Is it Luck?
Posted: Thu Nov 04, 2010 8:05 pm
I meet with a group of local traders about once a month to discuss various things. We had several folks there at the last meeting who astounded me with their lack of knowledge. This is not to be boastful but I realize perhaps I know more about trading than I even realize. Either that or the "average trader" is dumber than I realize. Either way it was quite a realization.
That being said I'm going to post a few threads with things every trader "should" know and mostly regarding risk and understanding your edge:
No this isn't a Primus song
So how many trades do you have to have to know that your trading expectancy isn?t due to just luck? There is a formula for the calculation of statistical error. This statistical measure provides helpful information regarding the adequacy of trade sample size. The larger the trade sample, the smaller the standard of error. The formula for standard error is:
a. Standard Error = 1/Square Root of N +1
Where N is the size of the sample. Standard error tells us the degree of accuracy of our results. For example, if the average win is $200 with a standard error of 25%, the average win is really $200 +/- 25%. In other words, our average win is most likely to be $150-$250. In the name of conservatism, it is assumed that the average win is likely to be $150. With a sample size of 10 trades, the standard error is 30%. A sample of 30 trades has a standard error of 18%. A sample of 100 trades has a standard error of 10%.
b. So how many trades are enough? In the case of trading system testing more is always better. You cannot get statistical significance without enough trades. This requirement can lead to a problem in the testing of long-term trading systems that trade infrequently. It is sometimes difficult with many types of trading systems to produce a large enough sample size. Other guidelines can help. When a test contains less than 10 trades, extreme prudence must be exercised in its evaluation.
c. That being said, for myself and many other professionals, we don?t consider anything less than 30 trades to be useful. However, if you have a high expectancy system (returning at least $.25 for every dollar risked), even with a worst case scenario of -18% (the standard deviation at 30 trades) average winner that still means we would make $.16 on every dollar risked. Therefore, while it might be nice to know what the last 100 trade results were, it isn?t necessary.
That being said I'm going to post a few threads with things every trader "should" know and mostly regarding risk and understanding your edge:
No this isn't a Primus song
So how many trades do you have to have to know that your trading expectancy isn?t due to just luck? There is a formula for the calculation of statistical error. This statistical measure provides helpful information regarding the adequacy of trade sample size. The larger the trade sample, the smaller the standard of error. The formula for standard error is:
a. Standard Error = 1/Square Root of N +1
Where N is the size of the sample. Standard error tells us the degree of accuracy of our results. For example, if the average win is $200 with a standard error of 25%, the average win is really $200 +/- 25%. In other words, our average win is most likely to be $150-$250. In the name of conservatism, it is assumed that the average win is likely to be $150. With a sample size of 10 trades, the standard error is 30%. A sample of 30 trades has a standard error of 18%. A sample of 100 trades has a standard error of 10%.
b. So how many trades are enough? In the case of trading system testing more is always better. You cannot get statistical significance without enough trades. This requirement can lead to a problem in the testing of long-term trading systems that trade infrequently. It is sometimes difficult with many types of trading systems to produce a large enough sample size. Other guidelines can help. When a test contains less than 10 trades, extreme prudence must be exercised in its evaluation.
c. That being said, for myself and many other professionals, we don?t consider anything less than 30 trades to be useful. However, if you have a high expectancy system (returning at least $.25 for every dollar risked), even with a worst case scenario of -18% (the standard deviation at 30 trades) average winner that still means we would make $.16 on every dollar risked. Therefore, while it might be nice to know what the last 100 trade results were, it isn?t necessary.