'Live' Accounts:
01-19-2011, FMT 4.2, Long/Loss, FXSolutions, dealing desk, 1.6023, 1.5983, -40.0
My stats since 11-08-2010: 11 No Trades, 14 Break Evens, 16 Losses, 12 Profits, Total:
-236.0 pips
Pips include slippage, using EA standard settings: 40 SL, 20 BE, 35 TP
01-18-2011, FMT 4.2, Long/Loss, IBFX, non-dealing desk, 1.60230, 1.59828, -40.2
My stats since 01-04-2010: 4 No Trades, 3 Break Evens, 3 Loss, 1 Profits, Total:
-89.4 pips
Pips include slippage, using EA standard settings: 40 SL, 20 BE, 35 TP
Demo Accounts:
01-19-2011, FMT 4.2, 6: Long/Loss at non-dealing desk brokers:
PFG Best, 1.60228, 1.59824, -40.4
FOREX.com, 1.60233, 1.59822, -41.1
MB Trading, 1.60233, 1.59806, -42.7
FXCM, 1.60233, 1.59832, -40.1
Alpari(US), 1.60228, 1.59823, -40.0
FXDD, 1.60235, 1.59826, -40.9
Modified hammy66's system:
01-19-2010, FMT 4.2 EA, set 5 separate GU M15 Charts each separated by 30 minutes using a different magic number.
Using PFG Best Demo Account, Settings: TP=30, SL=44, BE=20, BE+5 pips, UK time zone.
6:00 start, buy trade, hit SL= -44.4 pips
6:30 start, buy trade, hit SL= -46.4 pips
7:00 start, buy trade, hit SL= -44.0 pips
7:30 start, buy trade, hit SL= -45.0 pips
8:00 start, buy trade, hit SL= -44.2 pips
for today
-224.0 pips ... :-(
My stats since 01-11-2010: Total:
-114.5 pips
It was a blood bath . . . First time worse case happens.
Failing to accurately paper trade is a very common mistake. The trading system must be tested through various market conditions to ensure its future profitability and to gain the confidence of the trader. Although many traders do not paper trade their system, it rarely prevents them from proceeding.
Without reliable statistics about how the system performs, the trader does not know what to expect. After a string of a few losing trades, the trader will become disillusioned with the system. Most traders will stop trading the system and move on to the next “flavor of the month” trading system.
Although a trading system may be profitable in general, it is almost certain that, at some point, the system will sustain losses. This is a fact of life. . . Quoted from
http://www.iexpertadvisor.com/ib38.asp
BACK TESTING:
!!!YOU CAN NOT TRUST METATRADER HISTORICAL RESULTS!!! You will get junk . . .
I've been trading FMT now for over 2 months. Nov & Dec are hard months to trade.
But I traded anyway because I wanted to know how robust FMT would be.
I took my real trading data and Mark's real trading data during those two months and compared the data to the my back testing.
I imported M1 data to the MetaTrader platform. The data is free from
http://www.forextester.com/data/datasources.html They have good data from 2001 to 2010. They update at the end of every month. For Jan 2011 data, you would have to wait till the month is over.
Here are the real trading results between Mark and my trades using the standard setting:
Test period from 11-08-2010 to 12-31-2010:
System: Total number of trades, Total real trading pips at the end of the test period
RedGuerrilla: 32, -148 pips
Mark Fric: 29, -83 pips
Here are the back tested results for the time period that Mark and I traded in real time using the standard settings:
Test period from 11-08-2010 to 12-31-2010:
System: Total number of trades, Total back tested pips at the end of the test period
Back Test: 25, -30.7 pips
Mark's standard settings: TP=35, SL=40, BE=20, one trade per day at 6:30 UK time.
As you can see, even comparing real trading results with another trader, you're going to get different results because of differences in brokers. But all the trades were in the same direction, sometimes there were no trades or break evens, but never one trade going short while the other going long. But the key point is that the graphs were all very similar. In other words, the best you can hope for with back testing, with good data, is a very good estimate!
Also remember, Mark changed the parameters starting in Nov, to tweak FMT for better performance, which my back testing confirms did help. But these new parameters have only been in real time use for about two and half months.
Anyway, I decided to do my own tweaking. I tried all the ideas I've thinking about during these pass two months. But hammy66's ideas were the best. Because based on my back testing, I think he has the best settings.
I back tested the system that I've been testing on a demo account, where I trade 5 times every 30 minutes, starting at 6:00 UK time. The results were very bad. Plus, it was just too much risk.
But what I found out, thanks to hammy66, was that the best time to trade is at 6:15 UK time.
Here are hammy66's back tested results for both Nov & Dec period, using his settings:
Test period from 11-08-2010 to 12-31-2010:
hammy66's System: Time, Total number of trades, Total pips at the end of the test period
6:00, 23, +148.0 pips
6:15, 26, +237.4 pips
6:30, 25. +7.1 pips
6:45, 26, -38.3
Total: +354.2
Here are hammy66's results from when FMT started back in Aug, using his settings:
Test period from 08-02-2010 to 12-31-2010:
hammy66's System: Time, Total number of trades, Total pips at the end of the test period
6:00, 57, +491.4 pips
6:15, 65, +627.6 pips
6:30, 68, +539.8 pips
6:45, 63, +296.0 pips
Total: +1954.8
hammy66's settings: TP=30, SL=44, BE=20, BE+5 pips, UK time zone.
Here are back tested results of Mark's system compared to his real results from the time he started FMT:
Test period from 08-02-2010 to 12-31-2010:
Mark's System: Time, Total number of trades, Total pips, Real actual pips for this time period
8:30, 68, +526.0 pips, +267.0 pips
hammy66's system is great, but if you have a $2,000 account, that's about a 9% hit on your account, if you hit the worse case of 4 losses in a day.
I want to do what Mark is doing keeping it at 2% for a $2,000 account, which is $40 or 40 pips.
So starting now, I will discontinue testing the 5 trades per every 30 minutes.
Instead, I will start testing the new settings, SL=40, TP=30, BE=20, BE+5 pips at 6:15 UK time, for just one trade per day.
Here are the results of back testing these new settings for Nov & Dec:
Test period from 11-08-2010 to 12-31-2010:
System: Total number of trades, Total pips at the end of the test period
Back Test: 26, +234.0 pips
Here are the results of back testing these new settings from Aug:
Test period from 08-02-2010 to 12-31-2010:
System: Total number of trades, Total pips at the end of the test period
Back Test: 65, +580.9 pips
Remember this is only a good estimate. In reality, I might have end up with only half the pips, but the odds are good I would have profits instead of losses. Remember, Nov & Dec are the hardest time to trade, but it's the best time to test EA's in real time.
I've learned you should always back test any EA you want to try, with good data, and confirm the best settings.
Is it possible that many EAs on sale today base their back testing results on MetaTrader historical 'junk' data?
Confirm, confirm, confirm, especially when you're going to put real money on the front line . . .