[1] http://www8.gsb.columbia.edu/rtfiles/cbs/hermes/Buffett1984....
Imagine Deep Blue beats Kasparov 40 games in a row. "Oh, this computer is getting lucky, it doesn't really have a strategy. This is the unicorn computer that is beating our best player."
If the manager picked a single stock 40 years ago and that outperformed the market, that would be a unicorn. If the manager actively managed the fund for 40 years, buying and selling several times a year based on market conditions, and beat the market (the "best we have"), I'd say the strategy was sound.
You can always cop out and claim talent is luck (Steve Jobs was "lucky" at picking products, etc.).
So, yes it's possible that this guy is doing something clever or has some advantage that means he will continue to beat the market. It's also possible it's just that he got lucky, and the other 9,999 people who pursued strategies and didn't get lucky are forgotten about.
If (somehow) beating the market was a 50-50 chance (which is like saying a randomly written chess algorithm has a 50-50 chance of beating Kasparov), the chance of 40 years of gains is 1/2^40, which is less than 1 in a trillion.
I'd say that's sufficient evidence to reject the null hypothesis, that the drug is no better than placebo. Just because we don't understand the mechanism doesn't make it chance.
So, he very well could be the lucky monkey.
Beating the market (no matter the margin) is purported to be a 50-50 chance, and doing so 40 times in a row would happen by chance less than 1 in a trillion times.
(That is, if you accept the premise that stock market investing has no skill component.)
Roughly speaking, if it's possible to beat the market, then when comparing between the result distribution we have of actual result, against a hypothetical distribution where investor was using coin toss to invest, those two have to be different. The conclusion comes from people saying that those two looks the same.
Now, actually make the test and choosing all the variable in question is NOT a simple task, and definitely is a hard problem.
I just want to point out that it's possible to pick out winners better than random quess if they have advantage over others.
Case in point: Warren Buffet. He is good investor who has beaten the S&P 500 decade after decade. But he is not beating the market because he makes series of good decisions. He beats the market because he has structured his business so that he has several advantages that others don't. The heart of his empire is his insurance companies. They create steady stream of cash even in the bad times when company valuations are low. Buffet don't have to loan money when he finds good investment or sell his other assets when he finds good investment. This already gives him severals percentage points advantage over others and that advantage compounds over decades. Also, he is in the position where he don't have to play martingale strategy [1] to please investors, because he and his friends own so many shares in the company (he is not in the payroll, he is in the charge).
[1] The real cause of the financial crisis - An MIT Blackjack Team perspective by Semyon Dukach http://semyondukach.blogspot.com/2009/01/real-cause-of-finan...
Note, being a unicorn isn't necessarily a sinker for an argument, it just can't be your only defence (i.e. gravity is the unicorn of theories about why things fall :) )
The classification of anyone that beats the market as a lucky unicorn is logically impermeable. If you accept the argument, then no matter who says they're beating the market, you'll be able to dismiss them without further thought.
There needs to be a better argument, one that allows for the hypothetical fund that does manage to do something intelligent and beat the market.
I find that this style of logically impermeable arguments crop up quite often when dealing with peoples belief structures. Here yours is that the stock market cannot be beaten, which may or may not be true. A useful question to ask in these situations is; Suppose that your belief is wrong, what evidence do you imagine might prove that it is in fact wrong?
When you can answer that question, you'll have a more concrete argument for why this fund is indeed just a lucky survivor.