Fooled by Taleb

Quote from Maverick74:

It's a "little" bit more complicated then that. LOL. This is a controlled experiment you offered. You have the coin, you know it's fair and you know exactly the outcome possibilities. In the real world, you have thousands of coins you are flipping. You don't know if they are fair and you don't even know their distributions. So when this trader has a lucky run, we won't know if it's him or the coins. The coins in this example is analogous to the example I gave earlier regarding trading environment.
More...

Indeed. There are so many participants and so many "coins", that it's incredibly difficult to ascertain whether a given participant is skillful or lucky, even in the presence of consistent outperformance.

A decent approach that has been used to evaluate fund managers (with depressingly poor results) is the comparison of investment/trading decisions with the result of random choices made under similar constraints. For those that are interested:

Discovering the quality of portfolio decisions:
http://www.portfolioprobe.com/2012/11/26/discovering-the-quality-of-portfolio-decisions/

Not Fooled by Randomness: Using Random Portfolios to Analyze Investment Funds
http://papers.ssrn.com/sol3/papers.cfm?abstract_id=2143293
 
Quote from jack hershey:

........
I never really think about who gave me the money. It just keeps being made available.
More...

This turns out to be an important question to ask sometimes IMO.
 
Thanks Maestro... on caveat... I do not think I or anyone in our office got anywhere near a buck a share.

now in general I have tought about this...

I think one of the main points being missed and why I feel so strongly about this subject is that my scenario the trader were flat everyday. If you are flat everyday and making money day after day... you are fricken good.

But this may be what tilts the tables... (the traders in my office were not averaging down. The ones who did were let go. )









Quote from MAESTRO:

Ok, here are the rough calculations. Assuming that your average take was $1 a share per trade, your Drift should be as big as = sqrroot(2 * 2000 * 1 /pi) * 700 = $24,984. The fact that you have made 14 times more than the drift tells me that your trading has an undisputable edge compared to the normal random walk. Is that the answer were you looking for?
More...
 
Quote from jem:

Thanks Maestro... on caveat... I do not think I or anyone in our office got anywhere near a buck a share.

now in general I have tought about this...

I think one of the main points being missed and why I feel so strongly about this subject is that my scenario the trader were flat everyday. If you are flat everyday and making money day after day... you are fricken good.

But this may be what tilts the tables... (the traders in my office were not averaging down. The ones who did were let go. )
More...

I can not understand why my first drafts are so poor...
this is the corrected version.


---

Thanks Maestro... on(e) caveat... I do not think (myself) or anyone in our office got anywhere near a buck a share.

I thought about this... the formulas are not looking at the following...

we were flat everyday and we did not average down.
 
Quote from StarDust9182:

This turns out to be an important question to ask sometimes IMO.
More...

It is true I do glance at who I'm opposite when the day to nite margin goes into effect.

I know I am always pushed by the herd, so genrally my extraction is made possible by the CW followers; that has always been a given.

Big money doesn't make money; they just ride through the opportunities.

So what you believe to be important is as you say just an opinion.

Survivors push me; and like all others non survivors keep getting replaced.
 
Quote from Businessman:

All types of trading except guaranteed arbitrage have an element of randomness and luck involved.

Its not possible to say with complete certainty how much luck is involved in any traders results and how much is skill.

Lets take some one who does a neiderhoffer strategy and sells OTM puts. Now lets say this trader is better than neiderhoffer and manages to get out before every major market correction, thereby not blowing up like other traders who use this strategy.

Has this trader just been lucky or is he skilled. Is he a blow up waiting to happen or will he always have the super skill and discipline to not blow up?
More...

You're conflating topics. There is a difference between "luck" and randomness. If this Taleb guy is claiming that trading is randomness, then he is wrong and or conflating topics as well.

Randomness means there is no rhyme or reason to the outcome, luck is alltogether different. Suppose you are a Casino, one of the most successful businesses out there. There is absolutely no randomness in their outcomes. They have games/odds which are only slightly above 50/50. Some are 51% to the Casino and 49% to the player or less. Yet they still come out ahead, ALWAYS, on these games. It just requires enough runs. In 1 game it's almost a coin flip whether the casino will win, in a million runs it is almost 99% probability that the casino will win.

Luck can be defined in different ways. Suppose that you have a probability of 55/45 on a trade, and you win on one run --this can be called luck. Despite the fact that you're favored to win, on one run it is basically a coin flip. The fact that you won on this run can be atrributed to luck. On the other hand if you have a million runs at 55/45, and your probability of coming on top is close to 100%, luck plays little to no role. In fact it would have to be terribly back luck for you to somehow lose with a .01 chance of losing.

Randomness only plays a role when the odds are 50/50 --a million runs at 50/50 is still 50/50 for you to win. Randomness plays a role here. You'll have some players that somehow come on top after a million runs at 50/50, and you'll have some that lose --this is completely random, and its distribution is a bell curve.
 
an important question is "what is the chance that a given trader has performed this well due to skill?"

to answer that you will want to ask "what is the chance a random coin tossing chimpanzee could have achieved those results?"

but before you can even begin evaluating individuals you need to ask, "is there skill involved in trading as a whole?"

to answer that you ask, "how predictable is any trader's future performance given his past performance?"

so you measure what is called "serial autocorrelation," or the tendency for each trader to do similarly from one time period to the next.

if a population of random coin tossing chimpanzees demonstrates the same distribution of serial autocorrelation as the population of real traders, then Taleb wins.
 
You must log in or register to reply here.
Back
Top