Starting automated trading, what are most common setups?

The best ES trader I ever met made - career average - 50c /contract. BEFORE FEES. Trades 50k - 100k contracts a day.

That's considered pretty good for a mix of manual and automated on ES. You should do everything possible to trade as infrequently as you can. Anything that trades more than a hundred or so times a day has been picked over to a degree that's hardly imaginable. So, stick with low frequency - maybe a dozen or fewer trades a day.
 
dholliday said:
Mainly for more symbols. Sometimes I am watching more than 1,000 symbols for specific events to occur. If it happens great, otherwise that system won't have any trades that day.
More...

Is that some kind of swarm algorithm? If they all go crazy, then maybe it's not so crazy? Or does it look for meaningful divergence?
 
Last edited:
garachen said:
The best ES trader I ever met made - career average - 50c /contract. BEFORE FEES. Trades 50k - 100k contracts a day.
More...
Could you please confirm, 50 cents profit per contract and 50 thousand to 100 thousands contracts per day?

So the profit for this ES trader would be 25 thousands to 50 thousands per day?

Whoa, this is hardcore, I don't think I can do this on IB

garachen said:
That's considered pretty good for a mix of manual and automated on ES. You should do everything possible to trade as infrequently as you can. Anything that trades more than a hundred or so times a day has been picked over to a degree that's hardly imaginable. So, stick with low frequency - maybe a dozen or fewer trades a day.
More...

My current platform is IB. I will optimise between number of trades and fee, to maximise my net profit.
 
dholliday said:
Mainly for more symbols. Sometimes I am watching more than 1,000 symbols for specific events to occur. If it happens great, otherwise that system won't have any trades that day.
More...

Oh, I see, thanks, I will get IQfeed when I complete my work on Emini S&P.
 
tradingcomputer said:
Oh, I see, thanks, I will get IQfeed when I complete my work on Emini S&P.
More...
Don't go with IQFeed. They are shit when the market is volatile. Not slow, they just don't work, returning 'Symbol not found' errors all over the place. You don't want this to happen especially in fast market. Not reliable at all.
 
tradingcomputer said:
Could you please confirm, 50 cents profit per contract and 50 thousand to 100 thousands contracts per day?

So the profit for this ES trader would be 25 thousands to 50 thousands per day?

Whoa, this is hardcore, I don't think I can do this on IB



My current platform is IB. I will optimise between number of trades and fee, to maximise my net profit.
More...

That's correct. But fees are going to eat about 2/3 of that. My point is, I've never seen an ES trade that can function at the retail level, and the edge is pretty thin at high volumes. So I'd suggest a different product or very very low frequency.
 
garachen said:
career average - 50c /contract. BEFORE FEES. Trades 50k - 100k contracts a day.
More...
garachen said:
I've never seen an ES trade that can function at the retail level.
More...
They do exist. There were two guys at a British trading company keeping at least $7/round trip in ES and doing a few tens of thousands per day. So they could start up with $20k and an IOM lease.

One went to Hong Kong and the other is soon to be visiting the Northern District of Illinois.

Discussing these extreme outliers is mostly pointless in this context. Except to say that not everything which counts can be counted. Just because we cannot see a structure which would allow such high P&L over millions of contracts traded doesn't mean it isn't there. I'd venture to say that computer/maths guys are especially unlikely to discover it because of their approach. The information isn't in the data.
 
Ghost_of_Blotto said:
I'd venture to say that computer/maths guys are especially unlikely to discover it because of their approach. The information isn't in the data.
More...

Thanks for the info, at least, now we know it is possible.

The information is not obvious in the data. It is hidden.

Understanding the data, in this context, the ES data, the behavior of the order book, the latency of the setup is very important. In short understanding the domain is paramount, it will potentially simplify the problem significantly.

I believe the key is how to integrate the domain expertise to guide the algorithm so the search space is limited. Otherwise the problem to be resolved become too hard to solve (e.g. NP Complete), like in this old thread: http://www.elitetrader.com/et/index.php?threads/computer-generated-trading-strategies.117544/

That is, the holy grail is in how to integrate human expertise in automatic search for profitable strategies.
 
Last edited:
You must log in or register to reply here.
Back
Top