Princerise posted:
If you are using a strategy that gave poor results each year for the past 6 years then the probability for it to give bad results on the coming year is higher than average.
 

Mathematical just not true.

Forex is all about probabilities. No one know the futur, when someone enter the market it's all about probabilities of winning more money than what you lose.

True.
But this counters your first statement.
Probabilities = Stochastics.

You have seen the rest, now see the best!!
I'm starting to have some doubts when i see how ugly the charts are in real accounts compared to backtest.
Just look, it have gone beyond the 20% DD, it was not supposed to happen!!! And it happened just few months after forex cyborg start to trade, just imagine what forex cyborg will face on the long term.

I have done some backtests with 99% precision and i am still confident in forex cyborg to generate some money in the long term. So i willl run each chart independently (without the risk correllation manager) and with low risks of 2% per trade on each chart, the preset will be normal, not conservative, the backtests reveals that conservative mode is too conservative, it don't generate enough money. EA will be allowed to trade on friday evening and monday morning. I will let that run for 3 or 4 months and see the results. If results are good, i will increase the risks depending on how much % it generated each months and also depending on the relative drawdown.

even 99% bt are often wrong

My backtests are done with Tick Data Suite 2, commissions, slippage, variable spread, over a period of 10 years, that's as close as we can get to the reality. And Forex Cyborg is not only surviving on thoses kind of rude conditions, but he is also generating good profits with some really reasonable drawdown.
commissions = 12$ per round lot turn (yeah i know my broker is greedy, it should be 5$ if i had a good broker).
I applied some slippage on the GBPAUD only, there is no simulated slippage on the EURAUD and on the AUDCAD backtests, but it's almost sure that the results will still be good even with added slippage cause GBPAUD is the worst performing pair among thoses 3 and still the slippage didn't stop him from generating good and consistent profits.
Spread is the real spread of dukascopy (variable spread).
As you can see the results seems to be rather correct. That's the reason why i don't use the currency correlation manager, cause i know for almost sure that each pair perform well when taken individually, but i am not sure if the currency correlation manager could cause some pairs to miss some potentially important trades.
If someone know a EA that don't rely on martingale and perform as good as this one, then please let know cause i haven't found any yet.

Hello,

the 99% backtests are not wrong, they just can't predict the future too. No one can do it. Not even what will happen in the next hour.
So even the best trading methods and EAs with the best backtests and with 2-3 years of forward tests can fail someday. All we can do is to analyse the past, develop a strategy that would fit into this market behavior and see if the market will have similar patterns in the future.
If you look into the august, you will see that we had some unusual movements especially for the CHF-pairs.
All multi-currency asian scalper with a forward tests failed to make profit in this month.

We are still confident, that this is just a unusal move but we will be profitable in the future again.

Have a great day everyone!

You've mentioned earlier, that you have another pair in test.
Can you tell us a bit more?

Hello MicF,

it is GBPCHF and the backtest optimization is already finished.
Currently a small forward test (demo) is running.

Here is the backtest result:
Timeframe: 2011 till today
Risk: 1 lot (fixed) per trade.

95% winning trades is nice.
But on the other hand the average loosing trade is immense.

That's what we had in August.