A question about designing a medium frequency system, 10k trades per day

Quote from JuniorCTA:

ULLDMA applications cost anywhere between $25,000 - $100,000 to develop.

And that's outsourced; Ukraine, Pakistan, etc.

All your IP is at risk, even if you modularize development :)
More...

I think if one has a proven edge, the cost will be irrelevant in such a case.

On the other hand,IP proection is highly relevant. I think to keep your edge away from the newcomers is mandatory. To keep it away from the top institutional few, such as GS and others, is absurd because they already have it and have much much better than it.

ULLDMA software side is not from the few. They are serving the few.

My two cents. I might be wrong but this is my opinion till now :)
 
Quote from JuniorCTA:

Do it yourself or pay upwards of $100k
More...

To do it myself... it is an ideal solution for sure.

However, what if i know that this would take me relatively long time ( 2 years at min). In this business, as you know, time means money and more.

So, i have to make a compromise..

Ok, the question, to how much i should go?

To play slower ..... this is a broker compromise. This reminds me with people who swing to postpone loss ( or to be profitable as they believe).

To hire programmers ..... this is coding compromise.


To use already-made framework such as open source platforms or even API.

However, i started the journey and i like it the most.

Thank you for sharing your opinion
 
Quote from Random.Capital:

The second question is the important one.

Anyway, best of luck to you.

Cheers!
More...

I know the answer but i wish i could answer it.

Thank you for your kind wishes and thank you for sharing in my thread.

Highly appreciated.
 
Quote from amazingIndustry:

With all due respect, you are a moron! You could as well apply for a senior PM type role at one of the top tier hedge funds, you are about as qualified for that than for developing a >mid frequency trading and execution platform. Others have made their careers in this field (studying CS, working in financial services, or having traded for long time to understand market micro structure and execution related intricacies and then worked their ass off for years to either work successfully with programmers or have learned to code themselves (more than your "Teach yourself Java in 24 hours" book) but you are convinced you are as qualified as all of them to compete in this field? Give me a break, you are delusional if that is what you are thinking...

My honest and very sound advice to you is to start profiling mid-range frequency strategies, this is anyway where your edge in pure retail is the largest. Aim to execute a few trades per day at the absolute maximum. Once you have a strategy that looks promising, peruse an off-the shelf solution to start trading it systematically. You sound like you do not even have much experience in properly profiling and testing new strategy ideas. Do you even have properly cleaned/adjusted/filtered bar data (let alone tick data)? Only stupid people start to delve into a new field by attacking the most complex aspects. You should start with the basic and develop a firm grasp at that before you move up the tree. I do not think you get better advice in this thread in my honest opinion.
More...

Your tone is really pathetic.
 
Quote from amazingIndustry:

sorry if it agitates you, I prefer to tell you my true thinking. Hope some of it sinks in, its entirely up to you to take my advise or leave it...
More...

Your advice did not agitate me my friend.. but i have enough reasons to leave it. That is it. I do not want to engage in a useless arguments because this is simply will be a waste of your and my time.

What i care actually is to get constructive, open-minded and professional opinions and i got very good ones. Thanks for all who shared.

I have no doubt that this is your true thinking and i thank you for it. However, i disagree with most of it because either it is too general or too narrow. If i elaborate more, i would go into a fruitless endless discussion which i do not want to go.


Good luck .:)
 
Quote from mcgene4xpro:

No, it is not. Unless you consider 6.9 orders per 60 seconds is a HFT. :)

Even, i can not call it Medium FT.

My two cents.
More...

Dude,

Come down to earth from flying clouds. You are just not equipped to do it.

As someone said, putting in 10k orders a day is doable. What is harder is order management, and open position management.

Btw, 10k orders per day is HFT not MFT. MFT is like 10-50 trades per day. Don't go by your gut feel, get a grip by talking to listening to professionals. All the subsecond stuff in HFT that you hear about is ultra-HFT, not HFT.
 
Quote from amazingIndustry:

With all due respect, you are a moron! You could as well apply for a senior PM type role at one of the top tier hedge funds, you are about as qualified for that than for developing a >mid frequency trading and execution platform. Others have made their careers in this field (studying CS, working in financial services, or having traded for long time to understand market micro structure and execution related intricacies and then worked their ass off for years to either work successfully with programmers or have learned to code themselves (more than your "Teach yourself Java in 24 hours" book) but you are convinced you are as qualified as all of them to compete in this field? Give me a break, you are delusional if that is what you are thinking...

My honest and very sound advice to you is to start profiling mid-range frequency strategies, this is anyway where your edge in pure retail is the largest. Aim to execute a few trades per day at the absolute maximum. Once you have a strategy that looks promising, peruse an off-the shelf solution to start trading it systematically. You sound like you do not even have much experience in properly profiling and testing new strategy ideas. Do you even have properly cleaned/adjusted/filtered bar data (let alone tick data)? Only stupid people start to delve into a new field by attacking the most complex aspects. You should start with the basic and develop a firm grasp at that before you move up the tree. I do not think you get better advice in this thread in my honest opinion.
More...

OP, listen - Amazingindustry is giving you here a reality check. You are not taking his feedback, rather you are shutting him off. CS PhDs from MIT and Berkeley find it takes a long time and experience to code up a high performance hft trading platform and manage a HF operation. It takes years of experience at coding to even think of attempting such a thing. When you say it will take you probably 2 yrs to learn Java and use it to trade at HFT (your 10k orders per day frequency), you have got no idea of how hard it is in reality.

Attempt true mid-frequency stuff. Take 1-10 trades per day and you will do much better. Common man, think about it, if you taking 4 trades per day in Euro, you can move 100-300 million per side per trade. If you are taking 10k trades per day, how much size can you move? HFTs make consistent money but they quickly reach their capacity. Whats the point in doing HFTs, anyways!
 
You must log in or register to reply here.
Back
Top