虽然资金管理说得多，都有点逆反心理了，不过下面的东西可能对我们还是有点用。
这是hylt兄在Van Tharp的论坛上问的一个問題得到原版海龟交易规则的作者的回答（Tharp的回答我觉得不好，不拿出来了）：
Author: Curtis Faith (216.146.89.---)
Date: 01-29-04 06:11
The other thing about the Kelly formula is that it assumes you can keep on playing the game. In effect, that you have an infinite bankroll.
If you look at the original kelly paper, you'll find that the derivation comes from the mathematics of infinite series. A large string of losses and then an eventual turn around results in a positive outcome over time if you can continue to play.
Kelly was originally concerned with getting maximum throughput using digital transmission of voice signals. Losses don't matter so much there because the data just gets retransmitted.
With trading, there's real money involved, and I don't know too many of us with an infinite bankroll. If you go bust, knowing that you would have eventually made money if you had only had a little more won't be of much consolation.
My advice is NEVER, EVER, EVER use the Kelly formula. It is very dangerious to trade anywhere near the size it indicates. Ralph Vince, Ed Seykota, and the other proponents have done new traders a major disservice by promoting the idea.
http://mastermindforum.com/phorum/read.php?f=9&i=6660&t=6645#reply_6660
Curtis Faith:
另一方面，凯利公式假设你可以（译注：是说“可以”而不是“必须”）将游戏永遠玩下去，实际上，这是暗示你有无限的筹码。
如果你看关于凯利公式的原创论文，你会发现它来自无穷级数的数学推理。因此，如果你可以不停的玩下去的话，面临一大串连续亏损时你总可以等到最终来个大翻盘。
凯利（原创人）最初关心的是声音迅号传输时如何才能得到最大的通过量。迅号损失（译注：对应交易中的亏损）无关紧要，因为迅号可以重新传输。【交易知识macd.org.cn收集整理】
而对于交易，那关系到真金白银，我不知道我们有谁有无穷多的资本。如果你破产了，知道自己如果能够再多那么一点点钱就可以翻本，并不会起到什么安慰的作用。
我的意见是，永遠，永遠，永遠不要使用凯利公式。无论在何时何处，按照它指示的头寸进行交易都是非常危险的。Ralph Vince，Ed Seykota，和其它一些人对凯利公式的推崇，会对新入行的交易员造成极大的伤害。
Van K. Tharp的回答，但我不懂他为什么这样讲。
Author: Van Tharp (---.nc.rr.com)
Date: 01-28-04 15:22
Kelly only applies to a binomial distribution such as heads (50% pays 2:1) and tails (50% loses 1:1). It does not apply at all to an R-multiple distribution like this.
And most times, you''''ll lose your shirt betting Kelly.
Van
Van Tharp:
凯利公式只能用于二项式分布，比如（抛硬币）的正面（50%的可能，收益是2：1）和反面（50%的可能，亏损是1：1）。它完全不能用于R-multiple分布（译注：Tharp的R-multiple分布，他的书上有讲）。
绝大多数时候，用凯利公式会让你把内裤都输掉。
