Robust EA stands up to 10 year backtest
Here is another run with a very high take profit limit which relys on trailing stops and ends up pyramiding a lot. Trade settings are otherwise the same.
This run still risks about $1000 per trade but the ratio of profit/drawdown is not as good and the equity curve is not as cosistant although the over all profit is higher.
Bars in test 125169
Ticks modelled 21792507
Modelling quality 89.93%
Mismatched charts errors 0
Initial deposit 500000.00
Total net profit 214515.78
Gross profit 497404.62
Gross loss -282888.85
Profit factor 1.76
Expected payoff 518.15
Absolute drawdown 39329.87
Maximal drawdown 56460.68 (10.92%)
Relative drawdown 10.92% (56460.68)
Total trades 414
Short positions (won %) 153 (19.61%)
Long positions (won %) 261 (32.95%)
Profit trades (% of total) 116 (28.02%)
Loss trades (% of total) 298 (71.98%)
Largest
profit trade 21456.54
loss trade -1146.97
Average
profit trade 4287.97
loss trade -949.29
Maximum
consecutive wins (profit in money) 10 (79087.72)
consecutive losses (loss in money) 27 (-26978.81)
Maximal
consecutive profit (count of wins) 79087.72 (10)
consecutive loss (count of losses) -26978.81 (27)
Average
consecutive wins 2
consecutive losses 6
*** WITH NO MONEY MANAGEMENT
Initial deposit = 500,000, profit = 80,000 = 16% over ten years which is less than 1.5% when compounded annually, sorry but I can get better rates from the banks. And 200,000 profit is still only just over 3.5% compounded annually over 10 years. still not enough to consider putting any money into it. sorry.
Thanks for taking the time to read.
I did not try and sugar coat this EA by applying a moneymanagement system. I wanted to show the stability of the trading system by itself.
One can achieve the same profit with a lesser starting amount or achieve more profit by turning up the risk amount. One could also apply a money management strategy which adjusts the risk based on the amount in the account among other things to further increase gains by compounding the profits. The starting deposit in the previous test is somewhat irrelevant unless a 1.91% drawdown is your maximum tollerance and you want to skim the account of all proffits all the time.
I've posted the above results mostly to show the system (by itself) stability and robustness over 10 years.
But if I wanted to apply a quick money management system and reduce the starting equity then, depending on your risk tollerance you might choose to add this moneymanagement strategy (This is the same system with crude money management) I would imagine anyone interested in the exclusive rights would want to put their own moneymanagement strategy in or if not I can make a much better one than this I'm sure.:
***CRUDE MONEY MANAGEMNET
Bars in test 125169
Ticks modelled 21791470
Modelling quality 89.93%
Mismatched charts errors 6
Initial deposit 20,000.00
Total net profit 366,662.52
Gross profit 966562.33
Gross loss -599899.82
Profit factor 1.61
Expected payoff 885.66
Absolute drawdown 6568.05
Maximal drawdown 68689.11 (20.44%)
Relative drawdown 41.05% (23441.73)
Total trades 414
Short positions (won %) 153 (44.44%)
Long positions (won %) 261 (51.34%)
Profit trades (% of total) 202 (48.79%)
Loss trades (% of total) 212 (51.21%)
Largest
profit trade 31858.39
loss trade -13340.00
Average
profit trade 4784.96
loss trade -2829.72
Maximum
consecutive wins (profit in money) 8 (51541.59)
consecutive losses (loss in money) 8 (-13602.10)
Maximal
consecutive profit (count of wins) 82146.55 (5)
consecutive loss (count of losses) -43106.03 (6)
Average
consecutive wins 2
consecutive losses 2
Yes, of course you're right, the money management is very important. Money management should be designed to the specific risk tollerance of the account in my oppinion. The more smooth the core trading system is the more aggressive you can apply the money management. Here's a very simple money management strategy which simply risks 5% of the equity in the account on each trade. I tested a few which incorporated the current profit:drawdown ratio but the simple percent is smoother.
*** WITH SIMPLE MONEY MANAGEMENT
Ticks modelled 21791470
Modelling quality 89.93%
Mismatched charts errors 6
Initial deposit 50000.00
Total net profit 793356.30
Gross profit 2467942.38
Gross loss -1674586.09
Profit factor 1.47
Expected payoff 2013.59
Absolute drawdown 18072.15
Maximal drawdown 231770.57 (24.30%)
Relative drawdown 41.08% (22256.35)
Total trades 394
Short positions (won %) 151 (44.37%)
Long positions (won %) 243 (50.21%)
Profit trades (% of total) 189 (47.97%)
Loss trades (% of total) 205 (52.03%)
Largest
profit trade 121888.80
loss trade -45630.57
Average
profit trade 13057.90
loss trade -8168.71
Maximum
consecutive wins (profit in money) 8 (151947.75)
consecutive losses (loss in money) 8 (-37981.87)
Maximal
consecutive profit (count of wins) 245032.00 (3)
consecutive loss (count of losses) -129556.17 (3)
Average
consecutive wins 2
consecutive losses 2
Regarding how much time I spent; I guess the last month has gone something like this: remove filters then optimize, then replace filters and test, test and more testing, then remove filters and optimize other filters, then replace the filters (Always carefull to optimize only one set at a time) and test some more and right when I think I'm ready I dream up a new feature or filter so I then remove all filters, program the new features, test it, debug it, optimize it, put the other filters back on, test them all together, test them some more. Then I think of a new feature I must have etc.etc... Next thing I knew it's a new month! Time flies when you're having fun!
I'm sure running new currencies won't take as long. I'll have to test some pairs with the least amount of corrolation as possible.
Some of the optimization runs were 100 hours+. But how much time I spent exactly doing what I can't really say, I was having too much fun, too bad fun doesn't pay the mortgage.
endo,
I created a Trade Simulator in Excel several months ago to help extend my analysis of EA's. As you know, trading is a stochastic investment model. As such, it means that if you get a 49% win percentage for one 10-year period, it does not mean that you will get a 49% win percentage in every 10-year period. That is why I created the Trade Simulator. In the simulator, I can adjust the Trade Size, the Win Pct, the Reward-Risk Ratio, Max PIP Loss/Trade and the number of trades included in the stochastic test.
A formula that I use to help evaluate EA's is: Preliminary Profit Factor = win pct/loss pct x reward/risk
When the PFF is over 3.0 the EA is very stable is you do not abuse the Trade Size, where Trade Size = Account Balance/(# of Lots x 1000). Trade Sizes of 25 to 50 tend to work nicely when the PFF is over 3.0.
But when the PFF is hovering in the 1.5 range like your EA, the Trade Size should be 75 to 100 to make it "reasonably stable". However, with only 40 trades per year its hard to obtain a high ROI. Can you trade multiple currencies to increase the number of trades per year? If you trade 10 currencies and get 400 trades per year instead of 40 trades per year, your EA can produce some nice ROIs
In the very long haul (ie thousands and thousands of trades), a 1.48 profit factor work works out nicely. The casinos in Vegas work with a 1.05 to 1.10 profit factor in their stochastic model, but what makes that stochastic model work is the fact their "Trade Size" probably exceeds 1,000,000 when you place a $2 bet. The massive Trade Size coupled with millions of interations using that stochastic model make it very stable even though the profit factor is just a little over 1.0
I do not have enough info to properly run your EA stats through my Trade Simulator (i'm missing the Max PIP Loss/trade), but since you are using a breakout stategy I assumed 40 PIPs. What I found is you should be able to obtain an annual ROI of 3.4% with drawdowns staying below 5% using a Trade Size of 100. If you decrease the Trade Size to 50 (which means the actual size of the trade relative to your account balance increases) your annual ROI jumps to 7.2% and the drawdowns will increase to 10% max, and ocassionally 15%.
The 3.4% ROI in the previous paragraph would be 40% if you could achieve 400 trades per year, and the 7.2% ROI would be 100% if you could obtain 400 trades per year.
Note: Those numbers need to be adjusted if the Max PIP Loss/Trade is not 40 PIPs.
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
You agree to website policy and terms of use
Ticks modelled 21792507
Modelling quality 89.93%
Mismatched charts errors 0
Initial deposit 500000.00
Total net profit 89041.04
Gross profit 274756.43
Gross loss -185715.39
Profit factor 1.48
Expected payoff 215.07
Absolute drawdown 8249.14
Maximal drawdown 10102.53 (1.91%)
Relative drawdown 1.97% (9905.14)
Total trades 414
Short positions (won %) 153 (44.44%)
Long positions (won %) 261 (51.34%)
Profit trades (% of total) 202 (48.79%)
Loss trades (% of total) 212 (51.21%)
Largest
profit trade 3078.00
loss trade -1071.00
Average
profit trade 1360.18
loss trade -876.02
Maximum
consecutive wins (profit in money) 8 (10625.07)
consecutive losses (loss in money) 8 (-6858.75)
Maximal
consecutive profit (count of wins) 11232.45 (7)
consecutive loss (count of losses) -6858.75 (8)
Average
consecutive wins 2
consecutive losses 2
Notice how I show you the full 10 years data instead of showing only 8 or 5 years which would make this EA look even better. I wanted to show/optimize for the full ten years as anything less is insufficient for back testing and optimizing in my oppinion.
*** WITH NO MONEY MANAGEMENT-8.8:1 profit:drawdown Ratio w 10 years to date data.
-Simple BreakOut Strategy with filters.
-No money management strategy has been applied yet. (One could apply a money management strategy to eek out more gains as equity increases.)
-Lotsizes are adjusted depending on the distance of the stop from the order during the trade setup. (adjustable)
-NOT a scalping strategy, a few missed pips here and there from non actual tic by tic data should not matter. (Do you think the founders of the turtles had tic by tic data?)
-Stops and targets are placed/trailed for every trade.
-No hedging or rediculous floating losses.
-Each filter group is applied and optimized seperately instead of just dropping all filters on and curve fitting the crap out of it.
-Other common sense techniques have been used to prevent/test for over optimiazation such as described in the book "Way of the Turtle" (such as jiggle testing.)
-35+ different external settings to experiment with
-Works on other pairs but haven't done the carefull optimization yet for anything other than GBP/USD
-as a side effect, if there is another set up and the last setup is still active, it will fire again in the same direction occasionaly pyramidding adding no more than 1 trade per day and initial risks are calculated seperately. (I.E. max risk aprox $1000 per trade)
I wanted to just get some feedback on this data as I am considering selling the 1 time exclusive rights to use/resell this EA and its source as I am going broke working on this full time. Thanks for having a quick look.
endo@done.ca
Ben Dixon
www.done.ca