MA Cross Optimization EA (very cool) | Page 2 | Forex Factory

Options

Go to
Permalinks
Print Thread

Similar Threads

New very cool CandleTime and session time Clock indicators 189 replies

Rsi cross indicator? (NOT cross 50, etc) 44 replies

very cool Trading Panel EA needs some adds 19 replies

Royal Cross/Fresh Cross 7 replies

World Market Hours - Very Cool 8 replies

Attachments: MA Cross Optimization EA (very cool)
Exit Attachments

MA Cross Optimization EA (very cool)

  • #21
  • Jul 13, 2008 11:09am Jul 13, 2008 11:09am
ok, thank you for info.
keep up the good working!
  • #22
  • Jul 13, 2008 1:34pm Jul 13, 2008 1:34pm
Maybe read the whole thread next time.... not jump down past the meat of the debate. Thanks.

Quoting jane
M.A.C.Doug is 100% right.

genetic algorithyms create endless # of 'WOW!!!!' systems, perfect traders of yesterday. but all lose $$ in today. try if don't believe. you will learning hard way.

it is a fun game paint picture of past, but it is not real trading.
  • #23
  • Jul 13, 2008 2:17pm Jul 13, 2008 2:17pm
that's funny.... all charts are based on numbers, so somehow your "many fabulous EAs" are trading on alphabet soup and bob dylan? is that how it works?

ps) your modeling quality is N/A (DOH!)

Quoting jane
p.s. true profiting systems have no parameterization numbers ... only truth about markit, not numeraligy.
  • #24
  • Jul 13, 2008 2:28pm Jul 13, 2008 2:28pm
Quoting tdion
Maybe read the whole thread next time.... not jump down past the meat of the debate. Thanks.
i did read it all, and i was all the way ....
you're welcome.
  • #25
  • Jul 13, 2008 2:30pm Jul 13, 2008 2:30pm
Quoting tdion
that's funny.... all charts are based on numbers, so somehow your "many fabulous EAs" are trading on alphabet soup and bob dylan? is that how it works?

ps) your modeling quality is N/A (DOH!)
yes, alphabet soup (but in german), and bob who??

i use real tick-data for my backtesting, hence n/a.
  • #26
  • Jul 13, 2008 2:30pm Jul 13, 2008 2:30pm
right back at you sister!

Quoting jane
i did read it all, and i was all the way ....
you're welcome.
  • #27
  • Jul 13, 2008 2:31pm Jul 13, 2008 2:31pm
normally that produces 99% MQ....

Quoting jane
i use real tick-data for my backtesting, hence n/a.
  • #28
  • Jul 13, 2008 2:33pm Jul 13, 2008 2:33pm
Quoting tdion
normally that produces 99% MQ....
not in my setup, nope.
it is re-programmed for total accurate, including variable spread.
  • #29
  • Jul 13, 2008 2:36pm Jul 13, 2008 2:36pm
I highly doubt this being tick data.

What version of metatrader are you using?
  • #30
  • Jul 13, 2008 2:38pm Jul 13, 2008 2:38pm
Quoting Ronald Raygun
I highly doubt this being tick data.

What version of metatrader are you using?
version # is on statment.
no matter your doubt ... that is okay.
i can run it on standard backtester if you want, no problemo. (probaby better results!)
  • #31
  • Jul 13, 2008 2:40pm Jul 13, 2008 2:40pm
could you do it on a different thread?

Quoting jane
i can run it on standard backtester if you want, no problemo. (probaby better results!)
  • #32
  • Jul 13, 2008 2:42pm Jul 13, 2008 2:42pm
ok, sorry -
  • #33
  • Edited 3:54pm Jul 13, 2008 3:33pm | Edited 3:54pm
Say whatever you want to.... this is impressive..... Again, I was bored so I ran this thing on the USDJPY for about 60 seconds....

Daily USDJPY 16/18 cross: 1979 - 2008.... 22,000 pips.... is there an invisible edge? 703 trades is more then a couple of hundred. I've never taken MA crosses seriously EVER, but this has me thinking twice.
Attached Image (click to enlarge)
Click to Enlarge Name: TesterGraph.gif Size: 15 KB
  • #34
  • Jul 13, 2008 3:41pm Jul 13, 2008 3:41pm
I think your onto some pretty profitable stuff here.
Same Whore .... Different Dress
  • #35
  • Jul 13, 2008 3:55pm Jul 13, 2008 3:55pm
Quoting tdion
Say whatever you want to.... this is impressive..... Again, I was bored so I ran this thing on the USDJPY for about 60 seconds....

Daily USDJPY 16/18 cross: 1979 - 2008.... 22,000 pips.... is there an invisible edge? 703 trades is more then a couple of hundred. I've never taken MA crosses seriously EVER, but this has me thinking twice.
Looks impressive, can you post the MT4 stats page (number of wins, maximal drawdown, profit factor, etc) also?

Thanks,
David
  • #36
  • Edited 4:44pm Jul 13, 2008 4:01pm | Edited 4:44pm
MQ is irrelevant, as all trades are opened and closed on the close of the bar in the selected timeframe.

Going back here:

http://www.forexfactory.com/showpost...32&postcount=9

we start getting into the system surviving sideways and trending markets, and still keeping relatively linear output.


EDIT) the problem i see here is the R:R is 1:1.... it can go infinitely either way.... so it puts high pressure on the system to win around 50%.... i think it would do a nosedive at 40%....
Attached File(s)
File Type: zip stats.zip   28 KB | 665 downloads
  • #37
  • Jul 13, 2008 4:53pm Jul 13, 2008 4:53pm
Quoting tdion
MQ is irrelevant, as all trades are opened and closed on the close of the bar in the selected timeframe.
Q1. Excuse my ignorance, but what is MQ?

Also, just want to confirm the following:
a) These are SMAs, not EMAs.
b) A 3 pip spread has been allowed for on each trade.
c) Entry is on the open of the candle following the cross.
d) There is no SL, position is simply exited when the opposite position is entered. Hence system is always "in the market".
e) Pos size is always 0.10, so we're effectively counting straight pips, with no MM.
f) We are looking at daily candles only, i.e. no lower TFs need be considered.

Q2. How does it allow for swap?


OK, so we have a win rate of approx 50%, and ave win / ave loss of 1.618 (hey, Fibo - LOL) over 712 trades. Illustrates how a profit factor of 1.57 can be very profitable.

I'm not well enough versed in statistics to know how to measure how much a result is dependent on the optimization itself. However, using common sense, if performance collapses dramatically with only a slight adjustment to the 16, 18 values (i.e. which would cause the stop-and-reverse to occur just one day earlier/later on only a few trades), then I suspect that suggests the dependence is dangerously high. Hence I would check peformance using values between (say) ±2 for each MA, and see to what extent the results differ.

David
  • #38
  • Edited 5:46pm Jul 13, 2008 5:02pm | Edited 5:46pm
Quoting hanover
Q1. Excuse my ignorance, but what is MQ?

modeling quality

Also, just want to confirm the following:
a) These are SMAs, not EMAs.

yes

b) A 3 pip spread has been allowed for on each trade.

yes

c) Entry is on the open of the candle following the cross.

yes

d) There is no SL, position is simply exited when the opposite position is entered. Hence system is always "in the market".

yes

e) Pos size is always 0.10, so we're effectively counting straight pips, with no MM.

yes

f) We are looking at daily candles only, i.e. no lower TFs need be considered.

right.

Q2. How does it allow for swap?

MT4 automagically includes it.

OK, so we have a win rate of approx 50%, and ave win / ave loss of 1.618 (hey, Fibo - LOL) over 712 trades. Illustrates how a profit factor of 1.57 can be very profitable.

I'm not well enough versed in statistics to know how to measure how much a result is dependent on the optimization itself. However, using common sense, if performance collapses dramatically with only a slight adjustment to the 16, 18 values (i.e. which would cause the stop-and-reverse to occur just one day earlier/later on only a few trades), then I suspect that suggests the dependence is dangerously high. Hence I would check peformance using values between (say) ±2 for each MA, and see to what extent the results differ.

just tried 15/19, good for 19,500

David
Avg Pips/Trade: 31

i never realized what a trendy pair the USDJPY has been until now. (i am focused on the EURUSD mostly)..... i cracked open the chart and was pleasantly suprised with the continuous volatility (which is key)
  • #39
  • Jul 13, 2008 6:48pm Jul 13, 2008 6:48pm
Quoting tdion
Avg Pips/Trade: 31

i never realized what a trendy pair the USDJPY has been until now. (i am focused on the EURUSD mostly)..... i cracked open the chart and was pleasantly suprised with the continuous volatility (which is key)
Tom, thanks for your reply.

Agreed, the MA crossover will be robust (i.e. in less need of precise optimization) if the price movement is "trendy" enough, and the cycles sufficiently consistent in their character. Increasing the MA values to a certain point will then cause them to become "out of phase" with the price cycles, until they start to pick up the next cycle (a bit like tuning in a TV set ).

So, regarding future profitability, the key question is: to what level of consistency will the cycle-based character continue to reproduce itself?

A couple of observations, for whatever they're worth:

1. The balance between the nunmber of long and short trades, and also the win rate for each, is reassuringly consistent.

2. The slope of the equity curve was noticeably higher during the earlier years.

"Automagically" - looks like you've coined a colorfully self-descriptive new word there.

David
  • #40
  • Jul 13, 2008 9:23pm Jul 13, 2008 9:23pm
It's hard to believe at times, but the information is right out there for me to look at.

I'll code up my own set of EAs and we'll see how powerful this system really can be. I for one appreciate the simplicity I see here.
0 traders viewing now