An Edge is any differentiator that results in above average alpha. The timeframe delineates between "lucky guess" and "edge".
In a back-test setting, edge is how far away a strategy's P&L has been from randomness (after all transaction costs are taken into account), i.e. statistical significance. The p-value is conceptually little different from a sharpe ratio. So it is not only a positive expected value but also taking variability into account, i.e. whether this result could have been obtained by pure randomness. That's a much stricter requirement.
The edge lies entirely within the trader - not from the market or past data. I was using back-testing since the mid 90's, I am not biased just experienced. Experience is everything. All the back-testing, high probability, & positive expectancy under the sun means nothing until you have paid your dues to have forged the right psychology & have become strong with risk/trade mgmt. Entry signals alone will not win the game. Nearly all traders will have to endure a great deal of painful loses or blow a time or two so that their ego is shattered severely enough so they let go completely of their stubborn arrogance & are able to start truly learning.
Very true. Experience is the greatest teacher. Moreover, I have to question whether mechanical/algo trading is really better than discretionary trading. While mechanical trading does eliminate emotion (until losses mount), there is a big drawback: intuition. Intuition is the sixth sense of trading that comes only from long years of arduous struggle, namely, experience.
Best answer. A race car driver has an edge when he/she has tested a technique that puts them in the lead. A trader has an edge when he/she has tested a setup that results in a non-random outcome. Here's the question: is the setup profitable or unprofitable? Could go either way and still be non-random. The setup can include as many variables as you like as long as they're used consistently within the strategy. The problem with variables is that they often turn into "wheels within wheels". The variable itself can also have variables. The operative word is "back-test". And this turns out to be just a problem in statistics. If you apply probability theory as well, you may be ahead of the game, at least to the extent that you may have a better handle on the probable outcome.
“1) Edge is a pattern. It could simply be a price pattern on the chart, or it could be a behavior pattern (eg. the interaction between me and/or the market).” Edge is trading method: 1. statistically profitable, providing positive expectancy – this is objective part 2. usable for person, trading this method– this is subjective part Details: 1. I believe that (1) can be achieved using objective price action pattern in defined context. Pattern and context can be pure chart based, but also can be linked to market event and/or defined time intervals. I do not believe in universal pattern, providing positive expectancy in any context. I cannot postulate that it does not exist, but I never seen such pattern and this is so called holy grail for me. 2. The subjective part is about trader’s personality. The major factors are: working time, required capital and trading tools, risk appetite. I have trading system in my arsenal for FGBL future. Win rate is about 25%, risk is hard 4 ticks, average winner is 25+ ticks (20-40 ticks). The positive expectancy is obvious. But one should be focused 11 hours per day, 5 days per week without exception and never miss trading signal. This is not suitable for me, but can be used with success by other trader who is able to work 11/5. So, “Can it be defined? Is it quantifiable? Can it be repeated?” – my answer is Yes. Is it static or dynamic? – please explain what do you mean
90's, backtesting... me too Similar thoughts to yours. Was Chuck Le Beau's book that got me going as a young guy out of school. A person can capture all the statistical edge their coded back and forward testing allows. Great. I don't really care what the methods are, they maybe brilliant. It might be the beginning of the best trading record in history. Fantastic! But until you know how to lose, as a trader, a person is just a coder, and they wont ever realize the full future track record from their code. Which will lose.