很大的成交量平衡風(fēng)險(xiǎn)。交易者不原因交易奇怪的東西。我就是這樣開始的。似乎那里的錢不多。第一年我賺了35000美元，在當(dāng)時(shí)不算壞了。這些年來，我有回到了那個(gè)方法，但是對我來說，最安全的還是原則。

問：你不能僅僅依靠機(jī)械系統(tǒng)賺錢，是嗎？

丹尼斯：沒錯(cuò)。每年賺30%，長期來說是很難的。事實(shí)是，一旦我去了交易所，優(yōu)勢就不存在了，所以我要形成自己的判斷。

這是我做的最好的事，但也是我的負(fù)擔(dān)，因?yàn)槲乙嘈抛约旱闹庇X。在場內(nèi)的人10次會有9次搞錯(cuò)關(guān)鍵的反轉(zhuǎn)點(diǎn)，這更確信了你要相信自己的。你在場外機(jī)械交易不會那么差，但時(shí)不時(shí)也會有類似的沖動。我認(rèn)為我在場內(nèi)的經(jīng)驗(yàn)傷害了我的機(jī)械交易。在場內(nèi)交易的人通常都是很差的系統(tǒng)交易者。他們學(xué)的東西不同。他們關(guān)注于價(jià)格的微小跳動。如果在屏幕上，這些微小跳動沒有任何含義。

問：你說海龜?shù)某晒退麄儌€(gè)人有關(guān)——因?yàn)樗麄兪悄銍?yán)格挑選出來的。理論上，是不是任何人都能掌握機(jī)械交易方法？

丹尼斯：可以這么說，這和我們選什么樣的人區(qū)別不大。人們在學(xué)習(xí)海龜課程以后是否能做到，這和他們的個(gè)性有關(guān)系。如果有我們來控制，不管他們多么聰明，結(jié)果差別不大。

問：真的沒有差別？會不會有些人害怕使用系統(tǒng)原則？

丹尼斯：越是有創(chuàng)造性思維的人，越容易毀掉自己的表現(xiàn)。優(yōu)秀的交易是無聊的交易。成功的交易要求有能力實(shí)施限制。成功的系統(tǒng)交易者花了很多精力開發(fā)系統(tǒng)，但是在遵守系統(tǒng)方面就很差。你一定是精神分裂了。拼命開發(fā)了系統(tǒng)，然后又不用它。如果布什總統(tǒng)有一個(gè)系統(tǒng)，他會成為一個(gè)優(yōu)秀的交易者。

最后，你要去享受你的系統(tǒng)，否則沒有任何含義。

所以，你要有紀(jì)律，不能高估自己的判斷。一旦你把系統(tǒng)開發(fā)好了，你就不要再想辦法去優(yōu)化它。你可以在開發(fā)的時(shí)候優(yōu)化，但是在交易的時(shí)候優(yōu)化它會毀了很多東西。

問：一旦你確信不用海龜了，會不會有些人會抱怨這個(gè)系統(tǒng)？

丹尼斯：有1，2個(gè)人不出名，大部分都很出名。事實(shí)是，如果我告訴他們，他們會比我還難過。

問：你知道嗎？你的系統(tǒng)在互聯(lián)網(wǎng)是免費(fèi)的？

丹尼斯：我知道，但那不是我的系統(tǒng)。事實(shí)上，讓我困擾的是，人們以為我把自己不用的系統(tǒng)公布出去了。一次我在密歇根大街上走，聽到別人這么議論。很明顯，他們以為是我的。我希望人們明白，我很早就不用這個(gè)系統(tǒng)了。

我改變了很多東西。很多人就不知道去優(yōu)化點(diǎn)什么。比爾·埃克哈特還在用，但他瘋狂地優(yōu)化了很多東西。我則相反，我改變了觀念，如果你知道我3年前在做什么，我可能不知道我現(xiàn)在在做什么。我必須承認(rèn)這是一個(gè)主觀愿望。我想努力找到一個(gè)系統(tǒng)，它能和市場協(xié)調(diào)起來。我的思想不太現(xiàn)實(shí)。

問：如此說來，你不是靠管理基金發(fā)達(dá)的，而是靠個(gè)人交易發(fā)達(dá)的，是嗎？

丹尼斯：客戶沒有和我交流思想。問題是基金經(jīng)理很少會和有錢的人坐在一起?？偸怯羞@個(gè)代理，那個(gè)代理。人們有個(gè)誤區(qū)，都想找到通用的方法，但沒人想找到特別的方法。很難向最終客戶解釋。唯一的辦法是把客戶留在這個(gè)游戲中，這意味著不賺不虧也是不錯(cuò)的。

我的一些交易很糟糕。我不想解釋，但是人們有興趣知道，所以我一般不想管理基金。

問：為什么做系統(tǒng)工作？他們需要研究什么系統(tǒng)特點(diǎn)？

丹尼斯：系統(tǒng)對過去有用是因?yàn)橐恍╆P(guān)聯(lián)。也就是價(jià)格會想相同的方法移動。如果價(jià)格上漲了一天，你可能會預(yù)測它第二天還會漲。如果它漲了一周，你可能預(yù)測它下周還會漲。這就是趨勢的潛在原則。一個(gè)聰明的趨勢跟蹤系統(tǒng)就要盡力組合這些片段，這樣信號之間能互相確認(rèn)，這樣你就預(yù)測到了更多的力量。它有用是因?yàn)樗m用于不同的時(shí)間框架。

基本派滲透到價(jià)格中。政府行為或官方透露的信息都能改變趨勢。這樣模式就變得模糊了。如果你事后看，趨勢很純粹。這就是少數(shù)交易者知道的信息改變了價(jià)格的波動。這就是價(jià)格波動的不同。它扭曲了自然的趨勢。

問：你不是基本派，是不是？

丹尼斯：今年可能是基本派的天下。每次我打開電視，只要原油漲了5美元，我就能看見T·布恩·皮肯斯齜牙咧嘴地笑。當(dāng)原油漲到30美元時(shí)，他就在電視上說：“可能是40多美元，不會是20美元?？赡苁?0多美元，不會是40美元。”如此不斷循環(huán)，當(dāng)然了，人們一直在聽他的，他是對的。你也許希望這是技術(shù)派的倒霉日子。不管原油是10周的最高點(diǎn)，還是最低點(diǎn)，基本派都是看漲的。趨勢跟蹤者想在10周最低點(diǎn)買入，如果基本派多了，趨勢跟蹤者就倒霉了。

我更應(yīng)該成為基本派的傳教士，而不是基本派。對很多人來說，根本得不到基本面的信息。有時(shí)候，市場確實(shí)如預(yù)測的那樣，但長期是站不住腳的——看了電視以后才知道會發(fā)生什么。如果宇宙是這樣的，我就麻煩了。

問：假如說有人想以機(jī)械交易為生。比如他一開始有50000美元，想每年賺50000美元。

丹尼斯：我希望他能把50000全部投入，因?yàn)檫@是唯一的辦法。大部分系統(tǒng)會接受20%的虧損，希望你不會那么倒霉。50000美元虧20%是一回事，但是你要忍受這個(gè)可能，20%的縮水，一直到0。一般系統(tǒng)的縮水是最大縮水的1——1.5倍。如果你投入50000美元，你可能會靠一般的系統(tǒng)賺50000——75000美元。但我想這是上限。

請注意：一個(gè)系統(tǒng)在不同市場的表現(xiàn)應(yīng)該是一致的。針對不同市場定制不同系統(tǒng)也是可能的，甚至是必要的。

除了標(biāo)準(zhǔn)普爾，它們都是一樣。你必須當(dāng)它們是一樣的，不要找理由改變參數(shù)。你可以針對不同的市場優(yōu)化不同的系統(tǒng)，或者一起優(yōu)化。最好一起優(yōu)化。

問：為何標(biāo)準(zhǔn)普爾不同？

丹尼斯：它們是股票的平均值，如果你想研究它們，很難找到樣本。有人用棉花的系統(tǒng)去做標(biāo)準(zhǔn)普爾，一做就失敗了。3到5年前，有很多跟蹤股指期貨的系統(tǒng)。市場在研究如何從這些期望中套利。

問：你如何做的？

丹尼斯：這是你的思想。人們拿著系統(tǒng)找我。系統(tǒng)都很簡單，我不知道他們?nèi)绾蜗褡瞿M交易時(shí)那樣做交易。當(dāng)人們開始實(shí)戰(zhàn)時(shí)，他們就泄氣了。

這不奇怪。我也用短期系統(tǒng)，但那是我用心選擇的，且不會經(jīng)常交易。和其它思想相比，我認(rèn)為它們不行。我認(rèn)為那樣交易都是給別人貢獻(xiàn)傭金的。

另外一個(gè)問題是這些短期的東西好找。參數(shù)少。沒有大量的數(shù)據(jù)。沒有大量細(xì)致的工作。

你需要不同的方法。

你應(yīng)該嘗試不同的東西。我是一個(gè)真正的逆向投資者，比如人們說最簡單的就是避免持倉過夜。實(shí)際上，我想去做更難的事才合理——持倉過夜。3年前，似乎可靠的統(tǒng)計(jì)都在說要從開盤到收盤盯盤。如果你了解收盤——比如說當(dāng)天漲了——下次開盤時(shí)你有55%的機(jī)會是對的。

這個(gè)方法對我的早期生涯有很大幫助，尤其是在場內(nèi)交易時(shí)。有時(shí)我很沖動，但我愿意在收盤前建倉。

在過去沒有半夜的電子市場時(shí)，谷物在1：15分收盤，在第二天9：30分開盤。你沒有求助對象。

人們開始打賭。我知道一個(gè)人，過去每晚交易上百分債券，一年賺了100萬。但是我做的時(shí)候就發(fā)現(xiàn)很難做。我知道幻覺就是風(fēng)險(xiǎn)。沒錯(cuò)，半夜交易是有風(fēng)險(xiǎn)，但也能賺錢。

我強(qiáng)烈支持“做難事?！焙茈y把它塞進(jìn)機(jī)械系統(tǒng)，因?yàn)槟阌肋h(yuǎn)不用做判斷了?，F(xiàn)在的人都會做趨勢了，他們也不害怕?，F(xiàn)在很難知道什么是很難的事。如果你想想——你能想到一些。

———————————————————————————————————

Perhaps no one in the world of commodity trading has more lore attached to him than the legendary Richard Dennis, the founder, along with William Eckhardt, of the original Turtles and the system they established. Dennis’s stunningly successful involvement with mechanical systems came about when he realized that whatever worked for him in his trading could be reduced to and defined as a trading rule. Today, he pretty much eschews the public investment arena, confining himself to an exploration of mechanical trading ideas for personal use. A self-described “computer-illiterate,” he nonetheless has worked steadily with a handful of programmers for the last quarter century.

　　To separate the myth from the man, STOCKS & COMMODITIES contributor Art Collins spoke with Richard Dennis in October 2004 at Yoshi’s, a restaurant in Chicago.

　　How’s the current trading environment compared with your high-profile days?

　　It’s 10 times harder than it used to be. It should be. The market’s job is to derail the systems traders. Some of them are going to make money, but that can’t go on forever.

　　Why not?

　　Because the market is changing more dramatically than it would have 15 or 20 years ago. I think that’s because there are a lot more trend-followers involved in the market than before. It’s a game where you’re forever chasing your tail.

　　Anticipation doesn’t have a lot of place in the mechanical trading world, does it? The adage is that systems react rather than predict, and traders should have no input at all once they’re beyond a certain stage.

　　There’s something to be said for the dumb bunny approach of “I’m just looking at numbers.” I’ve done my share of talking myself out of systems that would have worked for years. You pays your money and you takes your chances. But ideas help if you’re thinking about what the market is likely to be like two years from now.

　　Let me give you an example. It seems to be increasingly true that volatility screens are a good idea; the trades with the lowest volatility for trend-following are best. You might decide to push that envelope even further by taking only a quarter of the trades that are the least volatile instead of half. That part wouldn’t be based on data, but on your idea that low volatility is good and the rest will be worse.

　　There’s got to be something different about the future. Otherwise, everyone will make money, and we know that can’t happen.

　　Any other reasons behind the change?

　　People should marvel at the difference between the returns of hedge funds and Commodity Trading Advisors (CTAs). The hedge funds have been much better over the years, with more money, which is a reason they should be worse. It’s harder to perform with huge amounts under management. I think it’s because hedge funds exploit trend-following systems to a large extent. A lot of the hedge funds are just trading on inside information. They’re not trading. They’re shooting fish in a barrel.

　　Think about what we had 20 or 25 years ago. The paradigm was that unknown information seeped into the market and gradually moved price.

　　You go to a currency market, and there’s the same fundamental forces but you also have this tremendous flow of inside information. It isn’t unknown inside information, but it’s bureaucratic. It’s bureaucrats at the Bundesbank telling the hedge fund guys what they’re going to do. Now as a trend starts, it explodes immediately so that the trend-followers can’t get in until it’s over. It isn’t slow. It isn’t indirect. It’s all at once.

　　System followers see a signal in financials that they think is the same as a soybean signal, but it’s not. The structure of the trend changes. Let’s face it, if you’re running any kind of a fund, 90% of your trades are in interest rates and currencies. It throws all the data of more than 10 to 15 years ago into question because it was generated a different way.

　　So your old bellwether trend-following systems don’t work anymore?

　　Nine out of 10 things that worked 20 years ago don’t work now. In the last year, only a subset of those breakouts worked relative to the 20 to 40 ranges that worked previously. I’m always looking at n-day trajectories – 10, 20, 40, up to maybe 280 days. Twenty-day trajectories worked for a while, but that started going downhill 15 years ago. If you go out to the longer-term ones, like a few years’ worth, they started out worse than the 20-days and maintained themselves nicely until last year. That’s what makes it tough.

　　Since you don’t regard any systems as likely eternal performers, do you anticipate only x number of probable profitable years before you have to switch?

　　That would be hard to know before the fact. What I try to do now may be a little subjective. I like to find something that worked in the last year when markets were bad, but also worked in the previous five when markets were mediocre, and also worked 15 or 20 years ago. That doesn’t mean you’re trading the optimal variable for any of those periods, but rather, something that has been passable in the good old days, the recent past, and in the very recent bad period.

　　Would you say a big part of what you’re doing now is not trend-following?

　　I like to take something counter-trending – maybe something that worked well this year but not so well 10 years ago – and pair it with a trend-following system. I like to make the trades depend on each other so that if you get one of one kind, you also get one of the other kind. It’s not the easiest thing to do. Part of the solution is to have a balance.

　　Are you still long term in your approach?

　　Currently, I’m running an online system that incorporates six different elements, the longest about 30 days. It isn’t behaving. I might have one mega-counter-trending system that might be for 60 days, which only makes money in periods like the last year. It’s balanced to make money when the trend-following stuff doesn’t work.

　　Do you agree with the old saying that simple is best at a system’s core?

　　I don’t. If you’re talking about risk-adjusted return, exclusion’s important. The way to make a system optimal is to throw things out. If it’s a trend-following trade and the market’s too volatile, throw it out. If you’ve had the trade too long, throw it out.

　　I’ve heard you don’t do your own programming. Do you understand the full significance of the results, and various components like standard deviation. and Sharpe ratio.?

　　Yeah. I’ve been doing it for 24 years. The math guys set it up and helped me understand it. I can’t type, much less program, but I understand the output.

　　I spend most of my time talking to programmers and waiting for results. I’ve been working with these people for 20 years, so I can get into that intermediate zone between computerese and real English. I tell them my idea of how it would go in the program, but they’re the guys who are going to program it. It’s quite an adventure, because even a good programmer will get the first it-eration wrong half the time.

　　Can you trust them to not make poor optimization decisions?

　　I pick all the parameters. I’ll say, “We have seven [variables], here’s the range. If anything shows any promise, go through it 10 times and try to narrow down the optimization.” It’s all my decision. They never just say, “Oh, let’s try some numbers.” My real interest is trying to figure the damned thing out.

　　You’ve been quoted as saying you could print your methodologies in the newspaper and no one would be able to follow them.

　　An individual trend-follower isn’t going to have people trying to mimic or corrupt him. But take the Commitment of Traders’ report, in which everyone can see how many contracts are committed, for example. It’s not just a straightforward accounting of the relationship of the number of contracts and how the market is behaving perversely. But knowledge of the report does give people an opportunity to change the trend at crucial moments.

　　Whatever crude oil’s price is now – and whatever it’s going to be one, five, or 10 years from now – won’t have much to do with speculation. Those using patterns of the past will encounter enough manipulation in the market to the point where their systems that worked before won’t work now.

　　So psychology affects the short run.

　　It’s less ethereal than that. There are contracts that have to be liquidated after a big trend. If you have some idea of what fundamentals will make a difference, you’ve got a big edge. If crude breaks the important levels, everyone’s got to sell it from there. If something fundamental can make that decline happen, countertrend funds will make it happen.

　　I’m an advocate of something most people don’t like, which I call “special situations.” I research keeping the countertrend element in mind. Here’s an example. We used to trade n-day breakouts. If I were creating the system now, one of the filters I would use is whether the n-day breakouts were winning or losing. I would only take the trades if they were losing. Suppose you want to buy at the 100-day high. I’d run a system in the background that showed me how a system is doing that buys at the 20-day high and sells at the 20-day low. I would only take those trades at the 100-day high if the 20-day trade in the background was losing.

　　Why does that work?

　　Because, first of all, the losing trade in the background guarantees a certain number of people with losses who have to liquidate. It also defines the market to be more in a consolidation than in a trend when you enter. If the last 20-day breakout has a profit, the market is not in a consolidation. So if the market is choppy and the 20-day breakout is losing, when you get to the 100-day breakout you know you were in a consolidation and people have been whip-sawed. You know that people are less likely to go with this trend than they would have many whipsaws ago. That behavior tends to make a trade work.

　　Interesting. Is that a lot of what you’re doing now?

　　I actually use it as a filter. Another filter I use is raw volatility – how big the ranges are relative to how big they’ve been in the last year or two. If the ranges are big, trades ought to be oriented to the countertrend. If you have something that works as a countertrend, go more for the markets where the ranges are big relative to what they were. If it’s trend-following, go for the small ranges.

　　Have you been doing this straight through? There were stories that you retired for a time.

　　I went six months without making a trade, but I never went six days without doing research. Some people like to wake up in the morning and figure out how they can get to be President of the United States. I wake up wanting to figure out a system that really worked well in the past, really works well now, and will continue to work in the future.

　　Regarding that initial argument you had with William Eckhardt that led to the formation of the Turtles – was it inspired by the movie Trading Places?

　　Oh God, no! I think the movie came after. At least I hope so! We did the turtle thing because everyone believed in intuition, including Bill, who is a very logical guy. I thought about intuition and trading and it didn’t seem right. It seemed like it should be rules.

　　Legend depicts the whole thing as kind of a playful whim. Did you in fact have a percentage of the Turtles? Did you make money off the endeavor?

　　Tons. I think they grossed $150 million and we made $110 million. We started out paying them 10%. Why give them a lot? It was our money, we took all the risk.

　　What kind of trading were you person-ally doing?

　　We’d traded systems for five or six years. I realized that the things I’d been doing that worked best were rules. The majority of the other things that didn’t work were judgment calls.

　　At the risk of drifting off into Japanese mysticism, when you’re not using rules, action still has to come out of some inspiration – something you’re doing with the spontaneous part of your mind. You can’t wake up in the morning and say, “I want to have an intuition about the market.” You’re going to have way too many judgments.

　　The adage is you need money to trade mechanically because you can only expect to earn a certain percentage of your capital. It would seem that the nonsystematic elements of what you were doing played a key part in your success.

　　When I started out, I had a system called “Having no idea whatsoever.” For four years, I was just taking edges. Every day, I focused on the fact that I knew nothing. If someone gave me a quarter-cent edge to buy an oat contract, I didn’t think he knew anything either. I just knew I was getting a quarter-cent edge, and at the end of the day, the edges would approximately equal my profit. I tried to be like the house in the casino.

　　This was the MidAm [Mid-America Commodity Exchange] and it was tough sledding because nothing offset anything. You’d buy wheat and sell cattle. This didn’t reduce risk: It just gave you the edge on each. When you make a lot of trades, you can drown the risk of those edges and come out with the residue of whatever the edges were to start with. [Initially] that was pretty much the system.

　　Which you pretty much intuited, right?

　　It wasn’t that novel – people at the Board of Trade had been doing it forever. But for the MidAm, it was revolutionary because no one would understand that you could balance your risk with a lot of volume. Other MidAm traders weren’t willing to bother with odd-lots or things like that. That’s how I started. It wasn’t like there was a lot of money in it. My first year I made something like $35,000, which wasn’t so bad back then. Over the years, I’ve drifted back to that approach, but the safe harbor for me has been coming back to the rules.

　　But you couldn’t have amassed the money you did just being mechanical, could you?

　　Sure. Even making 30% a year over a long-enough period. The truth is, once I went to the Board of Trade, the edges weren’t there in the same respect, so I had to develop some judgments and subjectivities.

　　That might have been the best thing that I’ve done, but in terms of going forward, it was an albatross because I’d learned to trust my intuitions. On the floor, you see people and know that they’re wrong nine times out of 10 at crucial turning points, which then confirms with near certainty any ideas about what you have to do. Nothing like that exists when you’re trading mechanically off the floor, but you’ll still feel similar impulses every now and then. I think my pit experience hurt my mechanical trading. People trading in the pit are generally very bad systems traders. They learn different things. They react to the tick in front of them. That’s nothing like looking at numbers on a screen.

　　You’ve suggested that the success of the Turtles was somewhat related to their personal makeup – that you selected them through an explicit screening process. In theory, couldn’t anybody follow a mechanical approach?

　　You could make the case it didn’t make much difference who we picked. The people who could sustain trading after the Turtle program did so according to their abilities. While they were under our control, it didn’t make much difference how intrinsically smart they were.

　　But shouldn’t it make no difference at all? Or is there something beyond people being too scared to follow system rules?

　　The more somebody is creative, the more chance they have of sabotaging their performance. Good trading is boring trading. Successful trading requires a kind of intelligence that can realize and enforce its own limitations. Successful system traders apply every ounce of intelligence they have into the creation of their systems, but they’re dumb-bells about following them. You’ve got to have a schizoid approach. Work like hell to make it good, and then ignore it. President Bush would be a great trader if he had a system.

　　Ultimately, you have to be on board to enjoy the bounty of your system. Otherwise, none of it means anything.

　　That’s why you need to have the discipline not to overrate your own judgment. You have to think of your systems as being beyond improvement once you’ve done the best you can with them. You improve them every minute you’re researching them, but the idea of improvising during trading has been the ruin of many.

　　After your confidentiality period with the Turtles expired, didn’t some of them start breaking the news on the system?

　　There were one or two who will remain nameless. The majority was exemplary. The truth is, if you talked to those people, you would find they’re way more upset with certain people than I am.

　　You know your system is available for free on the Internet now, right?

　　Yes, but it’s not my system now. Actually, one thing that’s a little bothersome is that a lot of people probably think I had something to do with posting my discarded system. Once I was walking down Michigan Avenue and I heard somebody talking about it. It was clear that they thought it somehow had my blessing. I hope people realize that it hasn’t been a system that I would have advocated trading for a long time now.

　　I’ve changed things a lot. A lot of people don’t even like to optimize anything. Bill Eckhardt stays in a very tight range, although he does re-optimize like a maniac. I, on the other hand, change the concept to the point that if you knew what I was doing three years ago, you probably wouldn’t have any clue what I’m doing now. I have to admit it’s subjective. It has to do with trying to anticipate what systems the markets are going to accommodate. I try to come up with novel ideas, ideas that others haven’t run to ground.

　　From what I’ve heard, you didn’t thrive in fund management anywhere near as well as in individual trading. Is that right?

　　Customers and I didn’t have a meeting of the minds. The problem was that a money manager rarely sits down with the person with the money. There’s always a representative of a firm of a firm of a firm. There are these levels of misdirection where everyone wants something that’s passable, but no one wants something that could be spectacular. It’s difficult to explain to the end user. The only objective is to keep the customer in the game, which means that breaking even is seen as doing okay.

　　I’ve had some bad performances that had nothing to do with that. I don’t want to make that be an explanation, but this variance of interests has been a problem, so I tend to not want to do it.

　　Why do systems work? What market characteristics do they exploit?

　　Historically, systems have worked because of serial correlation. There was an implication that the price would continue to move in the same direction. If a price moved up for a day, you could predict it would move up for another day. If it moved up a week, you could predict another week. That’s an underlying principle of why there are trends. A clever trend-following system just tries to overlap these momentum segments so each of these signals reinforce the others, so that you get a prediction with more force. It works better because it incorporates more than one time frame.

　　Fundamentals seep into price series. There’s a corrupting element to trends when markets get overwhelmed by government action, or officials leaking information. That’s when the patterns get muddied. If you look backward in time, the patterns tend to suggest more of a pure trend-following. This other kind of price movement is based more on short-term information known to a few trader groups. It makes a big difference in the price series. It distorts the natural formation of trends.

　　You haven’t generally been big on fundamentals, though, have you?

　　This is probably the year of the fundamentals. I turn on the TV and see T. Boone Pickens grinning every time crude’s up five bucks. He’s been on TV since crude was at the $30 level, saying: “$40’s more likely than $20. $60’s more likely than $40.” That ebbs and flows, of course, and people get their head handed to them along the way, but overall, he’s been right. You would expect it to be a bad period for technicians. Fundamentalists would say crude is bullish whether it’s at a 10-week high or low. They’re more likely than the trend guys to want to buy at a 10-week low, which means if they’re right, the trend guys are out of luck.

　　I’d have a better chance of being a fundamentalist preacher than a fundamentalist trader. Fundamental information just isn’t accessible to a lot of people. Sometimes, the market actually does what’s predicted, but still, it’s an untenable long-term paradigm – to watch TV and then know what’s going to happen. If the universe is structured like that, I’m in trouble.

　　Suppose someone wants to trade mechanically for a living. Say he has $50,000 for startup and will need to earn an average of $50,000 per year to survive。

　　I hope he’s willing to risk the whole $50,000, because that’s about the only way you can make that. Most decent systems build in about a 20% loss, which hopefully you don’t hit right off the bat. A 20% drawdown off $50,000 is one thing, but you also have to withstand the possibility of a 20% drawdown from zero. A decent system only makes one to one and a half times a maximum drawdown. If you have $50,000 to lose, you can probably make $50,000 to $75,000 with a decent system. But I would think that’s about the upper limit.

　　Please weigh in on the following: a single system should perform the same across an array of markets. Or a certain amount of custom tailoring per system per market is possible and maybe even inevitable.

　　With the exception of the S&Ps, they’re all the same. You have to treat them the same, for no reason other than sample size. You can optimize each system in each market individually or you can optimize them all together. It’s better to optimize them together.

　　What makes the S&Ps different?

　　They’re averages of stocks and if you’re going to do any research on them other than day systems, it’s hard to get enough of a sample. There’s a cottage industry of short-term S&P systems that just died the minute they were implemented. There were an immense amount of what I assume to be trend-following index systems available three to five years ago. The market figured out how to arbitrage out that expectation.

　　How do you stay above all that?

　　That’s where you need ideology. People would come to me with systems. They were pretty simple and it was hard for me to see how they could keep working the way they had on paper. The minute people actually started trading them, they just fell flat.

　　It’s not surprising. I use short-term systems too, but they’re very selective and don’t trade very often. Compared to other methodologies, I’ve got to say they don’t work as well. My impression is that that kind of trading is driven by margin considerations.

　　Another problem is that these short-term things are easier to find. There are fewer parameters. There’s a ton of data. It’s not a place where you find much subtlety.

　　You want the difficult approach.

　　You should be trying to do something different. I really am a contrarian at heart; for example, the easiest task is avoiding overnight risk. In fact, I thought it would make more sense to do what’s harder – take the overnight risk. Up until three years ago, it seemed to me that the most reliable statistics were follow through from the close to the open. If you knew something about the close – it was higher for the day, for example – you had about a 55% chance of being in the right position for the next open.

　　That approach helped me a lot in my early days, especially when I was in the pit. I would be brutalized sometimes, but I would make myself go back and take a position at the end of the day.

　　Back in the days when there were no overnight electronic markets, what you got was the 1:15 close and the 9:30 open price the next day in the grain pits. You had no recourse.

　　People started making those bets. I know a guy who used to trade a hundred bonds every night and make $1 million a year, so the market adjusted. When I was doing it, it was the hard thing to do. I thought the risk of it was illusory because I knew what risk you had during the day. Sure, the overnight trade seemed to have more intrinsic risk, but it also had more reliability.

　　I’m a big proponent of “Do the hard thing.” It’s hard to squeeze that into a mechanical paradigm, because you never make judgments. People are so trend oriented now that they don’t mind biting off anything. It’s harder now to know what the harder thing to do is. But if you can figure it out – you probably have something.