Thanks - I will do that today.  However, here is the weird thing about all 
this.  As I mentioned, after optimizing a strategy for say an Asia mutual fund, 
I can choose similar Asia mutual funds and get performance results that are 
pretty much ballpark the same.  Some are much better, some are a little less 
and many are very similar.  However, when I choose ANY Asia ETF, NONE of them 
even come close to the mutual fund performance.  Yet, the buy & hold 
performance of each is very similar and they all have 95%+ correlation to each 
other.  

Originally, I just thought because the mutual funds were actively managed that 
that made their performance better than the ETFs, which is usually not the case 
here.

Additionally, at initial glance, I notice that the trade dates for ETFs are 
different, as if their EOD closing prices are not heavily correlated to the 
mutual funds.  I don't see how that's possible, but it seems as if the 
optimization isn't even close for the ETFs.  However, when I re-optimize for 
the ETF, I still get subpar results.

My current thinking is that the +DI/-DI, Stochastics and EMA/SMA 
crossover-based systems I optimized for mutual funds do not work well for 
investments with Open-Hi-Lo-Close EOD data.  Is that possible?  I see "H" and 
"L" in the stochastic formula, but in none of the others.

When a system looks at EOD data and determines if a signal was generated via a 
crossover of an indicator, again while using EOD data and not intraday data, 
does the EMA crossover look at the day's high and lows to see if a trigger 
happenned or does it just look at the closing price?  I'm assuming the latter.  
If not, then I need to change the coding.  But how can I change for stochastics?

Long-winded, but I am desperate. - Thanks.

--- In [email protected], "bistrader" <bistra...@...> wrote:
>
> Hard to tell based on limited info provided.  Could do something simple to 
> check.  Take a trade for a mutual fund and for you selected ETF.  Write down 
> the buy price, sell price and performance for a few trades.  Check the prices 
> to make sure correct, in you case closing prices.  Check the performance to 
> make sure correct.  If the trades are correct mathematically, then you are on 
> the right track.  If not, then it should be straight forward to find out what 
> you are doing wrong.  Like, hey it is using the price for yesterdays close to 
> buy or hey it isn't suppose to sell until 2 days from now but it used the 
> price for 2 days prior.
> 
> --- In [email protected], "graphman27" <steve@> wrote:
> >
> > I'm a new user of Amibroker and am spending weeks converting my old Window 
> > on Wall Street simple formulas to Amibroker.  It's been a steep learning 
> > curve, but here is my problem.  An example:
> > 
> > When I take a simple strategy, such as a MA crossover system and run it 
> > with a mutual fund, say a Latin America fund, I'll get profit up 300% in 7 
> > years, as an example.  I'll then take a very similar ETF, that has a 99% 
> > correlation to the mutual fund and has the nearly identical buy & hold 
> > return as the mutual fund, but the performance for the ETF may be half or 
> > even only 1/3 of what the mutual fund was.
> > 
> > I use Yahoo Finance's EOD data, so could that be the problem?  Maybe the 
> > stops I have set up are based on more than EOD close.  I've tried changing 
> > the buy price settings every way possible (Open+1, Close+1, 
> > Average+1...etc.), but I can't get good numbers.  The only change I can 
> > make gets great results, but is unrealistic:  Open+0.  Amibroker support 
> > says that isn't right for EOD data anyway.
> > 
> > I'm hoping my settings are off or some formulas (like stochastics) are 
> > based on more than close and that could be my problem.  However, even the 
> > formulas that are 100% based on close are showing poor results.
> > 
> > Any advice?  Aren't ETFs going to perform better due to intraday trading?
> > 
> > Thanks!
> >
>


Reply via email to