See you tomorrrrrrrrrrrrrrrrow!!!!!!!!!!!!!!!!!
(an old English saying from the 60's)
- #982
- Oct 20, 2007 5:07pm Oct 20, 2007 5:07pm
- Joined Jun 2007 | Status: Unconventionable Conventionist | 2,768 Posts
Nite Nite Nat
See you tomorrrrrrrrrrrrrrrrow!!!!!!!!!!!!!!!!!
OK... we will get to this tomorrow also. Guess there is a good portion of risk still in this. But we may iron that out.
Ciao Bella!
Gnat
Wednesday I got in about 5 minutes before 17:00 since I am not "synchonized" with my broker time (I used my puter's time and it was lagging). And the price did go down a bit before the swap was hit. But at exactly 17:00 I saw the swap appear and had I hit CLOSE TRADE I would have had SWAP minus SPREAD (which is the minimum this will generate).
But I kinda froze (this was going real fast). I had to wait "some" minutes and saw the price go down (I was in for 10 standard lots OMG those numbers got my eyes out of my eyeballs) but eventually the price came back and I exited in expected profit
A bot would be much better than me at this, I am totally sure. And instead of entering the trade FIVE MINUTES before 17:00 (which is way too risky) it would get in/out much more efficiently.
- #984
- Oct 20, 2007 5:26pm Oct 20, 2007 5:26pm
...What if the trend is down...
...What if trend is up, but market dips during the minutes we'r in...
...What if the batteries of the fire alarm in our house are empty...
BTW. We don't have to wait until 17:05, the EA can test if swap already has been generously donated..
Nighty night, sleep tight
...What if trend is up, but market dips during the minutes we'r in...
...What if the batteries of the fire alarm in our house are empty...
BTW. We don't have to wait until 17:05, the EA can test if swap already has been generously donated..
Nighty night, sleep tight
The “triple wednesday” thing is really something which could be coded
...What if the trend is down...
...What if trend is up, but market dips during the minutes we'r in...
...What if the batteries of the fire alarm in our house are empty...
BTW. We don't have to wait until 17:05, the EA can test if swap already has been generously donated..
Nighty night, sleep tight
1. If trend is down, as long as we're in this less than a minute, it shouldn't be a problem. Actually wednesday when I tried this, the trend WAS down. We just need to enter and exit quickly. And as you mentioned, I'm happy to see that if we don't need to "wait" for pip donation confirmation, that's even quicker !!!
2. This is the same conditions as 1. Just GET OUT quick and collect your (3swap - spread) profit.
3. Bang yourself on the head with a skillet. Then go buy new batteries
Unfortunately this is "unbacktestable" (good word in scrabble) as I would have backtested it for sure. However, basic logic (or basic math actually) shows that it's a profitable approach in the long term.
Maybe some days we'll get small negative results, I suppose, but isn't that part of trading anyway with ANY system?
This one is just so "close" to being infallible it's scary. And it's the simplest thing in the world.
And as for "guaranteeness" (another good scrabble word) I mean in a few seconds the price doesn't usually catapult itself mega pips at 17:00 EST.
Here's my original math 101 calcs. I took those numbers from the "negative" swaps I was seeing in my short trade at the time. Later I noticed my broker (MIG) gets to charge me that BUT gives me MUCH less ($7.73 if I recall correctly instead of $10.00) so it's also important to find a broker who doesn't rip you off swap-wise for this to work.
http://i24.tinypic.com/9u4u2q.jpg
- #986
- Edited 2:48pm Oct 21, 2007 4:15am | Edited 2:48pm
Ok, here's one test with QQEA. It sure can give you extra incomes but It's all about finding good setups. I'm not saying these are the best you can use but it's in right direction Please notice that results are in dollars not pips.
- finnfx
edit. There was something wrong in MT EA-tester, I had 90% modelling quality but still results were wrong
And I noticed also human error in settings, they were different in EA-testing and in excelsheet... Sorry everybody...
Please notice that results are in dollars not pips.- finnfx
edit. There was something wrong in MT EA-tester, I had 90% modelling quality but still results were wrong
And I noticed also human error in settings, they were different in EA-testing and in excelsheet... Sorry everybody...
If you want to win you need a Finn
(an old English saying from the 60's)
(an old English saying from the 60's)- #987
- Oct 21, 2007 6:47am Oct 21, 2007 6:47am
- Joined Jun 2007 | Status: Unconventionable Conventionist | 2,768 Posts
Ok, here's one test with QQEA. It sure can give you extra incomes but It's all about finding good setups. I'm not saying these are the best you can use but it's in right direction
- finnfx
I am doing something mindbogglingly BORING: Visual tests on the QQEA. This really is a long term project as it seems. I can only look at the test to see whether it is good or not as somehow my tester locked in the last spread at "14". Well - if it works in that context, then the setting must be a smasher!
So I am watching Tester-TV a-a-a-again!!
Ok, here's one test with QQEA. It sure can give you extra incomes but It's all about finding good setups. I'm not saying these are the best you can use but it's in right direction
- finnfx
Could you tell us what pairs are involved in your tests + exactly end date for tests. Thx.
- #989
- Oct 21, 2007 7:24am Oct 21, 2007 7:24am
Very nice results there...
I am doing something mindbogglingly BORING: Visual tests on the QQEA. This really is a long term project as it seems. I can only look at the test to see whether it is good or not as somehow my tester locked in the last spread at "14". Well - if it works in that context, then the setting must be a smasher!
So I am watching Tester-TV a-a-a-again!!
My last quotes are from Friday night on IBFX: 235.02 - 235.16 So also spread 14
It cannot come from historic data, because that only contains HLOC and volume numbers.
Well looking at these visual screens must at least be more interesting than the "Music for people beyond 85" you were watching last Saturday...
- #990
- Oct 21, 2007 7:31am Oct 21, 2007 7:31am
Good morning Qpip. Slept well?
What version and Build number dreams did you have last night...
How did the tests end last night? Any conclusions we can draw?
If it should have anything to do with v2.0 to v2.1 mods, I will take another look at the differences.
What version and Build number dreams did you have last night...
How did the tests end last night? Any conclusions we can draw?
If it should have anything to do with v2.0 to v2.1 mods, I will take another look at the differences.
- #991
- Oct 21, 2007 8:03am Oct 21, 2007 8:03am
- Joined Jun 2007 | Status: Unconventionable Conventionist | 2,768 Posts
Good question: where does the Backtester get its spread from....?
My last quotes are from Friday night on IBFX: 235.02 - 235.16 So also spread 14
It cannot come from historic data, because that only contains HLOC and volume numbers.
Well looking at these visual screens must at least be more interesting than the "Music for people beyond 85" you were watching last Saturday...
Is there a way to code a "spread for testing" with an exchangeable number into the EA? Then we could test with 8, 7 guess Pipsy only has 6 at MIG.
And "Musikantenstadl" was great once more this Saturday... I just cannot tell the Stadls apart any longer - it always seems to be the same!
Good Day to all! Coffee??
Gnat
- #993
- Oct 21, 2007 8:11am Oct 21, 2007 8:11am
- Joined Jun 2007 | Status: Unconventionable Conventionist | 2,768 Posts
- #994
- Oct 21, 2007 8:16am Oct 21, 2007 8:16am
Is there a way to code a "spread for testing" with an exchangeable number into the EA? Then we could test with 8, 7 guess Pipsy only has 6 at MIG.
If I want to "Fake" the spread I do not only need to change the code of the expert, but also all used indi's.
- #995
- Oct 21, 2007 8:20am Oct 21, 2007 8:20am
- Joined Jun 2007 | Status: Unconventionable Conventionist | 2,768 Posts
Yes and No, the Expert itself is no problem Ask, Bid Open And Close values can be calculated by using a specified spread, but the indicators also use these values.
If I want to "Fake" the spread I do not only need to change the code of the expert, but also all used indi's.
During the week catching the right spread is not so miraculous though! OK then we better stick to that.
- #996
- Oct 21, 2007 8:25am Oct 21, 2007 8:25am
- Joined Jun 2007 | Status: Unconventionable Conventionist | 2,768 Posts
Good morning Qpip. Slept well?
What version and Build number dreams did you have last night...
How did the tests end last night? Any conclusions we can draw?
If it should have anything to do with v2.0 to v2.1 mods, I will take another look at the differences.
http://i24.tinypic.com/v5eemf.jpg
Also, a note you need to take into consideration. Because I want to be able to compare apples to apples, I will not correct/change the data for CAD and NZD. But since the beginning, I have divided their results by 8 (like all other pairs) since I tested from jan to aug. However, for these 2, I only have data starting sometime in may in somes tests, starting in march in others. So basically I should be dividing the results by 4.5 or 6. But since these 2 pairs are trouble-makers with poor modelling quality on top of everything else, I'm leaving their results divided by 8. And anyway, I am just not convinced of their overall performance and only foward testing will tell.
I am, on the other hand, quite puzzled with AUD's results with build 211 in the light of its extremely poor performance last week on forward tests. I'm reposting last week's results below.
http://i23.tinypic.com/2i6fodc.jpg
Hello!
...maybe I had to have posted in this thread, anyway Here there is a backtest comparison between Years 2005-2007...
http://www.forexfactory.com/showpost...postcount=1966
http://www.forexfactory.com/showpost...postcount=1967
bye
...maybe I had to have posted in this thread, anyway Here there is a backtest comparison between Years 2005-2007...
http://www.forexfactory.com/showpost...postcount=1966
http://www.forexfactory.com/showpost...postcount=1967
bye
- #999
- Oct 21, 2007 8:35am Oct 21, 2007 8:35am
- Joined Jun 2007 | Status: Unconventionable Conventionist | 2,768 Posts
Hello!
...maybe I had to have posted in this thread, anyway Here there is a backtest comparison between Years 2005-2007...
http://www.forexfactory.com/showpost...postcount=1966
http://www.forexfactory.com/showpost...postcount=1967
bye
Hi Ghido!
Actually it is all the same whether you post on the thread or here, but on the thread more people will see it. If you then would forward it here, it is optimal!
You are right about the exit and switching problem, as the trouble is getting biggerwith each late exit, leading to a late entry on the next trade and so on!
Greetings
Gnat
- #1,000
- Oct 21, 2007 8:35am Oct 21, 2007 8:35am
As you can see from the tables below, there is quite a difference between v1 and v2 on build 210 tests
Are you able to run this on both your 210 and 211 platforms in visual mode?
Doing this you will be able to see on both platforms their spread values (e.g. Show ask+Bid line)
Maybe this will shed some light on the difference..