Market: What's going on? - Page 2

haymus · #11 (**permalink**) 02-17-2008, 08:04 PM

on the 5th of frb it was released that the BIGDOGs are long on gbp and short yen well i take a long on gj at 208.55 looking at a probability of 220.00 then 240.00
enjoy
and ask for more

philt · #12 (**permalink**) 02-17-2008, 08:17 PM

If your refering to futures, they are all pegged against the dollar..so cable will be long pound dollar and short yen dollar, not long pound yen

haymus · #13 (**permalink**) 02-17-2008, 08:31 PM

well who could have guess 100% on gbp and 5% on yen
i cannot be confused

MrM · #14 (**permalink**) 05-24-2008, 12:37 PM

Hi,
One should wonder: what it the purpose of this forum: is it trying to find the next new TA indicator that seems to correlate with price on a certain timeframe & ccy pair? Is it discussing new technologies in trading? Is it about theoretical background? I think the latest article on the mql4.com website is a great way of finding trading rules that actually work, and this goes more into the direction of datamining than the average forum discussion about indicators etc.

I don't want this thread to be about trading rules, but about optimization, and optimization only.

The fact of the matter is that in-sample optimization is a matter of software ability: the better your technique (be it genetic algorithms to optimize technical trading rules or neural networks), the better you are able to fit your trading model to the in-sample data (the testing history). The problem for 99.99% of the systems is that once it goes out-of-sample (real trading on live data, or testing the optimized system on a new period), the performance drops back drastically.

Maybe we can find a quantitative methodology to express the difference between curve-fitting (overoptimization) and optimizing (adapting setting and rules to get a better performance, without capturing too much properties of the in-sample timeseries).

Now I once read somewhere that to avoid curve fitting, you should look at the sum of squares and once the graph starts turning back up, you should stop to avoid overoptimization. Something else I read is that you should test the variances (st dev squared) and once the difference between the variances of the in-sample and the test results on an out-of-sample timeseries is statistically different (with an ANOVA test), you have over-optimized.

I am looking forward to hearing feedback and opinions on quantitative (statistical) methods to avoid over-optimization!
(Mean square error, regression values, ...)

forexian · #15 (**permalink**) 07-27-2008, 02:55 PM

Ladies and Gentlemen, Good-Day!

A bit of analysis on the pair. EURUSD made another all-time high at 6037, only to rebound sharply. Taking a look at the monthly chart, we see some type of a double top, as well as a bearish divergence with RSI. This is more dangerous than a mere double top, since divergences in monthly timeframes normally lead to losses of somewhere near 500 to 2000 pips. Take a look at the historical divergences.

Another convincing thing. My wave count on the Monthly EURUSD reveals that the pair is currently making its 4th corrective wave of the bigger 5th wave. Confused? Take a look at the chart below.

First downside target happens to be 1.5200 and then 1.4700. After this corrective wave, the final thrust i.e. the 5th of the bigger 5th may lead all the way to 1.7080. I think all this is possible, infact probable, before the US elections later this year.

Further...

Fundamentally, anticipations of a squeezing interest rate differential between euro and dollar are creeping in the markets. Last week eurozone data revealed alerting weakness in the european economy, which might force the ECB to forget the idea of another interest rate hike. Plus crude oil, along with other commodities like wheat and soybeans, have slid in the recent weeks, easing inflationary pressures. This has also eroded the expectations of an increase in rates by the ECB. I dont think europeans will go for a rate cut either. Doing so can amplify inflationary pressures in the region.

Based on all this, i believe a sharp fall in eurusd is quite likely. But before that, the pair can rise till 5850-80 in the short-term. Take a look at the four hour chart below. A little divergence in RSI along with a bullish Gartley is quite evident.

I hope the analysis proves useful.

Happy tradin'