Cost Averaging System - page 7
You are missing trading opportunities:
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
Registration
Log in
You agree to website policy and terms of use
If you do not have an account, please register
Aaragorn,
Thanks for your reply. I know I have read it somewhere that the market ranges for a longer period of time than it trends. So, your observations are correct in my opinion. The clue is to find when the markets are trending and it has reached our pain threshold. That is the time to get out and take a loss and start the trades all over again. That is why I introduced the stoploss as a percent of equity feature.
Let us keep on testing and hope we get something going here. By the way, the system trades off closed bars, so any reasonable modelling percent during backtesting should be a good indicator of what really happened in that testing period. That is very unlike systems that are dependent on ticks, which are synthesized by MT backtester and hence prone to false results.Good work.
One thing to remember about finding that pain threashold when we simply take our losses and start over. The sooner we start over the sooner we are winning again. The rewards of holding on longer and increasing positions to try to recover diminish rapidly beyond a certain statistical point when you factor in that you could be making money if you just start over. Tell me more about the stop loss as a percentage of equity feature.
Have a question it lots reduced from 0.1 to 0.01, does that mean less risk.
Compare if you double with each step:
trade.....lots
01.....0.01 0.1
02.....0.02 0.2
03.....0.04 0.4
04.....0.08 0.8
05.....0.16 1.6
06.....0.32 2.56
07.....0.64 5.12
08.....1.28 10.24
09.....2.56 20.48
10.....5.12 40.96
Your margin has to be substantial with the second option.
if my studies of the multilotscalper bare any similarity to this system then the third position in the series has the highest win/loss ratio and the win/loss ratio declines after that.
In other words you can cost average two positions after the initial position with a high probability of recovering but after that the probabilities decline. I am expirementing with maxtrades=2 and seeing what else can be done to keep this trading and profitable without catastrophic drawdown over the long haul.
if my studies of the multilotscalper bare any similarity to this system then the third position in the series has the highest win/loss ratio and the win/loss ratio declines after that. In other words you can cost average two positions after the initial position with a high probability of recovering but after that the probabilities decline. I am expirementing with maxtrades=2 and seeing what else can be done to keep this trading and profitable without catastrophic drawdown over the long haul.
Excellent idea... I would love to see what you find and compare it with others running as is.
As for equity stop, the system calculates the equity when the first trade of a series is put on. It will liquidate the series if the open loss from that series of trades fall below a designated percentage of the equity calculated. Just a way to reduce risk (Corollary to this is that it cuts down on profits too).
Thanks again everyone for your ideas and hard work.
It appears to me that the best intervention of a catastrophic drawdown occurs after the third position in a series. Rather than take a 4th position I think it would be better to look at exit strategies that allow starting a new series with a minimum of loss.
I have looked at the catastrophic drawdown series to see what besides a stop loss could be used to exit them before they get significant. I see that they don't happen immediately but with current programming they don't close for the loss for between 4 and 18 days.
Perhaps the timeline itself could be used as an indicator of "time in series". It stands to reason that the longer any series is in play the smaller are the probabilities that it will win. Unless the trade is over a weekend it appears that most winning series resolve within... 3.5 - 4 hours on average based on the gbpjpy 1m chart results for those series which went to at least 3 positions.
2
6
2
3
1.5
5
5
8
1
3
1
1
1
1
9
13
4
0.5
1
1
7.5
4
1.5
2.5
2
9
0.5
1
10
15
_____
ave = 4.066666667 hours
[/PHP]
there are a small number of exceptions which went a period of days which closed for wins but VERY few. The majority of wins appear to fall inside this timelimit.
based on this I think it's worth taking a look at limiting timeinseries exposure to the market. perhaps scaling out of any series based on it's timeline.
for reference I pulled this data from the attached report. As you can see from the graph there were only actually 4 catastrophic drawdowns over this 2 month range. (plus one series which lost all 3 positions but not large $) However that was enough to really hurt all that good winning.
and what was the time duration of the catastrophic drawdowns?
14
4
1
6
7
_____
ave. = 6.4 days
Maji, do you remember this from my lowly gogetter EA?
[PHP]//+---------------------Close Based on Time function------------Thanks to 'Maji' for this-------------------+
void CloseOrder()
{
double Profit=ThresholdMove*Point;
int total = CountTrades();
for (int cnt = 0 ; cnt < total ; cnt++)
{
OrderSelect(cnt,SELECT_BY_POS,MODE_TRADES);
if ((CurTime()-OrderOpenTime())>MonitorInMinutes*60*MinsMultiplier)
{
if(OrderSymbol()==Symbol() && OrderType()==OP_BUY && Bid-Profit<OrderOpenPrice() )
{
OrderClose(OrderTicket(),OrderLots(),Bid,3,Violet);
}
if(OrderSymbol()==Symbol() && OrderType()==OP_SELL && Bid+Profit>OrderOpenPrice())
{
OrderClose(OrderTicket(),OrderLots(),Ask,3,Violet);
}
}
}
}
//+---------------------------end of close on time code---------------+what do you think?
Aaragorn,
I like your way of thinking. Most of my EAs have a time based exit, as I found that time is a great "indicator" for closing trades to protect profits or reduce losses. I will try to incorporate a time based exit with the system and see how it works. Once the first trade is triggered, the clock will start. After "N" hours, it will exit all the positions regardless of profit/loss. It will not be too accurate over the weekends, but if it is a robust method, then the influence of the weekends should be small and should not make a significant dent in the overall results. Also, GBPJPY is a volatile pair, so the time it takes to move 30 pips there is much less than it takes to move for a pair like EURCHF. Thus, it will be a challenge.
Forward testing
We are fortunate that very experienced members are prepared to invest their time in making this series of EA better for all of us.
I can only do my bit to help by posting forward testing of the basic settings.
After resetting the RSI system to 15 min as suggested things have dramatically changed around. The previous settings results are still in the summary.
John
Forward testing
24 hours on and Cost Averaging v3 Pyramid is showing a good result.
John
We are fortunate that very experienced members are prepared to invest their time in making this series of EA better for all of us.
I can only do my bit to help by posting forward testing of the basic settings.
After resetting the RSI system to 15 min as suggested things have dramatically changed around. The previous settings results are still in the summary.
JohnI think you are being an active participant in this exercise. While some members are spending their time suggesting improvements, and I am spending time in incorporating developments, you are spending your time testing and keeping us updated. Everyone needs to their part and we have a great team effort and hopefully success in terms of a profitable EA. I hope others participate actively and share their results so that we can find out what works and what does not.
Thanks.
Aaragorn, I like your way of thinking. Most of my EAs have a time based exit, as I found that time is a great "indicator" for closing trades to protect profits or reduce losses. I will try to incorporate a time based exit with the system and see how it works. Once the first trade is triggered, the clock will start. After "N" hours, it will exit all the positions regardless of profit/loss. It will not be too accurate over the weekends, but if it is a robust method, then the influence of the weekends should be small and should not make a significant dent in the overall results. Also, GBPJPY is a volatile pair, so the time it takes to move 30 pips there is much less than it takes to move for a pair like EURCHF. Thus, it will be a challenge.
Nothing says that it has to be used the same on every pair right? I hope you roll out the upgrade soon. I'm eagar to get my hands on it and see what it can do. I attempted to do it myself and ended up messing up my forward live test.
Note to self: turn off EA before making changes and recompiling.
I think the series I messed up will be ok. it's just got 3 .2 lot positions open instead of 1
...
I managed to get the two worst ones closed at breakeven minus the swap. So it's back to normal for this system...:D