I agree with Maverick, the mathematics of independent multi-system trading can give completely non-intuitive results. That is two bad systems can give a good result for example depending on the overlap of the winning trades and when they come. It's mathematically complex.
You are using an out of date browser. It may not display this or other websites correctly.
You should upgrade or use an alternative browser.
You should upgrade or use an alternative browser.
Fooled by Taleb
- Thread starter jem
- Start date
I agree with Maverick, the mathematics of independent multi-system trading can give completely non-intuitive results. That is two bad systems can give a good result for example depending on the overlap of the winning trades and when they come. It's mathematically complex.
Quote from Handle123:
Most will say we trade Price Action or TA, but when it comes down to it, we are trading emotions and we ALL Predict. We are predicting price will go a certain way, otherwise why do it? There is only one thing that is certain in life - Death, and everything is a prediction. No one can say 100% certain if their heart will beat one more time, the sun will rise or if the highs are in the ES.
I could care less what people say, we all know the "herd" are seldom correct for any length of time. People come in for several months, maybe a couple years and lose all their money, that's great for me. They say ones who making money are lucky, fine with me, I am one lucky son-of-a-bitch for over twenty years. Yep, the markets are random, ka-ching ka-ching, markets are manipulated, ka-ching ka-ching.
http://www.youtube.com/watch?v=ykuvHnNGHNc
Trading can be a bit like blackjack if you counting cards, you know when more facecards in deck you want to bet more. If Price Action showing a Secondary top, baby I am going to load up and sell sell sell. http://www.youtube.com/watch?v=wDI6SEIzwNM
I don't care about proving anything, yep, everything I trade is luck. Or I have figured out what to do based on emotions, granted, I am seeking the dumb and take away their money, the inexperienced who get all excited and get in when the market is near the highs, I will sell them all they want and more, and when total fear is upon them and they getting out, I will will buy all they want to sell but at lower prices than they wished they can get. Yep it sounds ruthless, but that is what the game has always been and will always be.
Trading is like playing Pacman each day, you playing the same game over and over and over, moments of luck, but eventually you learn 99% of rules to reduce the being unlucky. Best game in the world.
http://www.youtube.com/watch?v=tCcxr-fyF4Q
Well said. I have often said that most people have no idea what they really are trading.
Luck, skill or something else, who cares as long you made money. It's an angels on the head of a pin kind of argument. The hardest thing for me was to trust my own instincts. I don't always know why I made the money, but I always knew when I was making money in the trading game.
Quote from Maverick74:
I think you are confusing sales traders who make a nut on transactions with actual buy side risk takers. On the revenue line of most financial statements there is no distinguishable difference. It's lumped into one. Yes, the commission business is good, trading not so much.
Bang on. Those who made the most in the (HFT) gold rush were selling picks and shovels and spreading rumors.
Google confidence intervals, all this is high school statistics. Chance of it being random is a function of the number of players, and the probability of an exceptional result by luck. So, games with high luck inputs and large numbers of participants, like stockpicking or poker, it is harder to distinguish skill - this means that a higher number of exceptional performances are required before we start to consider someone is probably skilled.
A rank piker can win the WSOP, bust out Phil Ivey in a cash game, or make 1000% in one year trading futures, even though the odds are massively against it. And with enough participants, there is a good chance one or two will do it by chance. However, if someone kills it in cash games or hedge fund trading quarter after quarter, and places at WSOP final tables or top of the performance leagues year after year, the chance of it being luck drops dramatically. After enough performance, the chance of it being random is even lower than the chance of a professor losing tenure.
That's pure statistics...you can also increase the accuracy of the skill assessment by using other methods, like fundamental bottom-up analysis. E.g. if someone is au fait with poker theory, maths, and has lots of high stakes experience, is good at reading hands and people, and wins WSOP, it's probably skill. If someone who picked up poker 3 months ago and played with his buddies and a few 5-10 online games, wins WSOP, it's probably luck.
A rank piker can win the WSOP, bust out Phil Ivey in a cash game, or make 1000% in one year trading futures, even though the odds are massively against it. And with enough participants, there is a good chance one or two will do it by chance. However, if someone kills it in cash games or hedge fund trading quarter after quarter, and places at WSOP final tables or top of the performance leagues year after year, the chance of it being luck drops dramatically. After enough performance, the chance of it being random is even lower than the chance of a professor losing tenure.
That's pure statistics...you can also increase the accuracy of the skill assessment by using other methods, like fundamental bottom-up analysis. E.g. if someone is au fait with poker theory, maths, and has lots of high stakes experience, is good at reading hands and people, and wins WSOP, it's probably skill. If someone who picked up poker 3 months ago and played with his buddies and a few 5-10 online games, wins WSOP, it's probably luck.
Like GoC says, this is pretty well-covered by statistics. Maybe not quite high-school statistics, but anyways...
The problem you're describing, jem, is identical to a very classical subject in statistics. Specifically, given a set of dice, which you roll an increasing number of times, how do you "determine" if the set of dice is "unfair". There's a whole group of what's called theoretical "goodness of fit" tests that can be applied to the problem above. The most basic of them is the Chi Square test, but there are others of increasing sophistication (Kolmogorov-Smirnov, Anderson-Darling, etc etc). On top of that, there's a whole lot of techniques that people in computer science have developed to test random number generators that is somewhat applicable (Don Knuth's Art of Computer Programming covers a lot of the empirical tests and also, arguably, the most powerful spectral test).
The problem you're describing, jem, is identical to a very classical subject in statistics. Specifically, given a set of dice, which you roll an increasing number of times, how do you "determine" if the set of dice is "unfair". There's a whole group of what's called theoretical "goodness of fit" tests that can be applied to the problem above. The most basic of them is the Chi Square test, but there are others of increasing sophistication (Kolmogorov-Smirnov, Anderson-Darling, etc etc). On top of that, there's a whole lot of techniques that people in computer science have developed to test random number generators that is somewhat applicable (Don Knuth's Art of Computer Programming covers a lot of the empirical tests and also, arguably, the most powerful spectral test).
Quote from jem:
At what point or what metrics does a trader have to hit to prove non randomness.
Short term metrics on their own are not enough.
I would want to see at least 3 years (more obviously gives you more confidence) of trading results in different market conditions, not just 3 years from a roaring bull market say.
And also, probably more important, evidence of solid risk management.
Risk management is what will save them from talebs black swan.
You would want to see the specifics of the risk management being used, to ensure they are not using a texas hedge etc.
So long term track record under varied market conditions + evidence of very low probability of being hit by a black swan.
Quote from southall:
Short term metrics on their own are not enough.
I would want to see at least 3 years (more obviously gives you more confidence) of trading results in different market conditions, not just 3 years from a roaring bull market say.
And also, probably more important, evidence of solid risk management.
Risk management is what will save them from talebs black swan.
You would want to see the specifics of the risk management being used, to ensure they are not using a texas hedge etc.
what is the etc.
So long term track record under varied market conditions + evidence of very low probability of being hit by a black swan.
"A Texas hedge is a financial hedge that increases exposure to the risk one intended to mitigate. An example would be hedging the purchase of a call option by buying shares of the same underlying.
Origin
The origin of the expression is unclear. However, there are two competing theories among traders. One school of thought says that it refers to the saying 'Everything is bigger in Texas' because Texas hedging increases risk. Others believe that the expression is meant as an insult to the intelligence of the residents of Texas. It is the latter theory that is most popular among traders.[citation needed]. An alternative explanation is that the Texas hedge refers to Texan cattle ranchers who might buy cattle futures contracts while already owning cattle, thereby doubling their risk exposure.[1]
As an alternative to "Texas Hedge", modern traders use other expressions including the "Patrick Murphy, or, David McFarlin Hedge". This is in reference to the two Chicago thousandairs whose hedging strategies have resulted in compounded losses within their sports betting account."
http://en.wikipedia.org/wiki/Texas_hedge
Do you mean that all edges boil down either to detecting the right moment to trade a continuation, or to detecting the right moment to trade a reversion?Quote from StarDust9182:
... I think there are only a couple of edges ( other than arbitrage types) expressed in different ways but really the same edge at an abstract level...
Quote from abattia:
Do you mean that all edges boil down either to detecting the right moment to trade a continuation, or to detecting the right moment to trade a reversion?
That is certainly one way to look at it, but I would say further that there is no "right moment".
The market is a balance of opinions of buyers and sellers. If there was a method that worked 100% of the time, then who would take the opposite side of the trade? When I thought more about it, I concluded that markets are chaotic and statistics are very useful in dealing with my own uncertainty. Your own answer (your own "rightness" just like the your own confidence point made before) will be different, and that is a good thing for a healthy market!
So (for me) ultimately one trades against the herd and one decides what level of risk to take on that trade and when to get out. So whether you use moving averages, price patterns, RSI , VWAP or chicken bones to give you the confidence to enter a trade, is not particularly relevant for me. At a roulette table, we don't see charting or chicken bone analysis done. Yet many traders have shown me a myriad of clever methods to say, that the market will turn here. And indeed, sometimes it does, and sometimes it doesn't!
One edge is the ability to think clearly outside the herd and back your own opinion with confidence even in the face of contradicting evidence. (The 20 year chart of chicken bones says ......)
Novice traders are always looking for magic, there is no magic! Their trading will improve immensely when they see this.
Quote from abattia:
Do you mean that all edges boil down either to detecting the right moment to trade a continuation, or to detecting the right moment to trade a reversion?
I like this definition. Thanks abattia. This makes sense to me although it does not encompass 100% of all trading opportunities it is definitively a good percentage of them.
I also believe that the random versus not random and skill versus survivorship bias is a stupid issue for people who do not trade to waste their time on. If someone daytrades and makes 100 trades a day and he finishes the year up 15% only a moron who is very stupid to understand what is involved will call him lucky. Sample size is the most important criterion.