A question on required model's trial

[vs1]

Hi,

Could fellow members please share their opinion on the following:

Generally aiming at collaborating with High frequency trading players, together with a mathematician we’re completing a first set of model’s trials on 1-minute historical intraday data;

What father trials (on what type of data..etc ) should we conduct before I can interest HFT players ?

Thanks in advance !

vs1

 

[TheBramble]

Why are you asking here?

Call up your contacts in your client base and ask them what they want.

If you don’t have any contacts yet, then you’re in the wrong ball-park anyhow.

Eli Magid?

 

[vs1]

Why are you asking here?

Call up your contacts in your client base and ask them what they want.

If you don’t have any contacts yet, then you’re in the wrong ball-park anyhow.

Eli Magid?

----------------

Well, that's your opinion
I do hope that other members' attitude is way more positive then yours

 

[TheBramble]

----------------

Well, that's your opinion
I do hope that other members' attitude is way more positive then yours

That's very likely. Unfortunately, it will for that very reson be less useful.

The fact you didn't respond to the name I gave you tells me everything.

Zol zion mit Mazel

 

[anley]

What further trials (on what type of data..etc ) should we conduct before I can interest HFT players ?

I don't think you can really interest many of them because most of them will be so much more advanced than you, including their systems and the all-important investment in their operations.

As for what other data you'll need how about the last 5 years minimum NYSE tick data and volume of every print, plus any historical DOM you can get your hands on. That won't come cheap, I'm guessing here but I wouldn't sell it for less than $50,000.

Overasll it's a bit like you're playing with an MACD crossover system (for example) and are trying to tempt a big bank's prop desk to talk turkey and see if there's anything you're doing which will interest them.

Trading is a blunt business, and I'm being blunt, so please don't take offence.

 

[vs1]

Thank you anley.

I am fully aware of the fact that they are much more advanced than me, I might have an approach to managers of a major player in this field, I am trying to learn here what 'must' I bring with me in terms of performance proofs of our model, to avoid being sent back to perform more / other trials..

vs1

 

[vs1]

That's very likely. Unfortunately, it will for that very reson be less useful.

The fact you didn't respond to the name I gave you tells me everything.

Zol zion mit Mazel

--------------

I understand that you think I am someone that you know (eli something)
Well wrong - and as you can see, people do want to help

 

[anley]

what 'must' I bring with me in terms of performance proofs of our model, to avoid being sent back to perform more / other trials..

A 1-2 year real-money track record with drawdowns of less than 5% otherwise you're all talk and they've heard it 1001 times in the past.

Same with me, and anyone else reading this so don't take offence.

 

[anley]

Also, it will be extremly hard for you or anyone else to get a good HFT track record because most of not all your ultra short term profits will be eaten up by the cost of doing business.

 

[Pat494]

If your system is any good use it yourself - don't bother with the banks imho

 

[vs1]

cost of doing business.

Appreciate your insights!

Harder questions.. :

Is there a way for me to run such a 1 year real data without actually buying the low latency hardware and the other major costs ?

I have been trying to search the internet on HFT 'cost of doing business' and did not find sufficiant material, if possible I'd be happy to kearn that too

Thanks!!

 

[vs1]

If your system is any good use it yourself - don't bother with the banks imho

Thanks,

that is an option though if my only strength is the math. model and capabilities, dealing with all this field Difficulties, will, in the good case, very well stretch our 'time-to-market'.. correct ?

 

[Pat494]

Thanks,

that is an option though if my only strength is the math. model and capabilities, dealing with all this field Difficulties, will, in the good case, very well stretch our 'time-to-market'.. correct ?

yes indeed but dealing with the banks is costly usually. They need to make lots of money quickly. Would they have time for budding traders and their home-made systems ? Probably not unless you have it up and running and can show consistent profits.
If they want it they would tie you up in legal chains for a long time and pinch your idea !!

 

[anley]

Is there a way for me to run such a 1 year real data without actually buying the low latency hardware and the other major costs?

No, because you're trying to break into a billion $+ a year business, one where the best people will invest the serious money in people, research, computers and anything else you can think of.

HFT is a game for well capitalised companies, not for someone with a £400 Dell working in their spare bedroom.

But in trading a £400 Dell can be used to make serious money, just not via HFT. If I were you I'd forget about HFT and try to come up with some other trading strategies which aren't so resource hungry - like simple long term charting. That's always worked well and will continue to work well, but the learning curve is steep and long for most people.

 

[vs1]

No, because you're trying to break into a billion $+ a year business, one where the best people will invest the serious money in people, research, computers and anything else you can think of.

Thanks.


HFT is a game for well capitalised companies, not for someone with a £400 Dell working in their spare bedroom.

But in trading a £400 Dell can be used to make serious money, just not via HFT. If I were you I'd forget about HFT and try to come up with some other trading strategies which aren't so resource hungry - like simple long term charting. That's always worked well and will continue to work well, but the learning curve is steep and long for most people.

I understand what you're telling though the the mathematic model developed by my partner suits S h o r t terms trading.. this is where we might have an advantage..

Thanks again.

 

[anley]

I understand what you're telling though the the mathematic model developed by my partner suits S h o r t terms trading.. this is where we might have an advantage..

Thanks again.

It may well do but the cost of doing business will probably mean it's impossible to make money, net.

Your commissions for example will be say 0.01 cents a share, but you're up against people being billed at 1/10th of that, maybe even 1/50th of that and they also get major commission rebates. Then trrow in who gets filled first and last, you won't be first, you'll be last and that can easily be the difference between earning millions and losing money.

You're bringing a flick knife to a machine gun fight regardless of your strategy and that's the key point, the strategy is only part of the plan to successully HFTing. Setup and all the other things that go on behind the scene are critical which is why the profitable HFTers invest millions.

 

[vs1]

Understood.

Thank you anley.

 

[theknifemac]

A lot will depend on how short term you are talking about. Whats your average holding period for a trade for example ?

Mac