I saw many backtesting results in public areas with modelling quality less than 90. It's ok if we want to promote some EA or strategy. But if we want to create portfolio so we need this 90%. Because some people may use our portfolios/sets with real money and it is very serious things.
When we are talking that some EA is good or bad so it is just a talking with some backtesting results.
When we are talking about portfolio so it is something different. We need 90%. We need it to start creation of these portfolios/sets.
You're still not answering my initial question mate.
If you're getting 90% on these, then can you explain how you are doing it. What does it take to reach the 90%+ required level. I mean, you're posting presets on some of these, but I don't get the same results as you, nor do I get the 90%+ required rate that you speak of.
So there must be something here that is not right, obviously from my end, but I don't know what that is if the settings you used are the same ones as I have... just doesn't make any sense...
Am I understanding something wrong about Larry Williams Money Management. The whole point of the maximum loss is what is the maximum loss in dollars per lot that we can possibly have. This should equal the amount of worse case scenario which is the stoploss in pips x10 for dollars
so if stoploss is 30pips for 1 lot this would be $300 and this means that the maximum loss we have is $300. This is the value we should use to make sure we minimize our risk. Now for Pricechannel EA, I think the maximum loss is 250 pips in backtesting... This is $2500 and we should use this as the maximum loss amount in the MM input section. I know this will give us less profits than putting a smaller amount, but it is safer I think and will not let us use too many lots that will blow away our account if several trades go against us.
Am I right or have I understood this completely wrong?
What is your modelling quality percentage? Please let us know
If it is less than 90%, then please visit http://www.metatrader.info/node/67 and see how to get it to 90%. Only then will you have some backtesting worth something to think about because if it is less than this (for example 50%) this means that the system is only approximating 50% of the price that happens .. which means that your backtesting is 50% right or 50% wrong which means absolutely nothing!
I've did all this the other day when newdigital showed me the link to download the data , and I get less than 50% or just heaps of losses on the testing, to a point a 10k account gets blown out of the water. But for some reason newdigital gets perfect results in his testing.
These are only on the ones that he tests, others that I have tested, like, the new EA MA crossover that I posted in this folder, is giving me 87%+ and good returns on the account, and tried it live and also giving me fairly good results too.
More on Maximum loss option inLarry Williams MM.
Maybe we should make this a variable to be calculated... because in TradePowers EA, the stoploss is changing every trade according to volatility. So once we know the stoploss for the trade, we set this for maximumloss for that trade only... this way, if we have a low risk trade we can trade more lots.
Lots = AccountMargin * risk/maxloss
Lets say that we have an accountmargin of 10000, risk 0.15 (15%)
if stoploss is 250pips (maxloss = 2500) then 0.15/2500*10000 = 0.6 lots only to trade this time
BUT if stoploss is only 30pips (maxloss = 300) then 0.15/300*10000 = 50 lots to trade ...This means more chances to gain but keeping the risk the same everytime.
I know there is more to calculating Larry Williams Moneymanagement but I tried to simplify only the essential variables in above example to make the example easier to follow...
Again, am I right in this? If we only put the same maximum loss value everytime... sometimes we will be risking too little and sometimes too much... this way the dynamic changine maxloss for every trade follows better... Of course in EAs which have a standard stoploss everytime, then the maxloss will remain the same throughout... The TradersPower EA though changes it's stoploss for each trade and we should be careful.
If you are still getting less than 50% quality then sorry to say and no offence but you have not done it correctly... Make sure you import 1min data first in history center for EACH currency pair you want to test and then open an offline 1min chart for the pair you want and then use period converter to convert to EVERY time frame, 5min, 15, 30, 60, 240, and1440. Then try backtesting with ONLY the dates you have imported ... It works... I promise! . I get similar results to Newdigital and similar to forward testing.
Yes this should be how EVERY EA works when it uses a stoploss (and I think they all should). You should specify a variable called "Maximum Risk %" which should be 1-2% if you're a sane person. Then the number of lots that will be traded by an EA's order depends on this percentage and the calculated stoploss for that trade. This scales your lot sizes while maintaining the same risk as your account grows (or shrinks).
Any EA which does not trade based on this money management principle is a waste of time, because you are gambling, not investing. Most EA's I've seen don't have this, and they use 50% modelling quality and boast amazing results. Those are just a waste of time.
Any EA that's coded should work this way, and should produce spectacular results with at least 90% modelling quality over several years. Only then do you MAYBE have an EA with positive expectation to make you money.
I wish it were possible to achieve 99% modelling quality with 5 years of data to efficiently test EA's. That would give me some sort of confidence in them and I think it would rapidly advance the evolution of EA development because you have a solid foundation to work off. Our current testing methods are very flimsy and don't really tell the whole story. That's why this elite subscription is worth every penny, forward testing is the best thing we can do, it's too bad it takes so long before you find out your EA sucks.
I haven't done it incorrectly, I have imported the 1minute data and copied the data on to all time frames. I read the instructions carefuly and twice over before I tried it then the 3rd time as I was performing each step and they were exactly as they were laid out on that site and I had the same differences that were to be expected, ie, the data periods going up in each time frame within the history center.
So I'm not sure what's happening.
I might go and delete the history, and try it again.
Any way, thanks for your help.
Preset file tpe_h1_gbpusd_nomm_pre-set (attached).
1 lot size.
Every tick, modeling quality 90%,
since 04/08/2004 till 28/04/2006,
Profit factor 2.10.
Preset file tpe_h1_gbpusd_mm_pre-set (attached).
Using Alpari M1 data look at my results:
Preset file tpe_h1_gbpusd_mm_pre-set.
Profit factor 1.32.
What a difference hey?
It is because of MM and deposit size. I started with 25,000 deposit size the EA opened the orders by 3.8 lots from the beginning.
You used 10,000 deposit size and you started with 1.5 lot size.
It's because of MM.
Anyway I am doing optimization just to know the settings to use in EAs. As to MM so it will depends on how many EAs we will have in one portfolio. Somebody suggested to implement MM not for one EA but for All EAs in Metatrader (for one portfolio). May be. I don't know.
Anyway I am optimizing the settings.
And you are right: we should do it without MM first.
Please enable the necessary setting in your browser, otherwise you will not be able to log in.