Originally Posted by daraknor

Phoenix just seems to be picking bad trades so far this week. I have reviewed a few users trade logs and Phoenix just went the wrong direction...

Originally Posted by Wackena

This is a quick summary of results from 2006.11.27.

1. Timeframe: 15M
2. Broker: NF
3. Settings file: Default
4. Account: Demo
5. Version: 5_6_06
6. Mode: 3
Wackena

Originally Posted by daraknor

Part of this is due to a more stable signal, part of it is due to less stoplosses being hit through drastically jumping spread. In attempting to make Phoenix more reliable with its signal, it became more overall profitable as well. (to verify this, change signal_count to 1 instead of 5 and backtest.)

Originally Posted by cockeyedcowboy

Daraknor

In all sincerity, can you explain the use of, and purpose of, the statement above expecially the line in red. I have never looked at any of the newer codes versions of the Pheonix just the contest version.

Thanks

The Cockeyed Cowboy

Originally Posted by daraknor

It came from a user on the ForexFactory website and I implemented it with a default of "1" and then increased the number to "5". Phoenix generates many signals to trade, sometimes they are very consistent (all the same direction) and sometimes they switch between buy and sell. So the trade signals may look like this:
buy buy sell sell buy sell buy sell buy buy buy buy sell buy buy buy buy buy buy buy buy sell sell buy buy sell sell.

With a signal_count = 1, Phoenix would enter on the first buy signal and stay in the trade until it exited. When backtesting, that doesn't really tell you if the signal settings are 'statistically valid', it only tells you if it made a profit.

Having signal_count =5 the system would wait for 5 signals in a row, all buy or all sell. In the example, it would wait until signal #19 before it would execute a buy trade.

Since Phoenix is a trend reversal system, waiting longer to enter trades means needing to wait less time before the trend reverses. This reduces maximum drawdown. A signal_count of 30 or 50 is still valid and will still trade. 5 is a conservative number that avoids the choppiest markets. The signals are only traded when they are consistent, and a higher signal count means waiting for a more consistent signal - and waiting longer. The system isn't ideal because the signals could be hours apart - but they are the same signal.

If someone is doing backtest to find the best signals, setting signal_count = 5-50 and maxtrades=20 in Mode1 with no money management is a way to find the best signals for the time period.

Originally Posted by cockeyedcowboy

Intreseting. Has anyone done tests on it to see if there is a corrilation between the higher the number to its profits over the same dataset? As in useing 1, 5, 10, 15, 20, 25, etc. with the same data and leaving everthing else constant.

The Cockeyed Cowboy

Originally Posted by saat2006

Hi wackena:

May i know the risk that u use.Is it 0.05 or 0.3

Thanks