how do u determine if something is statistically significant?

Discussion in 'Trading' started by Gordon Gekko, Jul 11, 2003.

  1. say you have 20 trades. how do you calculate how meaningful they are? do i need 100? if 100, why 100? how do you determine when you have enough data to make a good judgement?

  2. Significance tests are performed to see if the null hypothesis can be rejected. If the null hypothesis is rejected, then the effect found in a sample is said to be statistically significant. If the null hypothesis is not rejected, then the effect is not significant.

  3. A sample size of 30 (minimum) is the statistically significant magic number. Seems low, but it's a fact. However, you can't go wrong by increasing the sample size to whatever you like.

  4. here is why i'm asking this...

    when i backtest something, the less i go back in time, the better the results are, however, they still aren't bad.

    but how far back do i need to go? don't i want to know what the market is doing NOW? the market was different 8 years ago, why should i include that in my tests?

    what i want to do is walk forward. test only on a fixed number of days, every so often. however, do i need a certain number of trades for the tests to become meaningful? i want to use the least amount of trades, but still statistically significant.

  5. do you know why 30 is the magic number?

    i remember when i read about bet sizing (optimal f), there was a optimum bet size according to your trading statistics.

    what i'm wondering is, if there is somehow an optimum magic number for test results. like maybe there is some calculation i can do to tell me how many trades i need for each test i do.

  6. I don't want to go there. I thought you were asking about tracking your trades.

  7. funky


    paper trade your strategy. if you are making money, then its statistically significant.
  8. I don't know "why", except that it is. You can increase the sample size, and it will not significantly deviate from the results of the original 30. That's the theory, but it 's based on many experiments by many people over many years, so it holds.

    I should add that pollsters and the like routinely use larger numbers (like 1,000+), but it's because they have to diversify their polling locations and population to avoid skewed results.

  9. Sharp


    I don't get the question Gordo. Maybe try looking at it in time frames instead of "number" of trades. For instance, if you are looking for daytrading strategy that you want to use to try and get a couple NQ points, a couple times a day then just go back 30 days or so, count the trades the set-up makes, see if you like the risk/reward. Lets face it, you are probably never going to find a set-up that will work for the rest of your life so, if it worked for the last 30 days then give it a try. If it continues to work then keep trading it until it stops working for you. Seems simple enough. If you are looking for longer term trades like 1-2 week swing trades then maybe check out 6 months to a year of trades with your set-up. If you like the risk/reward then give it a try. If it worked before chances are it will continue to do so. If it didn't work last year but it worked the year before last then why would you think it will work now? Try the set-ups you find that are working now. When they stop working find new ones.
    #10     Jul 11, 2003