If you took the the top ten performing monkeys from 2011 and created a new subgroup with just them in 2012 they would not be able to replicate their 2011 performance. Hence, they have no edge. Replication is one of the most important variables for testing edges, not raw performance.
Okay, let's change the experiment. You put 1,000,000 monkeys in front of trading screens for 5 years. Then rank order by average annual PL. Those average PLs will also form a Gaussian. Will those monkeys at the positive end be considered skilled traders? They would have replicated their performance for 5 years. Or perhaps you would say the chances of finding monkeys with a positive PL in this experiment is pretty much zero. I suppose that is plausible...
The longer they trade the poorer the monkeys (even the top 10) will be. Say you have $1000 and you go to a roulette table (yes ... it is a very bad bet) what is the best stake you can play for? The best bet you can make is a single bet for $1000. The more handle you give the house the more juice you pay. Since we can assume that each of the monkeys is playing a negative sum game you will have more winning monkeys if they trade for a week than for a month. And more winners if they trade for a month rather than a year. At some point it becomes an out bet that any of the monkeys will be profitable. It is much the same for at least 80% of the retail traders. The vig they pay (commish + spread ... as has been pointed out) outweighs whatever edge (if any) they walk in with. Breaking the nut every month for four or five months consecutively -- even if the profits are paltry -- is a milestone. If you are well capitalized and can get that far as a retail trader you may be 50/50 to make it. And even if you are the dog you are probably no more than 7 to 5 against. Not great but a lot better odds than the guy who rents his first retail store and decides he can compete on Main Street. I think over the long haul I can outplay the monkeys. Over a 30 or even a 90 day period, however, I am 100% convinced some of them will beat me. Put the million monkeys in a position where they can trade with no vig and I have no doubt that some will beat me over even a 50 year period. Overhead in trading, as in most businesses, is like a glacier. Even though it takes its toll slowly it profoundly changes the landscape.
Interestingly enough, I was once told that this is how most brokers get started as well. Make a prediction, wait to see what happens, divide your phone list 50-50. Make another prediction, wait to see what happens, divide your phone list 25% 75%. And so on. After 5 predictions, there are some people who think you walk on water. I have also heard that there is a monkey with a dart board who does very well. Unfortunately, he is completely booked up and not taking new clients right now. I side with the old maxim - the reason they call them brokers is because the more time you spend with them, the broker you will be. One of my first lessons was brokers. I held a bond fund in the 1987 crash. I made a great amount in one day. He didn't do anything (like sell some) to capture profits, didn't take my calls, and was off servicing his key clients. His excuses were BS and so I began to learn about trading. I am still here and he and the dishonest firm he worked for are both gone. Incidentally, are there others out there who believe that there is only one edge from an abstract level? That is my current thinking having studied all types of methodologies.
No, you are missing the point of replication. You could have monkeys trading for 1000 years and you will have some that make Soros look like a real monkey. My point is, to properly test ANY data, you have to remove the existing data from it's current subset and put it into new data that has not been contaminated by the original. Once you do that, ALL your monkeys will fail!!!!!!!!!! That is my point. Replication is the key to quantifying the edge. You can't replicate randomness.
This was another of my first lessons. I carefully charted and followed the top annual predictors in a newspaper I subscribed to and after 5 or so years realized that if it is meant to be, it is up to me. I stopped looking for others to trade for me. My account then was so small, that I realized that even if they did really exist, my account would be of no interest to the really really good ones. I assure you that they do really exist.
I believe it has something to do with what engineers call "entropy" - the tendency of energy to move to increasing states of disorder as the number of conversions progress. Winning traders do the opposite of this process - money flows from the many to the few. Simply reversing the signals of a losing trading system will not take care of the profit taking exits. I believe the exits are #1 in strategy design. Consider the ball valve in a windex sprayer which only allows liquid to flow in one direction, despite a bidirectional energy input from the pump handle. If you reverse the signals from a losing system you will still need to design usable exit strategy to take the profits. This isn't really that hard, however. Real world example: Trend following system is just a range system in reverse. Flip the signals around with a logical profit target in the other direction and viola - drawups.
Obviously you have never been to vegas or bet on sports. Vig is the cost of playing. It use to refer to the % the bookie kept on your winning bets, normally 10%. The Vig is also what the house (casino) keeps on your winning bets. In trading, you pay a vig whether you win or lose. So it's the bid-offer spread plus any commissions. http://en.wikipedia.org/wiki/Vigorish Vigorish, or simply the vig, also known as juice or the take, is the amount charged by a bookmaker, or bookie, for his services. In the United States it also means the interest on a shark's loan. The term is Yiddish slang originating from the Russian word for winnings, выигрыш vyigrysh. Bookmakers use this practice to make money on their wagers regardless of the outcome. To minimize their risk, bookmakers do not want to have an interest in either side winning in a given sporting event. They are interested, instead, in getting equal betting on both outcomes of the event. In this way, the bookmaker minimizes his risk and always collects a small commission from the vigorish. The bookmaker will normally adjust the odds or the line, to attract equal action on each side of an event. The concept is also sometimes referred to as the overround, although this is technically different, being the percentage the event book is above 100% whereas the vigorish is the bookmaker's percentage profit on the total stakes made on the event. For example, 20% overround is vigorish of 16 2⁄3%. The connecting formulae are v = o⁄(1 + o) and o = v⁄(1 − v) where o is overround. It is simplest to assume that vigorish is factored in proportionally to the true odds, although this need not be the case. Under proportional vigorish, a moneyline odds bet listed at −100 vs. −100 without vigorish (fair odds) could become −110 vs. −110 with vigorish factored in. Under disproportional vigorish, it could become −120 vs. +100. Contents