FPI - Fractional Product Inefficiency: The Impeccable Hedge

[michal.kreslik]

However whilst all the theory and fancy graphs are great, not one person on this forum has ever shown or proved that this works in reality and works on a regular basis.

I can't speak for others, but personally, I am posting here to get feedback and discuss various concepts so that I am able to trade better for me and for my clients. There's no motive for me to prove anything here.

Several years back, I bought a new BMW 745i to "prove" that I am successful. That was a silly thing to do and I later went for a less obtrusive car to not stir "public opinion".

I have made dozens of demo trades using different brokers, and i have not once made a profit using this system!

Obviously, you don't have an automated system for trading FPI. You can't profit trading it by hand with average brokers.

Michal

[Nicholishen]

@michal
yes, thanks for clarifying. you are the best. unfortunately, your calculus is still wrong because 29.6.2006 was a date with extreme low liquidity, and the spreads were extremely high. At that time, as you can check, on IBFX the EURUSD spread was 12pips (!!!), on FXCM the spread of EURUSD was 6 pips, on Oanda the same spread was 4 pips, and on SXB it was 4 pips.

Dear Bitcy,

a professor of geography does not feel obliged regularly to confute those who believe that the earth is flat. But I feel obliged to confute you since you are spreading this "earth is flat" notion in my forum.

First of all, saying that any arbitrary day in Forex might be a day with "extreme low liquidity" (I guess you meant "extremely low") is just plain ridiculous. A market with daily turnover of $2 trillion might theoretically lack liquidity if you want to go for a $800 billion position.

Secondly, I seriously doubt about Oanda having a continuous spread of 4 pips in EURUSD. Oanda publishes their weekly spread history, but there's no way to know what the spread was 4 months ago (unless you were sampling Oanda data).

I can't comment on the other mentioned brokers (huh, I don't even consider FXCM a broker), but concerning Oanda, I am pretty sure you are simply lying. The continuous 4 pips spread in EURUSD is simply a science-fiction.

Anyway, there are zillions of other examples of FPI trades. You just simply need to do your homework and do the calc for yourself. No one can do that for you. For example, May/5/2006 was an "NFP" day with lots of volume (not that you need those trillions of dollars volume, but let's assume you do). If you entered the FPI trade at 14:10 and exited at 14:35, it would look like this:

And more than that, you REVERSED the ring now, ha ha ha ha, that is not the same ring as in initial example.

It's the same ring taken in the correct direction. Maybe you didn't get that the rings must be opened in a particular direction. The "direction" issue was not clear at the beginning.

At the moment of opening the ring you DON'T know which direction is good for the ring.

Yes, I do. And I am currently programming FPI automation for several clients. Do you think the professional hedge fund would pay me to code FPI just for the heck of it?

Oh, ok, man, if you are saying that you are working to find a strategy to choose the best direction to play a ring, than all my respects !! (seriously, no irony) and good luck. I would also pay money for that strategy if is for sale!!.

Finally a good idea, Bitcy. PM me then and we might arrange for the deal.

I appreciate anyhow that even I used bad words against you, you were still conciliative and did not jump to my neck . Thanks.

Only people with whom I have a great deal of respect for could possibly offend me. So you may sleep well.

I have built an indicator that shows you the profit you can make on perfect hedging.

I could not find any profit combination, all are losing money, no matter where I put the line.

I can't comment on your MT indicator, I'm coding in C# and possibly in EasyLanguage. Anyway, it's apparent you made a mistake somewhere.

One final note: seems like you've got plenty of time on your hands. You should consider utilizing this extra time more constructively. You want:

http://www.shockwave.com/online.jsp

Michal

Michal,

Please allow me to play devil's advocate here and try to understand exactly what I am seeing in your above example. It would appear that you are calculating the fpi based on the historic close prices, adjusted with fixed spreads. I am assuming that you used the spreads as per your spreadsheet to calculate the appropriate Bid-Ask, correct? Assuming that spreads never widened this would be a perfect representation of actual results. However, during times such as NFP, high volume results in low liquidity thus widening spreads. Without actual tick data, how can you know that this spike was not a result of the spread widening? If you have downloaded the tick data that i posted from 11-8, you'll notice the spike around tick# 3700 or so (not exactly sure without looking). Incidentally it was the only significant spike for the day which happened to be a result of the spread widening by 2pips on one of the pairs.

When opening on the FPI calculated @ BAA, we have to close at BAA. We know that a widened spread will affect the FPI to give us a spike. Even though there is a spike that indicates ideal conditions, it may be in fact an error due to the fact that we actually cannot close in profitability.

Let's also look into the scenario that the spreads actually did widen creating the inefficiency. Instead of closing in profit it would be a loss, correct? I am not trying to be an *censored - swear word* here; I am just trying to test the efficacy of this theory. What do you think?

[michal.kreslik]

I am assuming that you used the spreads as per your spreadsheet to calculate the appropriate Bid-Ask, correct? Assuming that spreads never widened this would be a perfect representation of actual results. However, during times such as NFP, high volume results in low liquidity thus widening spreads.

Sure, but the spread certainly doesn't stay widened 5 minutes after the news comes out.

The true Bid/Ask history would naturally be the best data set for testing.

Michal

[michal.kreslik]

Another example, the same day (May/5/2006).

This time, the FPI goes from the high value to the low value. Thus, the ring is taken in the opposite direction:





Michal

[bitcy]

@makosgu
Ok, I got the point about throwing the dice. It seems to be very good reason for me, especially japonese are well known for their perseverance. Thanks. I will "build up muscle memory" to play the dice in MY WAY and I will not bother you anymore, and you can build it in your way. We play different areas, as you say, I am a guy too small for you, too small for this big ideas here, and "aquilae not capit muscam".

@Luke
My "darling" luke, I did not "went so long" just to prove that this strategy is wrong. You should be ehough clever to realize that spending my time on it means that I really wanted to give a chance to FPI. Unfortunately, the results I found let no doubt for me: it will not work. Ok, I do not want to upset makosgu, because I like his explanation with the dice, and let's say just "this strategy will not work for me". Maybe it will work for you, if you trading style is different, or it will work for michal if he "train his muscle's memory" to make it working. With patience and perseverance people can do a lot of things. But I do not have enough fast execution time, enough money, enough patience, or I am not enough clever for this strategy, so, for sure it will not work for me. Good luck using it.

@michal
sorry for bothering you man, this is my last post here, just to make a small corection of the profit line in my indicator, for the people who want to analize it and test it. I hope you will allow me to do that, before blocking my IP...

Just because I don't want to let behind me one tool with programming bugs. So here is last version of my trendline. I found a bug and have to change the "Point" variable with "MarketInfo(Symbol(),Point)" in some places. This bug affected the situation when all 3 pairs are croses, i.e. no USD involved, in this case the profit in pips was wrong calculated. The bug did not affect the situation when MainPair is EURUSD, and anything from what I said before were corect as I described. Who did already tests got the corect results: that are the REAL profits what you see in the comments, and all are negative...

Good luck in trading your strategies, whatever they are, FPI or not FPI, important is to make money. Michal is right, I have better things to do.
Making money, yaaaaaa....

[michal.kreslik]

I am a guy too small for you, too small for this big ideas here

You don't have to be ashamed of being a small guy, Bitcy! You're not the only one - as your Japanese sources say, "The average height of Japanese people is about ten centimeters shorter than that of Westerners."

and "aquilae not capit muscam"

Ok, seems like mangling foreign languages is your primary hobby. Correctly spelled in latin this is "Aquila non capit muscas" - The eagle doesn't catch flies. Let's hope you can speak Japanese well at least...

My "darling" luke, I did not "went so long" just to prove that this strategy is wrong.

You're right. You didn't prove this concept is wrong, Bitcy.

But I do not have enough fast execution time, enough money, enough patience, or I am not enough clever for this strategy

You should consider knitting instead of trading then.

Good luck using it.

Thanks. I wish you the same.

this is my last post here

We appreciate that. Please keep your word.

I hope you will allow me to do that, before blocking my IP...

Aquila non capit muscas. Why should I be bothering with a fly, blocking its IP? Moreover, you said this was your last post.

I could have easily banned you with a touch of a button, but it was more constructive and more fun to prove you're wrong.

Michal is right, I have better things to do.

Well stated. The same here.

Michal

[Harm]

I've been playing around with some true bid/ask data which was very insightfull and strengthened the idea that it is not tradable with out regular bucketshop brokers. As has been posted before. However I'm convinced it is applicable with access to proper price feeds and fast execution.
I also have some idea's on how to automate it, which is an absolutel requirement.

So now I'm at a dilemma, invest $1200 in a trading platform (or $90/month) to automate it with a ECN broker or leave it (for the time being) and continue with another project.

[michal.kreslik]

I've been playing around with some true bid/ask data which was very insightfull and strengthened the idea that it is not tradable with out regular bucketshop brokers. As has been posted before. However I'm convinced it is applicable with access to proper price feeds and fast execution.
I also have some idea's on how to automate it, which is an absolutel requirement.

So now I'm at a dilemma, invest $1200 in a trading platform (or $90/month) to automate it with a ECN broker or leave it (for the time being) and continue with another project.

Without automation and excellent broker, FPI is not profitable. I'm now working on an automated FPI in C#/NeoTicker for several clients. One of them has (very probably) access to the direct interbank quotes.

The problem of automating FPI successfully is very complex. There are many catches that haven't been discussed here. But the reward is sweet

Michal

[TheEconomist]

Simple question : Should I optimize rings for lowest FPI or for greatest swaps? (Thought about FPI cause a few cents/day won't make me rich)

[Harm]

I'm still to see an interest positive ring, could you post an example?

Also I noticed the swaps on EFX are nothing like it should be based on interest differentail, e.g. GBPJPY positions pay swap regardless of direction, or am I mistaking? So much for my beloved carry on this baby...