Would this work? | Page 2 | Forex Factory

Options

Go to
Permalinks
Print Thread

Similar Threads

Why do indicators not work/work? Why is Forex not truly random? 129 replies

Adding to profit, would this work? 0 replies

would this silly system work 6 replies

What broker would work well in this situation? 7 replies

would this strategy work? 1 reply

Attachments: Would this work?
Exit Attachments

Would this work?

  • #21
  • Jul 16, 2006 11:09pm Jul 16, 2006 11:09pm
Quoting dof
Why not buy only gbp/jpy, you will have the best interest
If you are going for the long run, it can go much highier, it's a double winner
Carry trades such as GBP/JPY have a tendency to unravel a great deal in a very short amount of time. The purpose of opening a position with a second pair is to minimize that risk while benefiting from the differential in rates. I have not yet seen a viable hedge for GBP/JPY.
  • #22
  • Jul 16, 2006 11:15pm Jul 16, 2006 11:15pm
Quoting mrmikal
I didn't read that part of the post...however equaling the pip value doesn't actually get you the correct hedge, does it? If that's the case, we're assuming that the pip movement is in synch, which I don't think is accurate. I think the value change is what is correlated.

I'm going to create a little indicator in wealth lab to see what the actual daily correlation is to see if I can't better understand how to do this.
The pip movement will not be in synch, but that is the beauty of this system. If you don't mind a little negative float, hold both positions until the difference in the P/L of the two positions is positive, then close the positions for profit and immediately reopen the postions, beautiful. It should be a small matter to create an EA that will do that for you. Oh, and currency correlations are well documented, you need only find two pairs with high negative correlations and an attractive yield. The only trouble so far I've had is finding a broker that offers fair interest rate payments but also offers fractional lot sizes.

Isufreak24 may have just given fire to the mortals.
  • #23
  • Jul 17, 2006 12:11am Jul 17, 2006 12:11am
OK...I back tested this thing and got some disturbing results.

The correlation of up days and down days may be correct, but the actual pip-movement varies quite wildly.

To do my testing:
- I bought 100 micro-lots of the GBPUSD
- I determined the number of micro lots of the USDCHF to buy by taking the daily opening price of USDCHF and multiplying it times 100 micro lots. So if the opening price was 1.3000, I would buy 130 micro lots. This gives you the roughly the same per-pip profit of $10.
- I then calculated the daily profit change.

Based on my back tests:
2002 - Max Drawdown of about $26,000, minimal updraw...ended the year near the low
2003 - Max Drawdown of about $12,000, minimal updraw...ended year down $9,078
2004 - Minimal Drawdown, Max updraw of $19,000...ended year up $4,286
2005 - Max Drawdown of $4,000, Max updraw of $6,600...ended year up $3,609

Not sure about you, but although the last couple of years results were pretty good, the first 2 years are absolutely frightening.

I'm beginning to wonder how correlations are calculated...I suppose there has to be something I'm missing...but I did take the rate info I had and compared them against both years, so unless the data synch was wrong, but I doubt it...

Anyone want to suggest something? or has anyone run a back-test with contrary results?
  • #24
  • Edited 4:49am Jul 17, 2006 4:22am | Edited 4:49am
Quoting mrmikal
OK...I back tested this thing and got some disturbing results.

The correlation of up days and down days may be correct, but the actual pip-movement varies quite wildly.

To do my testing:
- I bought 100 micro-lots of the GBPUSD
- I determined the number of micro lots of the USDCHF to buy by taking the daily opening price of USDCHF and multiplying it times 100 micro lots. So if the opening price was 1.3000, I would buy 130 micro lots. This gives you the roughly the same per-pip profit of $10.
- I then calculated the daily profit change.

Based on my back tests:
2002 - Max Drawdown of about $26,000, minimal updraw...ended the year near the low
2003 - Max Drawdown of about $12,000, minimal updraw...ended year down $9,078
2004 - Minimal Drawdown, Max updraw of $19,000...ended year up $4,286
2005 - Max Drawdown of $4,000, Max updraw of $6,600...ended year up $3,609

Not sure about you, but although the last couple of years results were pretty good, the first 2 years are absolutely frightening.

I'm beginning to wonder how correlations are calculated...I suppose there has to be something I'm missing...but I did take the rate info I had and compared them against both years, so unless the data synch was wrong, but I doubt it...

Anyone want to suggest something? or has anyone run a back-test with contrary results?
I'm not sure if backtesting this system would ever work. Negative float is to be expected, with this system you never have to take a loss. I'm not sure why you would take a loss. The only time you close a position is when you make a profit, even if it is small, and then you re-open the positions right away, the rate does not matter. The best part of this system is the acumulation of interest, were you able to add the payment of interest in your backtests?

How about this...can you backtest tick by tick, taking profit at a positive differential of 20 pips, never take a loss, and add the interest from rollover to your balance?
  • #25
  • Edited 8:21am Jul 17, 2006 8:09am | Edited 8:21am
Quoting SisterCurare
I'm not sure if backtesting this system would ever work. Negative float is to be expected, with this system you never have to take a loss. I'm not sure why you would take a loss. The only time you close a position is when you make a profit, even if it is small, and then you re-open the positions right away, the rate does not matter. The best part of this system is the acumulation of interest, were you able to add the payment of interest in your backtests?

How about this...can you backtest tick by tick, taking profit at a positive differential of 20 pips, never take a loss, and add the interest from rollover to your balance?
I did not add the interest rate roll-over, but I can tell you that from the results, it wouldn't matter. 1 lot hedged yields between $6 and $15 a day. Since I didn't maintain any equity in the back-test (I just did the math) and didn't adjust the lot size, I didn't see the need to add the $3750 or so during the test since it wouldn't make that much of a difference. The rollover would be constant.

Look, here's an easy way to determine how off I am:
- Take the opening price of the GBP/USD and USD/CHF at the beginning of the year.
- Buy 1 lot of each (or adjust it or whatever)
- Take the closing price of the GBP/USD and USD/CHF at the end of the year
- Sell both lots and adjust to determine the profit.

That algorithm won't show you the drawdown, but it will show you what you get with buy & hold. The system presented may have better results, but you're assuming that you're going to be able to sell most days with a profit and that on days that don't have a profit, you'll be able to recover, which isn't necessarily true.

I will modify the test slightly to exclude sundays since sunday movements are pointless and they only skew values.

However, I am warning you of putting too much stock in the correlations. Just because they exist doesn't mean that the pip variation or the profit variation will be exactly 0 or even close to 0. The fact that you don't "see why" doesn't negate the fact that the GBP/CHF could be (and is) moving and thus could be accounting for the losses.

The reason I'm a little wary is because if the system looks too good to be true, it probably is. Logically, of course, we want to believe that the correlations account for this, and it probably does. However we should really note that if you're expecting exactly 0 pip movement every day, then you're fooling yourself.

Now, I didn't run a tick by tick test, and I'm not sure I'm going to be able to on the entire year because it just takes too much processing power to try and process 500,000 records in WL. However, you're more than welcome to dispute the results with your own back test. I would definitely love for someone to come up with contrary results. Obviously, back-testing doesn't really mean a whole lot, but we can identify risks to the system by doing this. Let's just put it this way, if I find that a system performed poorly one year, I'm going to put more stock in that than if the system performed really well one year. If the system generally stays away from danger in all the back-tests, my confidence would be higher.
  • #26
  • Jul 17, 2006 8:46am Jul 17, 2006 8:46am
It's not really a buy and hold, and did you take into account the interest compunds as you add more lots as account equity increases? It's simple, you earn interest each day no matter what, and you hold positions through periods of negative float, until the hedge becomes profitable, which it is bound to. Unless your account is overleveraged, you will not get a margin call. I've been holding a hedged position with Oanda using $4600 out of $5000 in margin all day (at 50:1) and have not gotten a margin call, and cable has clearly outperformed swissy today. I have also hedged the pairs with fxcm by buying 2 lots swissy to 1 lot cable (using $3000 out of $5000 at 100:1) and have not gotten a margin call. FXCM is scheduled to pay interest, per lot, of $10.45 on swissy and charge $4.40, per lot, on cable. As I mentioned, I have two lots on swissy, I will do this each day, and earn interest each day and on Wednesday, rollover is triple, you do the math. In addition, I will close the position if and when P/L equals $300, if it ever does, but I have to at least cover the total spread which is about $100 bucks at 100:1. You cannot backtest this system. Is it becoming clear?
  • #27
  • Jul 17, 2006 9:19am Jul 17, 2006 9:19am
Quoting SisterCurare
It's not really a buy and hold, and did you take into account the interest compunds as you add more lots as account equity increases? It's simple, you earn interest each day no matter what, and you hold positions through periods of negative float, until the hedge becomes profitable, which it is bound to. Unless your account is overleveraged, you will not get a margin call. I've been holding a hedged position with Oanda using $4600 out of $5000 in margin all day (at 50:1) and have not gotten a margin call, and cable has clearly outperformed swissy today. I have also hedged the pairs with fxcm by buying 2 lots swissy to 1 lot cable (using $3000 out of $5000 at 100:1) and have not gotten a margin call. FXCM is scheduled to pay interest, per lot, of $10.45 on swissy and charge $4.40, per lot, on cable. As I mentioned, I have two lots on swissy, I will do this each day, and earn interest each day and on Wednesday, rollover is triple, you do the math. In addition, I will close the position if and when P/L equals $300, if it ever does, but I have to at least cover the total spread which is about $100 bucks at 100:1. You cannot backtest this system. Is it becoming clear?
I agree, that the system isn't buy and hold...but you've been holding all of a few days.

Making statements like "it's bound to be positive" isn't exactly accurate. I think there is a reasonable probability, but I don't think it's a high probability. There IS a high probability that you will have positive days, but being able to turn a negative float into a positive float is another matter.

Like I said, you're welcome to challange me on this with your back-test. However, you really need to see without the rose-colored glasses...optimistically speaking, $15 a day over 365 is $5476 you can add a lot every $4000, so you're compounding isn't exactly going to out-pace the drawdown in your account.

This isn't about taking a margin call, I didn't say that you were going to drawdown $26K in 1 day. I'm saying that the daily up and down adds up to that difference at some point. There are some days where you lose easily $500 (50 pips).

All I want is for independent confirmation from someone out there that my results are incorrect.

Like I said, I think there may be potential IF you can take a profit or break-even every day. However, it's not a certainty that you'll be able to do that every day...and there actually is evidence that historically, there has been a massive drawdown.

It's not as if I don't get the mechanics of the system, but let's just put it this way, you still have to deal with the spread on both pairs and you still have to hope that BOTH currencies move up together at some point in order to combat the times that it moves down.

Seriously, someone, please dispute the results with a back-test. I'm all for finding an edge, but not at a high risk ($26K drawdown to $5475 annually).
  • #28
  • Jul 17, 2006 9:20am Jul 17, 2006 9:20am
After a little further thought, I think you're just yankin' my chain, Mrmikal. Just to see how long I'm going to argue a point we both know is true. You really had me goin'. Well, I can take a joke as well as the next guy.

But, in case you are serious, will someone please back me up on this one?
  • #29
  • Jul 17, 2006 9:26am Jul 17, 2006 9:26am
I dont think Mrmikal is joking...
  • #30
  • Jul 17, 2006 9:43am Jul 17, 2006 9:43am
Quoting twinchell
I dont think Mrmikal is joking...
Twin-baby,

You know me :-) All about kidding around and sh-t.

Sis,

Honestly, I would love nothing more than to be wrong on this. However, I've been burnt more times to know that if a system looks too good to be true, it's time to try and punch holes in it.

However, let's say for the sake of argument, that someone does buy and hold. If correlations were -100%, we'd expect our hedge profit to end at 0, correct at the end of a year. However, if you compare just the year-end prices for the last few years, you will see that this isn't the case and that we do indeed end up with a few negative years (just holding for a year). Unfortunately, for the system, losses are greater than (by a LARGE margin) than the wins. Now, by reason of induction, holding for a single day has to yield a negative expactancy too for the years where we end up negative.

Now, it is entirely possible that we can "even" out the drawdown by doing what you say, but you also have to be willing to accept the fact that just because you lose 100 pips one day you won't necessarily gain 100 pips the next day. In fact, in the years where we turn negative, you have to expect that the number of possible positive pips does not equal the number of negative pips, which means that it's impossible to turn a negative expectancy into a positive one. Tick by tick may be another story...Furthemore, if that's true...you have to be willing to accept that there will be days where the losses will add up to more than the gains.

Here is how I will modify the test. On days where it is positive, I will take profit. If the day is negative, I will WAIT until the profit/loss becomes positive. I will not be surprised, though if this only worsens the results since we be terminating profits early and letting losers run.

Sis, nothing against you...I just don't want to see anyone get any false notions about this. It's the same thing with the hedgehog strategy that was touted earlier. It was a great idea in theory, but there was a hole to punch in that theory and in practice, it didn't work out as well after a string of losses. I'm not saying that it doesn't work well some of the time, but ALL of the time is another thing.

Point is, if we're trying to float a trade, we'd like to have a reasonable positive expactancy. Correlation tables suffer the same problems as back-testing...it's historically accurate but moving-forward inaccurate. This isn't a case of having a minimal drawdown...$26K...that's a big drawdown...even buy & hold (not adjust daily for the currency rate change) yields terrible results in a year. And while I agree that historical results aren't a prediction of what's to come, the fact that it HAS happened puts it squarely in the RISK column. You could very well accomplish your wildest dreams with this system. Maybe we'll be headed for some fantastic years with the system...but is it worth the risk of you blowing out your account? Just a thought. Of course, if you have a $100K account, you wouldn't blow it out...but for 5.4% APR?? dunno...sounds a little lame.
  • #31
  • Jul 17, 2006 10:31am Jul 17, 2006 10:31am
Quoting mrmikal
Twin-baby,

You know me :-) All about kidding around and sh-t.


In all seriousness, the best way to go about determining if a strategy new to yourself is worth spending time on is to try to prove it wrong. Mrmikal has both spent time on it and proved it wrong. Im not saying it isnt profitable, it might be; however, Mrmikal is an extremely smart person whos test I would back anytime. If he is wrong, Id like to see it as much as he would. It would be great if he was wrong, because the strategy sounds good in theory.
  • #32
  • Jul 17, 2006 10:34am Jul 17, 2006 10:34am
Quoting isufreak24


The way I've been calculating my positions to maximize this strategy on
with 50:1 leverage. I divide whatever amount I have on my account by 0.0588.
That gives me the lot number to buy on EUR/USD pair with still enough margin to
allow for deviation. So for $5,000, it's $85,034. Then I multiply 85,034 by
the
current USD/CHF rate, so in this case 85,034*1.2899, that equals to 109,685.
That way a pip on each pair equals to the same amount of money. And I open my
orders. "

sounds interesting.
Hi isufreak,

Can you elaborate on this last part,I don't unerstand where did you get the number 0.0588 and what does the number 109,685 means and how do i use it when i open oreders.

It realy does sound interesting,i saw some of those strategies before but i did't realy look into them.

Thanks
  • #33
  • Jul 17, 2006 10:36am Jul 17, 2006 10:36am
Well, I have to admitt Mrmikal knows his stuff. If he says it is so, it is so...

Scott


Quoting twinchell


In all seriousness, the best way to go about determining if a strategy new to yourself is worth spending time on is to try to prove it wrong. Mrmikal has both spent time on it and proved it wrong. Im not saying it isnt profitable, it might be; however, Mrmikal is an extremely smart person whos test I would back anytime. If he is wrong, Id like to see it as much as he would. It would be great if he was wrong, because the strategy sounds good in theory.
  • #34
  • Jul 17, 2006 10:38am Jul 17, 2006 10:38am
Quoting smjones
Well, I have to admitt Mrmikal knows his stuff. If he says it is so, it is so...

Scott
Guys thanks for the vote of confidence...but honestly...I WANT to be proven wrong here!! You're not helping :
  • #35
  • Jul 17, 2006 10:40am Jul 17, 2006 10:40am
Well, now you are just going to have to look closer aren't you?

(evil laugh inserted here)

Scott

Ps, I have seen you doggedly get to the truth work elsewhere and you know what I am talking about... So.. I will try, but you probably are correct...

Scott

Quoting mrmikal
Guys thanks for the vote of confidence...but honestly...I WANT to be proven wrong here!! You're not helping :
  • #36
  • Jul 17, 2006 10:43am Jul 17, 2006 10:43am
Quoting happy camper
Hi isufreak,

Can you elaborate on this last part,I don't unerstand where did you get the number 0.0588 and what does the number 109,685 means and how do i use it when i open oreders.

It realy does sound interesting,i saw some of those strategies before but i did't realy look into them.

Thanks
Happy,

What he means is he is normalizing the pip value. Thus for every pip movement in the USDCHF, it will move gain/lose the same value as the EURUSD.

Of course, he's buying $5000 worth of the euro which is different than saying he's buying 5000 euros.

To simplify it...when you open a hedge, buy x # of lots of the EUR/USD, then, to determine the # of lots you need to buy of the USD/CHF, simply multiply the current rate of the USD/CHF by the number of lots in the EUR/USD you bought. That "should" give you about the same pip value.

Here's a little math for you:

BUY 100K EUR/USD @ 1.2500 = 125,000 USD
SELL 100K EUR/USD @ 1.2501 = 125,010 USD
PROFIT on sale = $10 USD

LOTS of USD/CHF to buy @ 1.2200 = 100K * 1.2200 = 122K
BUY 122K USD/CHF @ 1.2200 = 148,840 CHF
SELL 122K USD/CHF @ 1.2201 = 148,852.2 CHF
PROFT on sale = 12.2 CHF / 1.2201 (rate) = $9.999180395 ~ $10 USD

get it?
  • #37
  • Jul 17, 2006 10:43am Jul 17, 2006 10:43am
Quoting smjones
Well, now you are just going to have to look closer aren't you?

(evil laugh inserted here)

Scott
Im with Scott here, pip samurais are only beaten by other samurais. Ok, sorry, this has been a test of the emergency broadcast system, now back to our originally scheduled thread...
  • #38
  • Jul 17, 2006 1:31pm Jul 17, 2006 1:31pm
Quoting mrmikal
Guys thanks for the vote of confidence...but honestly...I WANT to be proven wrong here!! You're not helping :
I think your interest-story is correct. But a high correlation between EURUSD and USDCHF doesn't mean that deviations will go back, like you wrote in the first post. That just depends how the EURCHF changes. Or did you mean something else?
  • #39
  • Jul 17, 2006 2:30pm Jul 17, 2006 2:30pm
Quoting mrmikal
Happy,

What he means is he is normalizing the pip value. Thus for every pip movement in the USDCHF, it will move gain/lose the same value as the EURUSD.

Of course, he's buying $5000 worth of the euro which is different than saying he's buying 5000 euros.

To simplify it...when you open a hedge, buy x # of lots of the EUR/USD, then, to determine the # of lots you need to buy of the USD/CHF, simply multiply the current rate of the USD/CHF by the number of lots in the EUR/USD you bought. That "should" give you about the same pip value.

Here's a little math for you:

BUY 100K EUR/USD @ 1.2500 = 125,000 USD
SELL 100K EUR/USD @ 1.2501 = 125,010 USD
PROFIT on sale = $10 USD

LOTS of USD/CHF to buy @ 1.2200 = 100K * 1.2200 = 122K
BUY 122K USD/CHF @ 1.2200 = 148,840 CHF
SELL 122K USD/CHF @ 1.2201 = 148,852.2 CHF
PROFT on sale = 12.2 CHF / 1.2201 (rate) = $9.999180395 ~ $10 USD

get it?
Hi,

Thank you for clearing things up for me,i now get it.
  • #40
  • Jul 17, 2006 3:25pm Jul 17, 2006 3:25pm
Quoting JosTheelen
I think your interest-story is correct. But a high correlation between EURUSD and USDCHF doesn't mean that deviations will go back, like you wrote in the first post. That just depends how the EURCHF changes. Or did you mean something else?
Nope that's exactly what I meant.
0 traders viewing now