Hedge and Correlation Strategy

Scott, you are absolutely right, this is the worst problem we have with this method; it is the Achilles's heel. I've seen one of the stochastics be at 90% or above for a very long time, meaning the price on that pair has just climbed and climbed, while the other stochastic has come up just a little, making the difference go down (when the difference in price is really going up, not down).

If you have a solution to this I think we are all ears; I know I am!

Looking forward to hearing from you when you can.

If you solve this question you deserve an Oscar !!!!!!.....

I agree 100% with stochastic's limitation. OB or OS conditions can still give you 100 pips of draw down after entry. Knowledge of historical correlation and de-correlation values for a given "trip" pair, in a given TF, could really help to prevent miss-timed entries. Again TF is really important here. The same "trip" pair could be correlated 75% on D1 time frame, but 95% on M30. In my experience trading D1 correlation might get 2 to 3 entries per year while trading M30 you might get 1 entry every 3/4 days. Identifying correlations and the relationships between them could really open up some new possibilities...

Ckp,

good to read your post. Switching to M30 I get only 1 trade last 3 days (excl. today) using 10 trip pairs. I wondered about the members here with their partly numerous reported trades. But mostly, I assume, this was applied to lower TFs.

But - other switchers to M30 made same experience?

josch

I guess I'm doing whatever time frame happens to look good at the moment. I got two 5-minute triples after NFP for a total of 19 pips. My wife took a 30 minute for a total of 48 pips.

Dreamliner

' Scott, you are absolutely right, this is the worst problem we have with this method; it is the Achilles's heel. I've seen one of the stochastics be at 90% or above for a very long time, meaning the price on that pair has just climbed and climbed, while the other stochastic has come up just a little... '


DL , I also think solving this problem is like finding the Key, to great Treasures...
I am usually getting better results on smaller time frames M5 .

Good to hear; the more testers the better.

I'm beginning to think like my son, just hold on and everything comes back, especially on the triples.

I think your son is right. Until Smjones could create a new weapon, maybe the only way to make this strategy profitable is to wait, wait, wait for profit. But for this we need good margin, patience and cold blood......

I haven't been to this thread since last night and there are some interesting concepts posted on here in that time. I have been experimenting with an idea that I think may have merit. I too have noticed that sometimes a pair moves to profit quickly and all is well.... but there is some luck involved in that and I hate luck in trading.

When this happens it is one of the correlated pairs moving faster than the other correlated pair. We improve our odds with the overbought/oversold stochs but probably just a little. Obviously the inverse is almost equally possible where one pair moves against us faster than the positive one. This creates the DD we all want to avoid.

Last night I entered a trade after 5 Eastern time based on the 1 hr stock being at 65%. I was experimenting to see how much DD and how long. I set the multi-trade manager EA to close it at 25 pips profit. Before I went to bed it had gotten out of sync by about 30-35 pips. This morning it was still open and still about the same amount out of sync. I made a note of the levels and let it continue. Before NFP the one was now negative about 50 pips and the other was positive about 20 pips. Still negative overall. However if I had added 1 position to the pair that hadn't moved much at the first check I would have been very close to BE. After NFP settled down I noticed that if I had taken that extra position I would have been out with my 25 pips profit.

I then took another quad this morning based on 80% on the 30M and watched it. It never really got very far out of sync. It got to where I could have taken 20+ pips of profit because one had moved positive 40 pips and the other negative by only about 20 pips. I continued to let it go and eventually it was the opposite about 20 pips in DD.

Sorry this is a long post but I am getting to a point . I still had the original trade open and noticed that the 1H stochs were down to about 4% meaning it was not going to get out of DD as that gap closed, but the 1 minute stoch was over 80 and looked like it was turning. I added a new position on both the original pairs with the idea I would watch them and try to get out at BE or small profit. I got distracted since I am at work and when I got back to my desk all 3 trade sets were closed. I then realized I had never re-set the EA to a new profit level so all 3 sets were closed for 25 pips of profit.

When I examined which trades had been in profit it was the third set that I had added to the first set (the extra position) off the 1 minute that had closed in profit. The original trade was still negative but the second position was positive enough to over ride it and get me out with profit. The other set that was opened separately was closed by the EA at roughly BE. It had no real bearing on the results.

I think this requires some testing which I will do starting next week. I am going to come up with some parameters somewhere along the line of:

If it takes profit of 25 pips with no intervention close (obvious I know)
If it gets out of sync by 25-30 pips I will add a position to the pair that hasn't moved as far as the other pair and then go for 10-15 pips
OR
I might also watch for a good re-entry point on another time frame to see if that gives me a better re-entry and if I do that I would add to both positions equally.

I realize this is much like a scaling in function but I wont do it if the pairs stay in close sync together even if they move to 90% or higher, which I have seen happen, I will only do it when they get out of sync. It may prove to provide bigger DD on some trades but I think it is worth studying. I'll post my findings after I have tried it over a few days.

Any thoughts would be welcome... it is just an idea that might help us avoid the longer running DDs. I realize it will have it's own pitfalls that may prove to be worse.

Hi guys,

All your brains are boiling by now ?

We all agree by now about the major flaw/limitation of using stochastics: OB/OS levels can carry on for long periods,
while "long" ranging/slow-market periods degrade stochastics performance by decreasing its signal/noise ratio, increasing its sensitivity to smaller price moves, giving much less return than would be for a similar "stochastic pattern" happening on fast-moving markets.

The other issue we are seing is about trading even number of lots for both pairs, and many now we feel like this is possibly not optimized, or even counter productive, because some pairs may have a much different dollar-per-pip value, and also much different price ranges, adding some more distorsion to the picture.

We are having some geometry issues here.

How do we determine a better lot size distribution over our 2 pairs ?
I don't have the answer yet, but we can explore some paths...

Finally, the third major issue we are having here is that we cannot backtest this system, at least with MT4, because MT4 won't feed 2 pairs into the Strategy-Tester (MT5 does).


Ok, here is the idea behind the new indicator that i am releasing here.

What we are doing here is merely trying to trade 2-Pairs with even lot sizes (until now), but opposite directions : buy 1*pair1 and sell 1*pair2.

We can represent that as a BASKET composed of 2 pairs:
this is identical to "buying" a BASKET of (1*Pair1 - 1*Pair2 ) ;

In order to solve our geometry issue, we must convert all pips/points into a common unit that will be independant on the pairs.
The value of 1 pip on EURUSD is different from that same 1 pip on NZDJPY, for the same lotsize.
And we even want to have different lotsizes, which makes it even more difficult to evaluate the result of price moves.

The best common unit i could find is something anyone will instantly understand and feel very strongly: DOLLARS (or EUROS, or Pounds).

So the idea is to convert convert each pair move into a Dollar differential.
That way, whatever pair and lotsize you choose, this breaks down into an amount of $$ won/lost on each pair, and we can add/subtract these directly as they are expressed in same unit.

In order to explore this more deeply, I have coded an indicator that will draw the Profit/loss line of our Basket in Account Currency.

You can select the lot size for each pair, and the direction of each trade:
- a positive lotsize means you are buying the pair
- a negative lotsize means you are selling the pair.

Examples:
- buy EURUSD & sell EURNZD, 0.1 lot each:
Pair1=EURUSD, Lot1 = 0.1
Pair2=EURNZD, Lot2 = -0.1

- buy 0.1 lot EURUSD & sell 0.05 lot EURNZD:
Pair1=EURUSD, Lot1 = 0.1
Pair2=EURNZD, Lot2 = -0.05

Now, how do we determine the profit/loss "reference point", i.e. where is you Break-Even point, which also happens to be when you enter the trades (not counting cost of spread) ?
The indicator draws a Blue Vertical line at 00:00 "today" by default, which is the BE point by default.
If you move this line to another location, the indicator will automatically adjust the profit/loss "0.00" value to that bar, so you can clearly see the resulting profit/loss over time.
You can play with that to align the BE reference line to Stoch Difference maximums and see the outcome of the trade in the account, not simply a pip difference that has no meaning:
a negative pip difference counted on the chart can very well turn out to be profitable, and you will instantly see this on the Basket line.
When the basket p/l line crosses the 0.00 line again, then you're at BE on the basket again.
If it goes below the 0.00 line, then you're in DD...


TIPS:

- Pair1 left blank will use the chart's symbol as Pair1;

- the indicator draws a label with the Pair2 name on the upper right corner of the chart. If you change the Pair name in that Label, it will automatically change the indicator's Pair2, and redraw the profit/loss line.
I will explain later the use for it... we don't need to worry about it now.
Just take it for now as a visual aid on what Pair2 you are currently using.


What is nice about this indicator is that you will be able to instantly see the trade results over time for any given lotsize/direction.

What is also quite interesting is that you can actually backtest your hedge, as it plots the p/l line for the complete history data of your chart.
Meaning that you can use it also when markets are closed, this week-end...

Finally, you can switch timeframes on the chart, it won't affect the result line, just make it more or less precise; plot it on M1 to have best precision.


So you can now play with the lotsizes, and even select negatively-correlated pairs and buy-both or sell-both to see if that would return better profits than hedging positively-correlated pairs!


Now, how do we find the better lotsizes ?
Well...

- Plan A: manual search... can be fun in the beginning, but also become very boring quite quickly... however i suggest you start with Plan A first to play with the indicator and see what impact it has compared to the Stoch Difference indicator.

- Plan B: use statistical tools to determine the best hedging ratios for our 2 given pairs.
What tools ?
"R"...
But this will be for another post...

Enjoy your week-end building your own new monster-pair that you will be trading next week !

Just don't forget to post back your findings and proposed hedges...


Sq

Could someone give me the difference on the GBPAUD/GBPJPY 15 minute?

I have 25 right now on the 15M

Thank you for these thoughts (glad to see you got good pips this morning).

I don't personally like unbalancing the overall trade by adding to only one side and not the other. That's just me. Part of my psyche says to stay hedged until the trade gets out of whack in my favor.

But if you test this then we may discover something beneficial. Let us know!

All totaled I made 2% on my account last week. That's not much, granted. But compounded it's over 100% per year, which isn't bad. I've just been taking very small position sizes.

OK this confirms my suspicion, my indicator does not work when setting stochastics timeframe to 15. It works on 1 hour, 30 minutes and 5 minutes, but not 15. I showed it over 90%. Now watch me try to get out of this small drawdown .

I´ve got now difference 24 (corr 94,2%, D1 @ 50) TF15

Closed 2 M15 trades:

EURCHF x GBPCAD (quad) manually at 32% 93+
GBPAUD x GBPCHF manually at 22% 64+

see charts

josch

... only - what a pity, this thread still is rated 4 stars.
Deserving one more - imo.

josch

Question: How many machines do you have to test out EAs? I have one (on constantly with 12g of ram) but I think I need another.

Way to go Josch, keep racking up the pips, you're doing well! That looks like 157 pips if I'm adding right. I've really enjoyed being profitable this week too!

SQ, I swear some people have probably lost more brain cells than I ever had. How you guys can come up with stuff like this is beyond me. Did you ever work on sending people into space?

Anyway, this does look like a delightful indicator. I'm going to play with it this weekend, and see what we come up with.

Thanks Squalou, and I hope this helps many people.