POOLER/SAN Stunning generosity

[dlewisfl]

1. How reliable is the EA: if the computer shuts down will it lose the buy/sell grid?
2. What happens when the grid is bypassed from a weekend gap?
3. During all the period I backtested it, the highest lot trade it opened was .8. Will it open higher lots if necessary? I think I figured out that it starts over with a .1 lot after it opens .1-.8 lots. Does the TP keep averaging with this, from the trade I saw it didn't look like it did, which may be a problem if that is the case.

He stores info in Empty4 Global Variables (which are used to save variable values in case of shut down/reload, among other things). He only references one of them in the code so the others are just out there doing nothing (he has code commented out that references the others). He's constantly reading through the open order pool and account history and counting open positions so it should probably work on reload (not 100% sure as I casually glanced through the code).

I saw trades hit the max #10 lot size. I ran tick data through about 15 months and there were 3 periods of large DD on EURUSD (relatively speaking - max trades were open) but they all recovered. I'll be setting up a longer test later today.

[bazze]

bm.gif

What does this message means?

[dlewisfl]

What does this message means?

Nothing. You need to remove the Print statement as described on previous page of this thread.

[Pooler]

This is the most ingenious martingale/grid type strategy I've seen (its better than Gertje+), thanks for sharing it. I backtested it on the 5M chart on the Euro from 1/1/2005-5/31/2012 and it survives! I also watched how it trades for a year or so visually: what makes it successful is that it moves the TP accordingly. I used a 50:1 criminal with starting 10k at first trading the default lots starting with .1 and this got margin called in 2008, so I switched to 20k and this survives.

Thanks. I figured to be safe I start with .01 lots per $3000

20k grew to 51,298k, which is a bit over +150% in 7.5 years = 20%/year without money management using a 50:1 criminal. So all it needs is money management: using a 50:1 criminal increasing the lot size by .01 lot per 2k works, using a 100:1 increasing it by .02 per 2k (or .01 lot per 1k) would work. This is the only feature that I think should be added first to see it's real gains.

Did you use the settings I posted earlier for the test? For EurUsd I find that starting with .01 lots and increasing with .01 each time except level 9 is .1 and level 10 is .12, aiming for 60% TP on each level and using the following OR increases with each level
1 2 3 4 5 6 7 8 9 10
25 25 25 37 37 50 50 50 75 75
I would like to see how whether the % profit is any better

By comparison during the same period:
Black Belt EA grew from 100k to 413,872k, a 400% gain, the best I know.
Dlewis's TrendOSOB grew from 100k to 160,392 using 2 delay, 16 TP, 70 SL, a 60% gain.

However, due to their money management both made very big gains. Unfortunately with a martingale/grid, money management works much more slowly because one only increases lot size by say .01 lot per $1k gained.

I don't see the usefulness in trading more than one pair, using the one with the best spread that backtests the best would seem the way to go. Two pairs may work though to avoid each's lengthy trades.

I accomodated the spreads withing the starting OR for each pair.

My questions are:

1. How reliable is the EA: if the computer shuts down will it lose the buy/sell grid?
It doesn't lose what is there but won't open any new levels if the price moves while computer is off
2. What happens when the grid is bypassed from a weekend gap?
It sets the next level accordingly
3. During all the period I backtested it, the highest lot trade it opened was .8. Will it open higher lots if necessary? I think I figured out that it starts over with a .1 lot after it opens .1-.8 lots. Does the TP keep averaging with this, from the trade I saw it didn't look like it did, which may be a problem if that is the case.
You can set the lots sizes as you wish. Please check the settings spreadsheet I posted earlier. DO NOT use settings as OTB in the first post. Profitability is affected by the 3 factors at play in the EA lot size x OR x % TP will determine where TP is placed each time. If you keep lot size at .01 for all 10 levels then TP will obviously lag far futher behind price as each OR level opens.

Thanks for assisting


I'll check out if I can increase profitability at all by fooling with the lot size structure next.

[dlewisfl]

I'm seeing some occassional #130 SL/TP price errors on order submission so there are more issues with this EA. It continues to run and close orders, probably once price gets back into acceptable range beyond stop level.

[dlewisfl]

Steve:

I'm just curious about your CloseEnough() function. I've seen Empty4 blow comparisons between double type variables, I've found that using the built-in NormalizeDouble() takes care of the issue. Admittedly, I've never had a need to go beyond 6 decimal places in any double type variables I've compared. Is there still an issue at the 8th decimal position even when NormalizeDouble() is used?

Also, what is the purpose of pFactor()? Is it there just for Gold/Silver? If so, why not just add an extra condition to "MyPoint" calculation for something other than 2 - 5 digit price?

[Generaal Goudvis]

First of all many thanks for sharing...it makes the forex world better and more beautiful!

I also use a martingale EA on live money. And to me the most important thing on this kind of EA's is:
- don't be greedy.... small wins, make easy money and your stomach will benefit form it
- moneymanagement; will you survive a max drawdown? and do you have enough to restart?
So a martingale/grid is not about winning, but staying alive.

I will definitely follow this topic!

[SteveHopwood]

Steve:

I'm just curious about your CloseEnough() function. I've seen Empty4 blow comparisons between double type variables, I've found that using the built-in NormalizeDouble() takes care of the issue. Admittedly, I've never had a need to go beyond 6 decimal places in any double type variables I've compared. Is there still an issue at the 8th decimal position even when NormalizeDouble() is used?

Also, what is the purpose of pFactor()? Is it there just for Gold/Silver? If so, why not just add an extra condition to "MyPoint" calculation for something other than 2 - 5 digit price?

The explanation is in the comment in the function.

[Pooler]

GU trade still open. Has reached level 4 with pending BL 5 opened. As you can I started widening OR/grid levels from level 4. Please us attached settings for currency pairs OR sizes. Do not use settings as in Post 1

Steve, please can you replace the spreadsheet in post 1 with this one attached here

Thanks, apologies for any confusion

[dlewisfl]

The explanation is in the comment in the function.

I've never had a problem doing something like this for example:

Code: Select all

if (NormalizeDouble(MA1,Digits+1) == NormalizeDouble(MA2,Digits+1))

I've only seen the problem when NormalizeDouble() is not used. I'm just curious if someone has seen the problem when NormalizeDouble() is used.

I suspect the Pfactor has something to do with pricing of Gold/Silver as $ value of one pip on those is not same as $ value of one pip on standard forex pair.

btw, I've seen problems with Empty4 StrToTime() function. I've used it to get midnight for a given date and it's returned something other than 00:00 for the hh:mm.