Luck vs. Skill
- Thread starter pak
- Start date
Really neat, thought proving post!
Naturally I'd rather be lucky than good.....
What would be the minimum amount of trades needed to get a good idea if you're profitable and how profitable you are to help determine the luck vs. skill issue? Just as an example you need to play a min. of 100K hands of poker to have a decent idea of a winrate. Is there any way to transfer this into trading?
there was an instance where I made almost 3k due to keystroke error.....that's was pure luck. and just about the only luck you can expect in this game of skillz
tradersboredom
Guest
luck: you buy one stock fully loaded in your portfolio and the next day there is a buyout offer for 10 times what your paid for it.
or you go straight long a out of money call at 10 cents and it goes ten fold because of buyout.
skill: you made your daily revenue quota.
or you go straight long a out of money call at 10 cents and it goes ten fold because of buyout.
skill: you made your daily revenue quota.
A widely used Blackjack simulator, CVData by Qfit Software (link), suggests that you simulate four hundred million hands of blackjack to get a reliable estimate of the mean ("Expected Value") and variance ("gut wrenching pain") of a blackjack card counting strategy.
However they are trying to measure a number on the order of +0.00513, (the player's expected value per hand), to three digits of precision. If you can be happy with less precision, you won't need as much data. They are also operating on computers which simulate tens of thousands of hands per second. A million hands of Blackjack can be played quickly, in simulation! You may not have this luxury.
Since trading is a non-Bernoulli-trials endeavor, there can be (will be) huge outliers on both tails of the distribution. You'll want to give yourself plenty of opportunities to experience some of those little beauties. So I'd suggest a good first start is to perform enough experiments (i.e. trade long enough) to get at least three "minus four sigma" events. The probability of getting a minus four sigma event is about 3.2E-5 (the odds are one in thirty one thousand). To get three of them, you'd need about one hundred thousand experiments. Same as in poker.