Back test starting from $100 to over 180 million | Page 3 | Forex Factory

Options

Go to
Permalinks
Print Thread

Similar Threads

Back Test vs Forward Test 6 replies

High Leverage Trading, Going For A 100 Million, 100% A Week 149 replies

High Leverage Trading, 100k To 100 Million, 100% A Week 23 replies

Why EA test is forward test better than back test?? 1 reply

Cash & Carry (Grid) took over 180 pips out of the market yesterday and under 1% DD 1 reply

Attachments: Back test starting from $100 to over 180 million
Exit Attachments

Back test starting from $100 to over 180 million

  • #41
  • May 16, 2015 6:25am May 16, 2015 6:25am
Quoting Robertk
"If the ea only makes a portion of the 100$ to 180.000.000$ on a consistent basis i might need to look for serious souls(buyers of this ea)"
Sounds like a sale pitch to me. Should be in the commercial section. If it could do the above, it would not be for sale. Not posting it, negates "real" testing...draw your own conclusions.
Sorry in the off chance I am wrong, but prove me wrong. Post the ea for testing...if it is commercial, put a 30 day expiration on it. Solved. Back tests in general are unreliable, and running them with virtually no spread produces fantasy results...gets the crowd going. I have a bucket full of such ea's, but they all fail in real market.
Just my opinion.
Best, Josh
  • #42
  • May 16, 2015 6:52am May 16, 2015 6:52am
Quoting ibthescottyb
I've been on this forum a long time and seen alot of strange things. Learning different things from so many different people from all walks of life. Using those tips and tricks and what I have gathered on my own research, I started creating my own MT4 expert advisers. To date, I have designed EA's, indicators, systems, etc. I will be the first to admit that each of the trading systems I have designed failed at some point. Martingale Trading systems eventually blow an account, Time Based Trading does not hold up to the test of time (no pun intended)....
A trader comes with a wonder EA that makes a million in a demo from a million random tries and he comes to FF and post his holy grail. He'll be careful not to ask for money on the open forum but you can bet your mothers grave that a zillion stupid newbies have flooded him with private messages asking for a piece of his magic portion which would not be free as usual. The shit ill finally hit the fan when all the clueless buyers bet their 401k in a real account with this same wonder EA and they get the shock of their lives cause they're screwed from behind with no lubrication!!!. Hell Yea! Ive seen this movie before
  • #43
  • May 16, 2015 6:57am May 16, 2015 6:57am
Go sell your martingale for billions to Goldman....

But it looks a scam to me. To 100 dollars to 180 millions...
COT: The precious data ignored
  • #44
  • May 16, 2015 7:16am May 16, 2015 7:16am
Curve-fitting
  • #45
  • May 16, 2015 7:29am May 16, 2015 7:29am
Spread is in points not pips in the tester so set it for 20 and run the test again...

Absolute CRAP!!!
Solutions ONLY!!!
  • #46
  • May 16, 2015 7:47am May 16, 2015 7:47am
Quoting tablet8
{quote} now I tell you the truth that this guy robot is Scam: first: his broker is a 5 digit broker(look at his backtest you will see it) and the spread for backtest is 2 in the photo that means the spread is almost Zero in his account and fixed during backtest, this is impossible in live account (if u dont accept zero spread: example: bid price= 1.11750 ask price =1.11752 how much is the spread? 75-75=0 + 0.2 ) I have ecn account even in ecn account the minimum spread in a complete flat market is 0.8 +mimimum .1 second reason: spread slippage this...
Solutions ONLY!!!
  • #47
  • May 16, 2015 7:55am May 16, 2015 7:55am
spread is 2 and modeling quality is 25%. Download full tick history and set real spread and you will see how quickly you will loose all account With this EA
  • #48
  • May 16, 2015 7:58am May 16, 2015 7:58am
Quoting Dezil
spread is 2 and modeling quality is 25%. Download full tick history and set real spread and you will see how quickly you will loose all account With this EA

As I mentioned in a earlier post:

2. LOL. Tablet was also right on the spread on the backtest. I have been testing EA's for so long (prior to the NFA changes of 2009 and 5 digit brokers), I never knew to use 20 instead of 2. Talk about a lightbulb moment. My EA today does not work when setting to 20 with CURRENT settings.

So, looks like the will be spending working on the setting to take into consideration proper spread. In design it is just a breakout EA that can be used on any TF.
  • #49
  • May 16, 2015 8:18am May 16, 2015 8:18am
I am a beginner and a little intermediate in MQL4 programming I have coded most of my idea myself , but now I have a new idea which I can not code myself, it need a very very experienced programmer , I need fibo retracement + pattern recognition + res/sup lines for my idea (really advanced mql4)
if anyone can help for free !! I explain the idea ,
u can have it for yourself and even sell it.
i only want an mql file for personal use.
  • #50
  • May 16, 2015 9:32am May 16, 2015 9:32am
EDIT TO FIRST PAGE- One of my fellow traders discovered a flaw in my testing. So I am sorry to report that this back test is flawed and would never work on a real account. Happy Trading.

If was a fun exercise though!
  • #51
  • May 16, 2015 9:56am May 16, 2015 9:56am
Quoting ibthescottyb
EDIT TO FIRST PAGE- One of my fellow traders discovered a flaw in my testing. So I am sorry to report that this back test is flawed and would never work on a real account. Happy Trading. If was a fun exercise though!
STOP!

The first thing you should learn about backtesting, is that the parameters and settings are irrelevant!

The data obtained from these is completely worthless!!

The value of running backtests is to verify your strategy. ONLY!

Now...you have an idea, and it seemingly works as far as your triggering and profiting methodology are concerned, this is a good thing. If you throw the baby out with the bathwater now, you will never know if it could have, or may have produced anything close to acceptable profits.

100-180 million=pipe dream!!

But that doesn't mean that it can't be profitable with a reasonable amount of risk.

Using stats from the BT to either pass or fail an EA from forward testing is a fail in itself. Drop that sucker into a demo account, let it run for a couple of months and then make a decision.

Better yet, if you drop it into a the forum, others will work with you on the idea.

If you stop now, you are left with 2 fists full of fuck all!
If you continue on, you will both learn a lot about market, gain knowledge from others optics and possibly turn this into a steady gainer with a reasonable risk.

Keep on keeping on, Scotty.

Never Say Die!

Money Can't Buy Happiness. Poverty Can't Buy SHIT! You Choose!
  • #52
  • May 16, 2015 10:05am May 16, 2015 10:05am
Quoting Magix
{quote} STOP! The first thing you should learn about backtesting, is that the parameters and settings are irrelevant! The data obtained from these is completely worthless!! The value of running backtests is to verify your strategy. ONLY! Now...you have an idea, and it seemingly works as far as your triggering and profiting methodology are concerned, this is a good thing. If you throw the baby out with the bathwater now, you will never know if it could have, or may have produced anything close to acceptable profits. 100-180 million=pipe dream!! But...

Thanks Magix - Oh, I'm not giving up by ANY means. The possibilities are endless. But, I did want to make it clear that the purpose of the thread was to show off a backtest. But that the test itself was flawed as I was using an incorrect setting thanks to my fellow traders. Finding out things like that was one of the reasons I created the forum.

I have seen live forward testing results that are awesome. But whether that was just a lucky day is yet to be seen. I will be posting the EA here with an explanation of the settings for people to play with once I get them written up.
  • #53
  • May 16, 2015 10:18am May 16, 2015 10:18am
Quoting ibthescottyb
{quote} I have been testing EA's for so long (prior to the NFA changes of 2009 and 5 digit brokers), I never knew to use 20 instead of 2.
I am sorry but... did I understand correctly - you have been testing EA's for 6 years without knowing how spread is being calculated?
  • #54
  • May 16, 2015 10:24am May 16, 2015 10:24am
Quoting Indrek
{quote} I am sorry but... did I understand correctly - you have been testing EA's for 6 years without knowing how spread is being calculated?
Embarrassing I know. but, I always thought it was in pips when back testing. But, you learn something new everyday. Just goes to show why all of my EA's did great in testing but lost in real environments.
  • #55
  • May 16, 2015 11:11am May 16, 2015 11:11am
Quoting tablet8
{quote} Hey Guys here is this robot: never trust this robots I can give u many similar robots making 1 million dollars from 30 dollar in a monnth in backtest and sometimes in demo forward test but in live account they all fail the reason is spread slippage which is fixed in demo and backtest but in live accounts it is killing every scalping robot another similar robot to this is angry beaver u can find it in mql5.com it makes 1000000 dollar in a month but in live account it help you loose faster {file} if u don't know about robot and want to buy...
Thanks tablet8 I agree with you! And I appreciate your help.
  • #56
  • May 16, 2015 11:25am May 16, 2015 11:25am
Quoting ibthescottyb
{quote} But, you learn something new everyday.
And if not every day then every six years. I have seen much BS in FF but still some people can surprise me. Wonderful community
  • #57
  • May 16, 2015 11:35am May 16, 2015 11:35am
Quoting Indrek
{quote} And if not every day then every six years. I have seen much BS in FF but still some people can surprise me. Wonderful community
Being able to adjust spread in the strat tester is still a relatively new feature...especially if you've only recently came back to working with this tool.

Prior to the new upgrades, it was always fixed on current, leaving weekend bts worthless due to most brokers having a spread that was insanely higher than normal...Seeing a 20 pip spread on a backtest on older versions, wasn't actually that uncommon...so, if the last time somebody was fucking around with the Strategy Tester in their terminal was an older version and they have just come back to using this tool, it's easy to see how somebody could get confused about the spread now being in points vs pips.

He never said he was confused about calculating spread...just that he was mistaken with how spread was being applied in the strat tester.

Busting Scotty's nuts because he is unfamiliar with this tool is of no value.
Money Can't Buy Happiness. Poverty Can't Buy SHIT! You Choose!
  • #58
  • May 16, 2015 12:23pm May 16, 2015 12:23pm
Ok - now that I have a basic user guide written, should I keep it on this thread and post it with the EA or start a new thread that can be specific to the EA and not a faulty backtest?
  • #59
  • May 16, 2015 12:49pm May 16, 2015 12:49pm
New thread started for EA testing only.

http://www.forexfactory.com/showthre...11#post8266611
  • #60
  • May 16, 2015 1:12pm May 16, 2015 1:12pm
You can not randomize your way to a profitable 'EA'. You can not stumble upon a profitable set of parameters and indicators for an 'EA'. You can not trade with "Expert Advisors" that are somehow experts without you yourself being an expert. The only profitable expert advisors you will find were 'profitable' in backtests due to chance, not due to an understanding of the market. Once you become an expert, you don't use "Expert Advisors", you use algorithms or models. The algorithms don't come from stumbling around, they come from years of research and exploration of market data. You don't start with algorithms, you start with data research. If you haven't opened up an excel sheet, R instance, or used Julia, C, or C++ or some other tool to do data research then you most certainly have 0 chance to become profitable using "Expert Advisors".

The same can be said to those that stumble around with machine learning. You do not fumble around with neural network settings hoping something works. If something sticks, it sticks very little through years of old data, likely meaning little for new data. Yes, even if you use cross-validation.

If you do not look at each and every single trade your "Expert Advisor", algorithm, or neural network / machine learning model took and evaluate what went right and wrong for each one, then you most certainly are doing something wrong and are relying too much on brute force chance to get your backtests to signal in bright neon lights "Profit!".

With this in mind, I am not an expert in this field so take everything I said with a grain of salt. I hope I have saved many a lot of time and anguish with this reality wakeup call, that I myself could also have used a while ago.
1 trader viewing now