Re: Live Trades different than in FSB

Any ideas why FSB trades completely differently to FST?
Even opposite direction!

Re: Live Trades different than in FSB

here better view if you like pics:
MT4
http://my.jetscreenshot.com/1298/20100507-e9zz-225kb.jpg

and
FSB
http://my.jetscreenshot.com/1298/20100507-scto-187kb.jpg

Re: Live Trades different than in FSB

http://my.jetscreenshot.com/1298/20100507-zxmi-78kb.jpg

Re: Live Trades different than in FSB

I know it is even not a good strategy to trade, but before spending any more time one creating a good one, I would like to make sure that I'm doing it for nothing and results will be equal between live and FSB.

Re: Live Trades different than in FSB

here are updates,
!!!!!!----------FSB is Not ==== FST, !!!!!!----------


http://my.jetscreenshot.com/1298/20100509-fgkd-119kb.jpg

Re: Live Trades different than in FSB

These are trades from MT4/FST
http://my.jetscreenshot.com/1298/20100510-3hvh-237kb.jpg
and this from fsb:
http://my.jetscreenshot.com/1298/20100510-tkub-274kb.jpg

Re: Live Trades different than in FSB

This's FST:

http://my.jetscreenshot.com/1298/20100510-r5iv-231kb.jpg


This's FSB:

http://my.jetscreenshot.com/1298/20100510-xakk-215kb.jpg

33 (edited by fxwinner 2010-05-10 20:05:57)

Re: Live Trades different than in FSB

Why on earth it trades This way on FST!!!
All indicators the same and the same strategy!

try it yourself Footon, it is the same broker the same account all the same except trades!!!


Some trades from the MT4.

Guys, any feedback on your trading ???
http://my.jetscreenshot.com/1298/20100510-ez1p-102kb.jpg

Re: Live Trades different than in FSB

side by side trades: COMPLETELY DIFFERENT!!!

http://my.jetscreenshot.com/1298/20100510-deyi-145kb.jpg

Re: Live Trades different than in FSB

Fxwinner,

Are you still using the same strategy posted here on 5/7?

Re: Live Trades different than in FSB

yes ID7 on here.
http://www.mt4i.com/users/fsb02/stats

Go to EA section and tack it down.
No changes to the strategy

Re: Live Trades different than in FSB

Footon,
Sorry to here that, you probably was to hasty to move to life trading (real money) unless you use Cent account.
FSB lacking Cross validation, it is just dirty optimisation tools which is useless in real world, I have EA it picks up random trades it is only a few lines and it can produce enormous profit (MT4 optimisation), as soon as you put it on live feed it dies.

If you would like to understand why strategies stop working, read tops on optimisation http://www.mql4.com/

Cross validation would bring FSB to the first line of testing tools, it is when stratagy is tested on new data, which has not been used to produce optimisation.

Search for Curve Fitting and you will find zillions strategies which can work very well on backtest and never works on life.

If you compare strategies, use only new data it has never learned. This is what you supposedly should get on live trading. (but not a fact)

if you would like to keep your account, never risk more then 1-5% on one trade, with auto trading it should be even less about 0.1 but need to produce a lot of trades to get good results.

Wish you all luck. And if Mr. Popov is not coming soon, I had to put on hold this project.

Re: Live Trades different than in FSB

how did it manage to wipe account if it is conservative ?
you need about 20-90 trades all in losses to wipe account!!! some trades trade 0.1% of account, in this case you need 1000!!! conservative losses to wipe account, with is even theoretically imposable!

Re: Live Trades different than in FSB

just an example of risk management for u footon,
http://www.ninjatrader.com/webnew/TradePerformanceAnal_LG.jpg

Strategy wins only 46%!!! and still in profit !

Re: Live Trades different than in FSB

Footon,
Due to luck of communication with Popov, I might abandon this project till he's in.
I sent you PM , can you please reply.

Re: Live Trades different than in FSB

Speaking of MM, DD doesn't not reflect Risk level in most cases.
if strategy did not lose money on backtest and has risk per trade more then 1% it is already not conservative.
For instance you back test and strategy X has only 10% DD, is it conservative???

however, if you risk 10-15% per trade (SL levels) and you run across 5 consecutive losses which is possible, then what!!!??? you can into position that to recover account very very difficult.

speaking of conservative strategy, it wins all the time over aggressive, if we take larger period of time.

Re: Live Trades different than in FSB

Speaking of cross validation- it's the same as out of sample testing I think,

Yes, it is when backtest is conducted on new data, almost the same as life trade.

Correct validation of strategy has 4+1 stages:
1. Optimisation
2. Cross validation on out sample data
3. Demo trading
4. Trading live on cents
5. Trading live with good MM like <1% of account with less then 10% DD on stages 2-3

Would be interesting to speak to you on-line, send me PM, what is you IM?

Re: Live Trades different than in FSB

Crap, this is really discouraging.. got five strats forward testing on demo but i guess im back to http://sufx.core.t3-ism.net/ExpertAdvisorBuilder/ as i also thought they took a bit too few trades..

Re: Live Trades different than in FSB

ichimoku, they are different worlds.
And not the same weight category. But yes, if this is very discouraging.

Re: Live Trades different than in FSB

Fxwinner,

I took another look at your strategy and noticed you're using enter on previous bar closing for entry.  I would be very hesitant to trust these tests especially on 1 min data...at least for now until FSB backtester is improved.

It is very possible price skipped the previous bar closing value in live trading due to price gaps, spreads...etc.  At it's core, previous bar closing entry is a limit order, so maybe some limit order coding is needed if price is above or below the previous bar closing...?

As for now, I would stick to bar closing and bar opening only if using such small timeframes.

I'm testing a 4hr bar closing strategy right now in FST and everything matches up.

46 (edited by fxwinner 2010-05-19 23:08:53)

Re: Live Trades different than in FSB

Welcome back dmm301, and many thanks for your comments and time to analyse it .
If you confident with bar closing / opening then I will test it now and let you know. yes FSB has a lot of limitations / bugs which needs to be addressed.
so do you think this one will be discreet? http://my.jetscreenshot.com/1298/20100519-gzts-42kb.jpg

This is just a dummy strategy, doesn't matter if it losing or winning, I wanted to have confirmation that, FSB can be = FST as have never seen it yet.

Re: Live Trades different than in FSB

That should work.  How are the results so far?

I would expect a little difference because of data errors.

Re: Live Trades different than in FSB

I cannot compare them any more as this Strategy is 96% on the market, and when I set up FST it opened orders while FSB had them already, so they can't get the same start point now, even theoretically.
I need to build something which  only 30% or less on the market so they can have fair start.

Re: Live Trades different than in FSB

From my testing with other Conditions than "Bar Opening" for Open and "Bar Closing" for Closing conditions, I can say that ONLY those two have created same results in FSB/FST for me. All other conditions do trade almost completely different in FST than they do in FSB.

So make to use as Opening Condition:
Bar Opening

and as Closing Condition:
Bar Closing


Only then the trades to match up perfectly for me.

Re: Live Trades different than in FSB

geektrader, thank you for your confirmation, I would like to hear what other users have say on this matter.

If this is the case, I will add this to "To-Do list of suggestions", ideally I would like to see FSB trades on tick data, (to by pass mentioned by you limitation), the same way MT4 does backtesting whey you do 99PerCent_modelling_quality_MT4_hack
.hst file holds Tick data even though it is M5/M15/etc.

However, FSB would not be able to process so much data on this stage, particularly when there are 2-3 million ticks involved, it has trouble with 100.000 ticks, at least from my experience.