Systematic trading

If you want, go ahead and describe some of the phenomenons you mention at the above post and I'll promise to dissect them for you (if my data is good enough - I don't have book data for longer periods) so you'll see where the edge lies.

When you are "Hedging" ie buying/selling the same currency what you really are doing is altering your position size, it has nothing to do with anyone looking it with contempt (except maybe those people whose broker takes double margins or even double swaps).

Hello mikkom, sorry to interrupt your talk here.

thanks for all your posts, they are really interesting.
im studying the possibilities of using GP in forex atm, so this thread comes in quite handy.

id like to ask you some questions, which, of course, you dont have to answer

 

1. how long have you been working on this specific project?
2. i guess you are seeding your initial population with more or less complicated strategies?
3. how large is your pop?

if im completly in the wrong direction, simply tell me

thanks for your time and effort

Mikkom,

I strongly believe in what Master tapereader George Douglas Taylor told us back in 1950s that all speculative markets - whether regulated or non-regulated - are engineered. Neither he nor I say it is illegitimate. I have empathy for them because I am a business person too and want to be a competitor.

I understand technical analysis well but have NO FAITH IN IT. I use charts to compile statistics . I use indicators to define trends, trading zones, tops, bottoms, and pauses by ways of graphics or statistics tables. I have used some of your indicators which fall into this category.

I put my own money online (not others) and do spot trading as a business and not as a game of probabilities. I ignore all kinds of analysis and concentrate on watching statistics charts and use my brains to give out signals. News releases are market makers' trigger points for generating volatility and the data releases are important to me for that reason. What is the content of the data releases is of not so much importance to me because I know market makers will achieve the levels they want to. The data will be interpreted to suit the market makers objectives.

I am an entrepreneur and understand the importance of planning, dedication, hard work, and exercising patience. For me most important is controlling risk and loss. If I do this successfully, I know profit will always be there. The 70% focus is on money management and risk control. Hedging, drawdown and margin planning is crucial to my money management. I can't describe it as well as a good Excel or Strategy tester user can do. I am not well versed with those two software.

I am writing the outline of a market makers business plan to let you know from which forest land (started out as an NGO founder, later became a daily newspaper journalist, then turned entrepreneur and now post-retirement functioning as a business consultant) I come from.

If I want to be a market maker this will be the outline of my business plan:


Macro view

(a) Industry: Information/finance-related gaming

(b) Product: Wealth distribution with a huge profit bias for myself.

(c) Customers: Get-rich-quick seekers (speculators) and customers for real-money exchange

(d) Nature of business: OTC , self-regulated rules. Heavily dependent on Governments' support. If governments are not satisfied they will make it a regulated market or give a right to fix daily exchange rates to IMF or some new world body.

(e) Competitors: Because of the huge profit potential I will attract various kind of competitors - large (sovereign funds, central banks and big investment bankers - I will have tacit understanding with them that their interests will be protected), medium (hedge funds and medium-sized investment bankers), small (commercial banks and big brokers) and tiny (retail professional traders ).

(f) Advantages: Huge - OTC nature of the business gives me big advantages over my competitors. My computer servers can tell me about my customers' positions, volumes , the clustered levels, trading behaviour and patterns while my customers will always remain in the dark about this.

Greed is the driving force in today's capitalistic world. Do I have to say that there will be no shortage of worshippers at a temple of greed?

(g) Disadvantages: If I play by rules and don't upset the government then everything is OK. I can't keep market continuously volatile for hours, days, weeks, months, years in a row or for a longer period. If I choose I can do it but it will be like killing the golden goose. Government will not keep quiet if I upset them.

I can tinker with High-low for a short period. I have to make sure I must keep open-close levels balanced to keep the government happy. At times I get orders from my big customers , Central Banks and Sovereign Wealth Funds, to execute their transactions. My spread income will be huge . For those transactions, I don't act as a market maker.They give me specific levels and I have to bring open-close levels to suit their needs. For you it's intervention, for me it's a brokering job.

Second disadvantage is that I must keep all mighty people happy by sharing spoils with them. That's not a big deal anyway. After all their share will go from the profits and not out of my pocket.

Micro view

What will be my main activity?

Find, exchange, compile and create information

Set up infrastructure to interact with the decision makers (government executives), policy makers (politicians, bureaucrats, think-tanks) , the media, competitors and customers.

What will be my marketing tactics?

Let everyone in the industry know there is something in for everyone. Brokers, charting software providers, system developers, signal providers, advisories, analysts, researchers, educators and the media. Word should spread so that bounty-hunters come to me.

What will be my tactics with competitors?

Create uncertainty in the market. Make it appear random. If I don't create confusion I will lose in this business. So my entire focus will be on achieving this objective. If I can't create uncertainty in the market I should not enter this business.

What will be my tactics with customers?

Create market noise constantly so that the customers remain in a state of confusion.

Create feelings of scare (fear) or overconfidence (greed) among competitors/customers so that they become careless and fall in my traps. I am not content with just having spread income. I want to have huge profits. I must evoke strong emotions so that my customers make mistakes and I profit from them.

What kind of resources do I need to have?

Set up internally

* Lobbying team to liaise with the government to keep them happy so that they don't change their mind to hand over the currency exchange fixing to a regulated exchange or international agency

* Research team to understand customers psychology and behaviour so that human emotions can be triggered which will benefit my business.

* Coordination team to keep relations with other market makers and form an invisible syndicate to act for mutual profit. I know if I try to grab all profit for myself I won't last for long. I have to work with a lot of other mighty people and ensure everyone has a slice of the cake.

* Statistics research team to obtain information through open sources: I must have information on all markets - stocks, futures, options, which are regulated and the information is available publicly. Basically it will be a research team to compile information and find some patterns and use the information in sync with the market makers of other co-related markets and work together with them again for mutual profit. I want to ensure the movements in all the markets, stocks, futures, options, spot currency support my business. The market makers of these markets will need my support too . So it will be a two-way affair with one single goal - MAKE PROFIT. Someone can call it insider trading but can it be proved?

* Media/community relations team to drum up massive market noise to make sure my customers and competitors remain confused and lost all the time. One person is saying buy, another person is saying sell and third person is saying wait. Thank God I don't need to create an independent structure to do this job. Various Government organisations are churning out statistics day and night out happily . My job is to provide interpretation to those data releases as it suits me.

Market appetite or risk aversion? I decide what suits me. The whole analysis and media industry is happy to go along.

How do I do it? Well, there is a group of people who have a lot of time on their hands to think, talk and write about. Inforrmation is their staple food. Academicians, analysts, researchers, advisory service providers, educators and media can't function in vaccum. I will give them information/mis-information food in plenty . Let them create incessant market noise and information overload. I need to divert attention of my competitors and customers . I just don't want them to use their own brains.

set up externally

* Industry liaison team to obtain information through secret sources: I must finance an organisation which will give the appearance of an independent think-tank or research organisation but their job will be to get information from my competitors and the government and pass it to me. I want it to be a covert operation. If syndicate/cartel/monopoly group (known as association) is with me then it becomes easier to fund a so-called independent organization because the source of funding can't be traced to an individual.

Technology team plays a crucial part in my business

My trading robots can't function randomly. Can they? Without definite timing and choice of start/end points, volatility can't be triggered. I provide liquidity to the market but I want my customers to think that the liquidity is coming from the market. If I stop providing liquidity to the market do you think the market will move by itself? No way.

I must say I have an advantage (not illegitimate) over my competitors and customers. I know at what levels their positions are parked, clustered and ready to become live. I know how they are moving second by second.

When I see big roadblocks in my way, the robots freeze trading screens all over the world for a while and i use 25-100 pips balls to remove them. My tech teams say if I want I can use a huge ball of up to 2500 pips. (picture attached) and if needed the size can be greater than that too.

( It seems in case of Euro/Nzd a bigger pips ball than the one pictured was used in a period of 15 minutes to crush hedge funds who were trying to cash in on carry-trade interest differentials. If someone finds this chart please post it here.)

My Pac-Man is trained to trigger stop losses, entry/trail orders and jump in and out at stunning speeds. I can unleash Tsunami waves (analysts call them multi-year trend reversal) when fund managers threaten my existence itself through options-related trades.

All these patterns undoubtedly will be complex but I for one surely know that a competitor/customer can recognise it in varying degree from 100% to 1%. I must use tactics which should not allow them to use their brains against me. I must make them numb. I must create market noise so much that even a smart person should start doubting his/her recognition skills and refuse to act, particularly during the high volatility wave. After all, I create the wave to benefit me and no-one else.

I would love to know the reactions from tech guys who have worked for market makers. If they can share their knowledge it will be of immense help.

Market makers have such a big clout that they hardly care if their secrets are spilled out. Educated retails traders are handful and they are no threat to them. Neither in last 100 years nor now and not in future too.

PJ

Mikkom,

I don't believe in risk/reward concept. I can't take any loss more than 5 pips. I will not go on taking losses in the hopes of making bigger profits to offset those losses. This concept doesn't appeal to me at all.

Basically what I do is buy and sell n same currency using single margin. So my drawdown is controlled and my margin remains in tact. In order to control my costs, I have negotiated with my broker not to charge me rollover interest rates by promising big turnover.

When you are hedging you have to make sure that a minimum of 10% is left in usable margin. Once I had left 3% as usable margin and saw some of my positions closed out because of margin calls. While I was asleep, the market makers widened their buy/sell spreads and I lost my supposedly safe hedged positions. I have made it a rule to keep not less than 15% as usable margin, which is quite safe. During news trading also the spreads widen and I must be cautious.

Hedging of 20-40 pips for first time is ideal. I don't use stop loss and lock up my loss at 20-40 pips and forget about the position for some time.

I have to be good at identifying retracements/pullbacks/reversals So I wait for the retracement/pullback to happen and then liquidate the unfavourable position at my chosen time. I am in no rush to liquidate my hedge in the same hour or same day or even same week. If I don't feel confident enough to identify the reversal I would wait for another opportunity.

Liquidating one position amounts to opening a new position and it should be planned and executed with patience and clear mind.

Anyone who wants to use hedging should use demo to do a few loss making transactions and lock them up by hedging in same currency. Once hedging is done, wait for opening/closing hours/data release - times when the volatility is expected to be high - and liquidate a position by identifying the reversal point.

Hedging will get you immense profits if you succeed in identifying the reversal pattern correctly. If you fail to identify the reversal pattern and instead get caught up again in stop hunting then lock up the loss again and wait for another day to act. This process will be never ending. You make a mistake, just lock it up and wait for another day.

Hedging has literally forced me to concentrate on my skills to identify reversals because I don't want this unending chain of hedging to continue.
I like hedging immensely for this reason alone because it enforces discipline in me.

Hedging costs: 2-4 pips spread. Is it a big cost to grudge about? Even if you lock up five times the cost is worth it. Just one good timing of reversal during high volatility time can fetch you 50-80 pips profit in 5 minutes for gbp/jpy, which is my favoured currency. It is considered a beast but with hedging it can be tamed.

Edge: I lack tech skills. I already value your indicator coding skills and now your focus on systems trading is timely. I will be glad to share with you any real-life experience of trading fx options (14 years) and spot fx (less than one year). I lack energy to watch charts and maintain a log but not the passion to trade.

If you can define a pattern of breakout and connect the same breakout with number of pullbacks/ reversals/ retracements in minutes, hours, days, weeks, months, years it will give a good idea to work out an hedging edge.

I can make a structure of the simple system based on watching previously live statistics and charts to catch the aggressiveness of the market makers .

I want to have a few indicators which can :

(a) measure the velocity and size of the pip balls the market makers use to scare buyers and sellers. I am using my chart watching to determine it and as well as how often my trading screen freezes.

(b) identify the sessions pattern : how many ups and down (in terms of certain pip levels) are achieved in a row and during selected times. I am doing it manually though I know a simple indicator can do this job.

(c) define shifting of trading zones based on parameters of pips or percentage. Past Trading zones, transit zone, and new zone being formed. I am doing it manually.

(d) capture movements during open/close hours for major pairs and the rest of times so that a daily time pattern (measured in terms of volatility) can firmly be established.

(e) establish timings of cross currency pairs: When major pairs are active, they are sleeping. When they wake up, major pairs go to sleep. This is a confirmed pattern. Again backtesting will be helpful.

(f) point out which currency pairs are liquid. I suspect majority of Nzd currency issued is hoarded by market makers. How can you justify big spikes in one bar unless you know most of the currency is with you and the spike will hardly affects you. Again I feel this currency can move in one direction for months together and in a row with small retracements in between. I have just noticed this pattern and I don't know how to use system tester to backtest it and identify as a definite pattern based on historical data.

(f) identify market sentiment - sellers and buyers volume on a platform of a big online broker like Interbank fx or so. It will not be a true representative of the actual volume happening on the market but nevertheless it has always guided me in showing the herd mentality and I take it as a representative of the trader community as a whole. It does acts as a live sentiment indicator though I know it is not accurate. I use a defective indicator "boxingislife" to find out volume of buyers and sellers.

I am used to writing long worded documents (old habits die hard) - which is not suitable for blogs like this. I will try to change this.

Cheers

PJ

Can you eplain this in more detail. Do you buy and sell 2 currency pairs against one currency?

This particular project? I think the time can be seen from the posts I have made to this thread - however the ground work (what I use and what I don't use) are something I have learned during years of studying the markets.

I don't use genetic programming in it's usual form at the current framework, I use something simpler (I don't want to go into details at this point) at this point.

If you want to use gp, the larger the population, the better the result but be sure to restrict what an algo can do and the most important thing: find a good fitness algo or you will get real curve fitter strategies - gp and all evolutionary algos are really good at curve fitting.

If you really have been able to negotiate no rollover interests, then by all means play old good broker hedged interest game too. Buy plus interest rate pair at a broker that gives you rollover and then hedge with one that you have negotiated the zero rates with.

Free money - who could resist that

Anyway, I'll see what I can gather from the rest of your posts - this thread is mostly about things you can quantify and I don't think there is much stuff on your posts that I can quantify so it might be a good idea to discuss about non-quantifiable aspects of trading or general trading philosophy on some other thread..

One pic from a bit different algo I'm using at the moment, in the past runs I used a risk scaling so that when a new algo was added, risk was split between the algos. No more - now I'm using a stricter selection method and not scaling the risk, I'm instead using a random risk of 0.5 to 3% per trade and checking the correlation between each new algo and the results are even more impressive. I think in couple of weeks I'm moving to live with this - first probably with a algo to mql4 converter and after that when I have time, I'll do a real ecn connection, most likely to IB ot MTB.

How can any broker agree to this ? Are they fools? Our horizon has to be big and not think of petty profits.

It is very clear: If I don't pay interest then do I have any right to expect to be paid. Come on. How can it be like this? Are you joking?

PJ

I was inviting tech guys to come out and share what they do at your thread.

The whole exercise was aimed at that.

Stakes for tech guys like you are high and you need to understand the whole business not just some part of it if you want yourself to be a professionalt trader rather than just be happy doing systems development job.

Now it appears the thread is aimed at the professional systems developers rather than traders.

It was my mistake. Any way, thank you for bearing with me.

PJ

thanks for the answers mikkom!

I'm not joking - you have to use 2 brokers/accounts for the hedge, not one. This is not a new way of making profit.

All you have to do is to check that you have enough margin in the both accounts all the time, margin call on either accounts will expose you to directional risk.

Yes this thread is much more about finding quantifiable edges and trading system development than market chitchat in general - there are many threads about general market structure.

That said, you have many research points in your second post that actually fit pretty well into quantifiable part in your second post so maybe that is where we can go deeper to some parts, for example the parts of session patterns (I have had a plan to do some research about this in the past) and time-volatility part.

About what pairs are the most liquid, there are some BIS (bank of international settlements) documents that tell the actual spot trading volume for different pairs, I'll see if I can find the link.

edit: Here it is, "triennial central bank survey" http://www.bis.org/publ/rpfxf07t.htm

Direct link to some statistics http://www.bis.org/publ/rpfxf07a.pdf

About the kiwi spike, I think that's an error in data, so called bad tick, not actual spike.

This is the most interesting thread I've read yet on FF after a few months of browsing off and on.

I have some thoughts on algorithmic discovery and selection that I have been mulling over for a while. Essentially my idea involves formulating the design as a combinatorial optimization problem where a "solution" is a particular trading strategy. The idea is then to use local search metaheuristics to search the solution space and evolve good trading systems.

I've programmed the framework in java and have about 8 years of tick data. I'm currently working on coding up all the "building blocks" that the strategies will be composed of.

When I get some results I'll share, and I'll look forward to more updates in this excellent discussion.

Just for starters, here is a 2001-2009 average hourly range (hourly max-hourly min) range list as a text chart. My data is gmt+1.

This is sorted version

For those not aware of it, kreslik.com has a few threads on GA and fitness functions:

http://kreslik.com/forums/viewforum.php?f=6


Also check out

http://kreslik.com/forums/viewforum.php?f=14

for statistics similar to the hourly range posted here.

I have read some of the stuff done by TRO but I'm not quite sure if I can take him seriously - for GA and fitness functions I would much rather read some university material than his writings, I'm not sure if this is too harsh but some of the stuff I have read seems a bit amateurish (I'm sure someone thinks the same of my stuff too )

The biggest problem with evolutionary algorithms and fitness functions in general and trading is that computers are really good at optimizing - therefore they will always find the easiest route to curve fit the results - that's what they are meant for, finding a correct solution for the given set of data.

Fitness function alone is not the key but if a good (and usually simple) fitness algo is combined to simple enough algos, curve fitting is harder (but not impossible) for computer. Also look ahead factor is really hard to avoid when using historical data.

The Kreslik guys post are worth reading. The site was doomed when you-know-who took over.

My thoughts exactly.

I didn't know this forum is for gossiping and badmouting people