FPI - Fractional Product Inefficiency: The Impeccable Hedge

NeoTicker indicators

Moderator: moderators

User avatar
michal.kreslik
rank: 1000+ posts
rank: 1000+ posts
Posts: 1047
Joined: Sat May 13, 2006 2:40 am
Reputation: 36
Location: Monte Carlo, Monaco
Real name: Michal Kreslik
Gender: Male

Postby michal.kreslik » Fri Nov 23, 2007 1:39 pm

daniil wrote:Is there any way to calculate the expected returns on the fly based at FPI value?

Once the FPI ring is open, you can calculate what is the current profit/loss if you would close the FPI ring at the current set of live prices.

daniil wrote:What entry/exit levels of FPI you use? i mean 0.9998, 0.9997.... (for not exotic pairs).

It depends on:
  • total costs for the selected FPI ring - the more costs, the more deviation you need to cover those costs
  • individual settings of the automated FPI framework user that tell the framework to either take:
    • more arbitrages with a lower average profit (and possibly having to wait longer for exit or having to exit at a loss), or
    • less arbitrages with a higher average profit
daniil wrote:If the FPI value is abnormally high or low, does your framework proceed with orders?

Sure. The more extreme the FPI value, the bigger the arbitrage profit potential. But as I said earlier, the FPI framework does not initiate any orders during the time of day / week that has been set as being too risky based on expected volatility.

daniil wrote:What is the percentage of successful Triarbs of the framework you are developing?

If properly set up, the percentage of profitable FPI arbitrages can be 100%. If you risk more by setting up the framework to trade during extra volatile times, letting it open the FPI rings before the rollover, allowing it to open FPI rings at near-equilibrium values etc., this percentage obviously goes down. It depends on the risk appetite of the user. You can safely go for the above mentioned 100% if you're ok with less trades. We'll see how this fares in a prolonged live testing.

daniil wrote:Is it better to focus at 3Rings or 4Rings?

More pairs in a ring mean more execution risk, more costs, but also more arbitrage opportunities as there are more elements that can break the equilibrium. So once again, it depends on what do you prefer. Quid pro quo.

Michal

Please add www.kreslik.com to your ad blocker white list.
Thank you for your support.

Oli
rank: <50 posts
rank: <50 posts
Posts: 26
Joined: Sat Feb 17, 2007 10:50 am
Reputation: 0
Gender: None specified

Postby Oli » Sat Nov 24, 2007 4:59 pm

michal.kreslik wrote:The FPI concept is not groundbreaking. It's equally commonplace as a wheel.

Michal


Oh don't be so modest... I mean who would have thought that three currenecy pairs imply a fixed price ratio? And then that someone can tell when that ratio is being ignored and so trade a perfect arbitrage profit? I mean ... that is just genius.

And then, to code it into NeoTicker using C-Sharp... well no wonder those HFs are lining up to pay big bucks... I mean that is technology at the cutting edge man!

gustavodemelo
rank: <50 posts
rank: <50 posts
Posts: 4
Joined: Wed Nov 14, 2007 3:01 am
Reputation: 0
Gender: Male
Contact:

Postby gustavodemelo » Mon Nov 26, 2007 4:53 pm

Hi Michal, it's very interesting your article. After I read that, I decide to developper this technique to MT4 (Meta Trade 4).

First, I make a routine that combines any currency to mount a pair. I made another loop to mount the rings. I decide to work with rings using 3, 4, 5, 6, 7 and 8 pairs.

After that, I read a file that contain the spread and the swaps (buy and sell) to make an array. In this array I sum the swaps values and sort the array by sum(swaps). Well, now I know the most important rings to open orders.

All of this routines are in the INIT() session and is very fast to calculate.

In DEINIT() I have some routines that updat some files (controls).

I make FPI functions (for each ring) to calculate and decide when is the better moment to open the orders.

The advantage of this combinations is that I can use on any broker. You just need inform the pairs, spreads and swaps (buy & sell) to EA, by CSV file, and then, the EA do the rest.

In my first test, I take 14% in just 7 days. Really very well, but it could be good luck. If a take a half of this every week, it's great, because it seams 21% per month.

I have some questions about this great technique:

1 - I use the sum of swaps to open orders just if the sum is > 0. Is it correct?

2 - About FPI, the ideal is open the orders from a ring when FPI is the lowest?

3 - Does exist some relation betwin FPI and the sum of orders profit? (Same ring, of course).

This EA works on brokers that doesn't have swap, because we can configure to open and close the rings in the wright moment, causing positive profit when the FPI is high.

Regards.

User avatar
daniil
rank: 50+ posts
rank: 50+ posts
Posts: 61
Joined: Mon May 07, 2007 6:40 am
Reputation: 0
Gender: Male

Postby daniil » Tue Nov 27, 2007 5:35 am

Michal!
Sorry for disturbing again.
But do you have a success with testing FPI with the MBT/EFX at single account?
For me the problem is the deviation of FPI calculated for entry/exit and actual FPI after positions opened. (of course this deviation can go in your favour too). Even cutting off the high volatility, news, rollover periods it seems there is no way to prognose what FPI value will be at the very next tick.

and one more:
do you ignore if partial lot size actually mismatch the size it was ordered, i mean:
we sent the order to sell 10 000 EUR/USD at 1.4878, so we must send the order to sell 10 000 x 1.4878 USD/JPY.
But actually SELL 10 000 EUR/USD was filled at 1.4877, so there is the exposed position for 1 dollar.


If you are lucky with MBT/EFX it will encourage me to do the further job with my own FPI framework.

User avatar
michal.kreslik
rank: 1000+ posts
rank: 1000+ posts
Posts: 1047
Joined: Sat May 13, 2006 2:40 am
Reputation: 36
Location: Monte Carlo, Monaco
Real name: Michal Kreslik
Gender: Male

Postby michal.kreslik » Tue Dec 11, 2007 8:17 pm

gustavodemelo wrote:1 - I use the sum of swaps to open orders just if the sum is > 0. Is it correct?

If you mean swaps as rollover interests, then the sum of these swaps is always negative for a correctly hedged FPI ring. It's because the borrowing and lending interest rates always differ. If you borrow money from the bank, you'll pay higher interest rate than if you lend your money to the bank (by opening and funding an account).

gustavodemelo wrote:2 - About FPI, the ideal is open the orders from a ring when FPI is the lowest?

First of all, there are 4 ways on how to calculate the FPI. Have a look at my earlier message in this thread with the graphs. Thus, with one calculation, it can be the lowest while at the same time with another calculation it can be the highest. The point is to open the FPI ring at one extreme and close it at the opposite extreme.

Michal

Please add www.kreslik.com to your ad blocker white list.
Thank you for your support.

User avatar
michal.kreslik
rank: 1000+ posts
rank: 1000+ posts
Posts: 1047
Joined: Sat May 13, 2006 2:40 am
Reputation: 36
Location: Monte Carlo, Monaco
Real name: Michal Kreslik
Gender: Male

Postby michal.kreslik » Tue Dec 11, 2007 8:56 pm

daniil wrote:But do you have a success with testing FPI with the MBT/EFX at single account?


Yes, I'm testing my fully automated FPI framework mainly on EFX.

daniil wrote:For me the problem is the deviation of FPI calculated for entry/exit and actual FPI after positions opened. (of course this deviation can go in your favour too). Even cutting off the high volatility, news, rollover periods it seems there is no way to prognose what FPI value will be at the very next tick.


Why, sure. There is no way to predict the future :) You don't have to. It's sufficient to open the FPI ring at one FPI extreme and close it at the opposite extreme. That's it. Precisely as I said in the previous post.

daniil wrote:and one more:
do you ignore if partial lot size actually mismatch the size it was ordered


Now this is a serious issue. There are basically two evil things that might happen during the execution:
  • the particular pair's position only gets filled partially (or not at all)
  • it does get filled at a different price
How do you cope with that? My automated framework attempts to fill the entire order a predetermined number of times and if it does not succeed, it closes the entire FPI ring regardless of the current FPI value. If you are having a really bad day, your closing order might not get filled properly neither. In such a dismal case the framework is issuing the closing orders until all get filled properly. Having a bunch of non-hedged pairs is not an arbitrage anymore, but plain trading. FPI is not a trading system, but an arbitrage, "risk-less" profit generating one :)

If it does get filled at a different price, it's a better case since if you are fully hedged, you're not exposed to any price movement-related risk. In this case you can sit and wait up until you get an opposite extreme FPI value and close the ring then.

daniil wrote:If you are lucky with MBT/EFX it will encourage me to do the further job with my own FPI framework.


There is no place for a luck in an automated trading systems business. You either do your homework meticulously or you are a gambler.

Michal

Annu
rank: <50 posts
rank: <50 posts
Posts: 18
Joined: Mon Mar 19, 2007 12:25 pm
Reputation: 0
Gender: None specified

Postby Annu » Wed Dec 12, 2007 3:22 pm

Michal: just out of curiosity, what is your approximate average holding duration of such a FPI-position? (seconds, minutes, hours)?

Regards,
A.

davidf
rank: <50 posts
rank: <50 posts
Posts: 24
Joined: Fri May 19, 2006 6:10 pm
Reputation: 0
Gender: Male
Contact:

HI

Postby davidf » Thu Dec 13, 2007 8:51 pm

Are you able determine, which pair is out of balance form ring?

Example: eur/usd, eur/chf, chf/usd, = 1. Which of them is ot of balanece?

DavidF

SetFreeByTruth
rank: <50 posts
rank: <50 posts
Posts: 4
Joined: Thu Nov 16, 2006 6:57 pm
Reputation: 0
Gender: None specified

Re: HI

Postby SetFreeByTruth » Thu Dec 13, 2007 8:57 pm

davidf wrote:Are you able determine, which pair is out of balance form ring?

Example: eur/usd, eur/chf, chf/usd, = 1. Which of them is ot of balanece?


All we're interested in is that the ring, as a whole, is out of balance. A pair by itself cannot be said to be "out of balance."

Hope that helps!

User avatar
daniil
rank: 50+ posts
rank: 50+ posts
Posts: 61
Joined: Mon May 07, 2007 6:40 am
Reputation: 0
Gender: Male

Postby daniil » Mon Dec 17, 2007 3:13 am

Now this is a serious issue. There are basically two evil things that might happen during the execution:
the particular pair's position only gets filled partially (or not at all)
it does get filled at a different price


Michal! it seems you use Limits to fill, aren't you?
do you use IOC type?

do you use DEMO for testing?
if so, i suppose the Limits behaviour at REAL is much better than on DEMO with MBT/EFX/

Please add www.kreslik.com to your ad blocker white list.
Thank you for your support.


Return to “NeoTicker indicators”