This is from statistics textbook.
ranchgirl 发表于 2010-7-5 18:37
Sorry, this is BS!
Let P1 = failure rate of indicator 1 (I1), P2 = failure rate of indicator 2 (I2).
Success rate of I1 = 1 - p1
Success rate of I2 = 1 - p2
Assuming I1 and I2 are independent, then joint probability of I1 and I2 failing simultaneously = p1 * p2. The probability of success for only taking actions when both indicators are aligned in the same direction = 1 - p1*p2.
In the case of p1 = 0.5 and p2 = 0.5, the probability of success = 1 - 0.5 * 0.5 = 0.75
Fade move-up or move-down is called mean reversion trade. It is quite a valid trading strategy when market is in range bound environment. Or in market profile speak, one fades the range extreme. Considering that market spends 60% of the time in range bound environment (using a daily time frame), mean reversion trade can be very profitable if executed correctly.
On the other hand, if you try to sell the low and buy the high, you are guaranteed to go broke faster than 散户. Why don't you go and grab data from Yahoo Finance or S&P 500? Set up a simple study - something along the line of selling when price beaches N day low (or N day MA) and buying when prices breaches N day high (or N day MA). Just a hint - numerous studies have been conducted buying high and selling low and numerous studies all show that such approaches yield negative returns and massive draw down. This is even the case when you are in a trending environment because market's tendency to retrace.
At least have an idea what you are talking about and come up with some empirical studies that can back up your belief.
Let P1 = failure rate of indicator 1 (I1), P2 = failure rate of indicator 2 (I2).
Success rate of I1 = 1 - p1
Success rate of I2 = 1 - p2
Assuming I1 and I2 are independent, then joint probability of I1 and I2 failing simultaneously = p1 * p2. The probability of success for only taking actions when both indicators are aligned in the same direction = 1 - p1*p2.
In the case of p1 = 0.5 and p2 = 0.5, the probability of success = 1 - 0.5 * 0.5 = 0.75
csw2002 发表于 2010-7-6 07:43
Please do not use dirty word! Thanks!!
Your calculation is wrong, wrong. If you don't know anything about statistics, then you'd better read a book before putting some non-sense here.
Winning and losing are just two exclusive events. actually, to be more accurate, we should add the third one - break even or within a small range result.
You are mixing up 2 separate events. Losing based on indicator 1 is an independent event from losing based on indicator 2. The joint probability function is given by http://en.wikipedia.org/wiki/Joint_probability_distribution in this case and it is not a Markov process with dependent events as you try to make it out to be.
If you flip a coin to predict market, you have a 50% correct rate.
Now you can flip two coins to predict market. They are independent and each has a 50% correct rate. But the final output is not 75%:25%, it should be 50:50.
If you flip a coin to predict market, you have a 50% correct rate.
Now you can flip two coins to predict market. They are independent and each has a 50% correct rate. But the final output is not 75%:25%, it should be 50:50.
humbletrader 发表于 2010-7-6 12:35
It is definite a better way to teach the same concept!
I know nothing about statistics, but with common sense, I don't really get humbletrader's idea, even you flip thousand coins, there is nothing more than 50:50, then what's the point we are here for.
The key is independent event/indicator (of course, you can argue that most of the indicators are derived from price/volume action). To me, flip coin is one indicator, no matter how many times you try or what kind of coin you flip...
A very serious question that most people talk about here but didn't prove is: how do you say the indicators or factors are independent? Some of you are saying my words are nonsense, no problem, but you have to prove your basic axioms or support your methodology. If you can't prove two things are independent, then your words or calculations are also nonsense. Thanks.
狗仔卡
发表于 2010-7-5 07:38 PM
发表于 2010-7-6 08:53 AM
