8
$\begingroup$

I want to study if the odds of an up or down day in a forex pairs is 50-50. I just count the total number of up and down days in X years and compare it with the total days. The results are very similar to a 50-50 chance. Now I want to see if by applying an EMA200 filter you have more probabilities of an up day if the closing price is above the EMA, and vice-versa for a closing price below the EMA. The results show that it´s more probable to obtain an up day if the closing price is above the EMA. The question is: Does the test have any bias? I am worried that the results aren't true because of a bias in the test. Because the EMA depends on the price, maybe it´s just obvious that there are most up days if the price is above the EMA. 

$ema = EMA(Close,200); 
foreach (NewDay) { $Totaldays++; 
if (Today(Close) > Today(Open)){ $Totalup++; } 
if (Today(Close) < Today(Open)){ $Totaldown++; } 
if (Today(Close) == Today(Open)){ $Totaldojis++; } 
    if (Today(Close) > Today($ema)){ $Totalabove++; 
        if (Today(Close) > Today(Open)){ $Upabove++; } 
        if (Today(Close) < Today(Open)){ $Downabove++; } 
        if (Today(Close) == Today(Open)){ $Dojisabove++; } 
    } 
    if (Today(Close) < Today($ema)){ $Totalbelow++; 
        if (Today(Close) > Today(Open)){ $Upbelow++; } 
        if (Today(Close) < Today(Open)){ $Downbelow++; } 
        if (Today(Close) == Today(Open)){ $Dojisbelow++; } 
    }

}

Improve this question
$\endgroup$
3
  • $\begingroup$ Hi tn240, welcome to Quant.SE! I find it hard to figure out what the exact experimental set-up is you have, other might too. Can you maybe provide some pseudocode of your algorithm so it is easier to understand what you're doing? $\endgroup$
    – Bob Jansen ♦
    Aug 27, 2015 at 6:11
  • $\begingroup$ Please add this to the question, comments are not meant for this. $\endgroup$
    – Bob Jansen ♦
    Aug 29, 2015 at 11:53
  • 1
    $\begingroup$ FX runs 24/7 depending on when you choose to start/end your days, your result will be completely different! $\endgroup$
    – pyCthon
    Feb 25, 2016 at 20:48

1 Answer 1

Reset to default
-1
$\begingroup$

The design of your experiment may be false.

It may have several biases including, the precision of your experiment (e.g., 0.001), number of days, and this list goes on. I can safely say that the probability of up and down days to be exactly 50-50 or 0.5 up or 0.5 down is very near 0, considering such factors.

The probability may be very close to 0.5 (50%), such as 0.49129 (49.129%), 0.49171 (49.171%), etc., yet those are not 0.5 and that precision is how it would make it uncertain and thats where all the competition is.

Have seen so many retail traders gamble under such false pseudo-math assumptions and have lost large accounts including some of them jobs and houses.

I have been told, what they don't usually consider is that they do not take into account that majority of retail traders may loose (based on bell curve probabilities) if they continue to play for certain amount of time, which is usually the case (for early winners).

Or you might consider EMA as just one indicator, however not everyone make decision based on that.

Improve this answer
$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.