Try My EA! Need Group Feedback...

Isn't it lunch time where you are?

It's got to be happy hour somewhere!

http://www.kitco.com/images/live/nyg...07787682455684

~dgodfrey I was looking at your V1 explorer and it seems it hasn't had a first cycle, is this correct? Or did the first round net a loss?

It hasn't cycled yet. I was trying it on 5% target to see if the MinGridGain function worked, and it missed the mark a little ($-60) so it's still on the first cycle.

Slow night on the thread, guess most of us are in drawdown while the market is back at the 1210-ish level.

Anyway, here's my $10 grid image and some thoughts on maybe how to handle this particular version of the program.

Took most of this post down. It's tedious and hard to follow, so I'll try to put it up in a more succinct form later today.

I like your idea. So, you're saying close all un-hedged, positive $$$ trades every $10? And close enough hedged trades to net a profit, leaving open the other hedged trades?

Your idea may be a good compromise to try to take some trades off the market at profit, while setting up the next move to have some meaning, instead of just a retrace on already opened trades.

Yeah, every trip $10 away nets some positive trades and usually a few hedges. Use whatever's positive to pay for as many of the hedges as you can with the profit, or all of them if that's the case. Don't realize a loss, just reset the hedge at the next level and let it go to the next $10, netting another batch of positive trades. That way you take all the profits off the table when they are still profits and before they become hedges themselves on a reversal in the market.

Just trying to see how we can keep the number of hedges down to keep from eating up the accounts. I'll post the details in a better fashion this evening.

Here's another enthusiast of the idea. Who could possibly complain to work with a bot that has intelligent, proactive and preventive MM built in!

What you also may want to watch out for is if it starts looking like just a standard grid, it won't work. A standard grid being, for example, buy and sell orders every $10. When price goes straight for $80, it loses huge. I don't think that would be a problem if you're just trying to net a profit, but leaving orders open if profit cannot be obtained.

I think the difference here is the reset, there are buy stops and sell stops always available on reset at each $10 level. A straight line move would only be adding trades that favored that direction, whatever the direction was. You wouldn't be adding contra trades except in volatility around the $10 levels, which would be hedged off immediately once the move continued, and hopefully resolved at each $10 level as it continued.

Two observations for the $10 grid:

- draws the vertical line on the first cycle only; no lines for subsequent cycles;
- when closing a cycle, it closes open trades first, which is just as it should be, but then closes the pendings outside in, not inside out.

.....

I think I've got those issues fixed in the next version already. I'd like to wait on getting this next change in place and release it all at one time if possible.

A Note: I'll release it under a different version. But if you rename it to the old version and import the language file into metaeditor, you should be able to compile it on the fly and not effect your current grid. I recompile a lot while writing code to try out and it doesn't reset things. It will overwrite the old version, though. You'll just have the one renamed one on your program at that point.

Sounds perfect, many thanks!

So, to recap, here's the idea once again: To reset the grid every $10, closing all unhedged open orders (all will be at a profit), use the money from the gain to pay down your hedge debt, or pay it off completely. Don't spend more money on hedge debt than the profit from closing the unhedged orders netted you. Start fresh with a new grid and hope for a clean run and no hedges, which pays off handsomely. Here's a worksheet excerpt to show what's possible. There are choices to make.


These scenarios are for GridSteps of 30, 50, and 75 for comparison. The Gain No Hedging column is what a clean run between 2 $10 bands would net, with some loss possible due to spread. As you can see, grid step 30 pays off at over 13% assuming 0.1 lots, and that's because there are 27 steps along the way, so you're carrying 2.7 lots assuming 0.1 each. Higher settings pay less because there are less trades paying out.

For when the volatility around the $10 line ranges out to $5 on either side of the line before proceeding, the $1 Buffer vs the $3 Buffer is this - no trades are set closer than $1 away from the $10 line, or $3 away in the $3 case. As you can see, the returns diminish with a $3 buffer, but the payoff is in the net when you do have hedging - the $3 buffer will always carry less hedge loss - there are always less trades to hit in volatility.

This also assumes that no buystop/sellstop trades will be placed within a $1 band around the target $10 line. There's no point in picking up trades that close to the target.

The two main columns to look at are the Gain No Hedging and the Net Profit/Hedges. The last assumes that the market moved just shy of $5 against you during the step between $10 levels, and hedges were picked up. It shows the money remainder you would keep after paying off all you could, how many of the hedges you picked up will remain, and what their net loss is for each one.

Just to be clear, the loss mentioned for hedges in the last column is for each hedge pair remaining.

So what if you had a $5 buffer? Well you could still pick up some hedges, there's nothing magic about $5 away. But lets say you didn't because your buffer was so large.


Not bad, you still make over 1% on the clean runs, and any run with less than $5 volatility will be clean.

Well which ones are more frequent? Here's a chart from the last couple of weeks or so.



There were 18 crossings with 5 clean runs, meaning no volatility ranging more than $1 out from the relevant $10 line.

There were 9 that had less than $5 volatility around the price band and would yield various amounts of hedges. The $3 buffer might cover some of these to yield no hedging. Many would have covered the hedge loss completely and still been profitable. (It's only when you get out near $5 volatility that the move will leave some hedges left over).

And 4 that went heavy, more than $5 in volatility and a lot of hedging. Ugly stuff. They're not all maxed out, but the first one looks pretty rough.

What is the max DD due to hedge loss from a complete hedge between the bands? (IE when it goes just over $9 to one side and doesn't trigger the next $10 level, then reverses to the level on the other side) If you always trigger on the $10 bands, this is the maximum DD possible in a single run, and there are no profitable trades to offset the losses. The worst case scenario not counting slippage. Naturally all this hedge load will be carried on the next ride between the $10 lines.

30 GridStep would be -26.5%, $2646 DD on a 10k account at 0.1 lots.
50 GridStep would be -17% DD
75 GridStep would be -10.5% DD

You can mitigate the risk of this considerably with the Sensitivity setting, making it high enough so that you gather in the near misses anyway. I'll code it so that the grid won't build over top of the sensitivity window.

Now I'm not suggesting that these GridStep settings are in any way the best, they are just the numbers we've talked about.

I'm going to start on the code for this, I'll have to figure out the best way to selectively close hedge pairs. (BTW there aren't any two opposing trades paired together in reality, this is just a fiction I use to make the math easier). One thing that worries me is all the trade overhead coming at once around that $10 line. I'm hoping that it doesn't get backed up too much trying to get all those orders out. There won't be as many though, with only the next $10 to cover on either side.

Hope this helps visualize what we're talking about.

@dgodfrey, I saw your Steady Gold BET V2.1 demo, that is based on ur BET v2.1, am I right?

Just wondering, FinFX in your Trade Explorer, suppose to be standard spread which is avg 38. So, is the low spread thing is still applies now? or now even we can actually use this with standard spread?

It's 2.11 or 2.1, they're the same except for a automatic lot size correction pretty much. And some people wanted fixed lots, that's in 2.11 as well.

The spread is a number you can select in the inputs, just put in your broker's average and it will work fine.

FinFX is pretty liberal about saying the fills are on the money, perhaps too much so. FXCM on the other hand gives really bad fills, so the truth lies somewhere in between I suspect. Let me know how your broker does on that if you decide to run it.

Good luck!

Looks like we're tee'd up for NFP Friday. Should be interesting to see what comes next.

My broker gives higher spread avg actually, abt 49. Not sure whether it is ok for Steady Gold. But so far, I've applied it with the normal Steady Gold version (no BET, no hedging, this is the version I downloaded from the first page). It was OK actually, with modified inputs by making the grid more tight.
33% for 21 days. But at this times, it will be just like flat.

I guess I need to wait and see, interested to wait for NFP....

Yeah, v2.11 doesn't really mess around much with how the system works, it's more about getting it stable so when you switch timeframes, turn off the computer, or do anything to cause it to re-initialize it doesn't crash the grid. And it adds BE and Target hypothetical lines to the chart.

You really should be ok about the spread, I can't think why that wouldn't work. The figure you mention is the same as FXCM and it runs fine on that. v2.11 has a 50 cent spread by default. With FXCM, it's just that they have so much slippage as to make a grid meaningless.