Team programming for better algo-trading results

Programming is the easy part, the hard part is having a theory. Then verifying that theory, in a controlled manner, avoiding common pitfalls like survivorship basis, bad data, overfitting, liquidity constraints. Then applying good practices like including trading costs, diversification, risk management. Be prepared to have your theories debunked and thrown away, this is the learning curve. Out of this cycle will emerge a set of best practices and tools you can utilise to quickly verify ideas.

So the question is. Are you looking for ideas, or do you have ideas and you'd like to calibrate with others in order to verify these ideas?
Both, I'm looking for ideas plus I do have working (live trading) ideas and I'd like to calibrate with others in order to polish them and optimize.
Thank you for all the brave souls that decided to join in
 
op says he has has strategies that are doing just fine so he wants to know are you in or out, no other discussion welcomed.
Don't take it to the extreme,I would like to hear more opinions,yours included of course, I would also like to take it to a more practical side,
Send me your email so I can add you to the project members group
 
Hi guys,

8 years ago i left bank world to set up arbitrage-technology.com
since time i done it all alone and i confirm, its sometimes very difficult to not collaborate.

i am only interested in intraday and i dont think all mushrooms frameworks you can find on internet now are valuable to trade intraday and compete with hfts...

i built up an hft framework (http://www.arbitrage-technology.com/offheap-trader.html)
and working on equity stat arb.

my concern is that the biggest challenge is not techno or strategy but how to be able to fight predator hfts which prevent you to execute your orders, i mean monetize your back test.

i tried to set up a serie idea on that issue, if the group willing to explore this area, would be happy to participate!
 
Hi guys,

8 years ago i left bank world to set up arbitrage-technology.com
since time i done it all alone and i confirm, its sometimes very difficult to not collaborate.

i am only interested in intraday and i dont think all mushrooms frameworks you can find on internet now are valuable to trade intraday and compete with hfts...

i built up an hft framework (http://www.arbitrage-technology.com/offheap-trader.html)
and working on equity stat arb.

my concern is that the biggest challenge is not techno or strategy but how to be able to fight predator hfts which prevent you to execute your orders, i mean monetize your back test.

i tried to set up a serie idea on that issue, if the group willing to explore this area, would be happy to participate!
Hi,
As quoted before, we are going to explore end-of-day, US equities,
and not intra-day HFT.
best of luck.
 
my concern is that the biggest challenge is not techno or strategy but how to be able to fight predator hfts which prevent you to execute your orders, i mean monetize your back test.

Top HFT firms nowadays rent servers close to the exchanges, and develop their own hardware (FPGAs usually) to compete on microsecond-level. Many strategies rely on this speed, in an all-or-nothing manner. Unless you are prepared to make such investments (in hardware etc.) - then it seems you will definitely have a hard time competing to true HFT shops.
 
  • Like
Reactions: 931
Unless you are prepared to make such investments (in hardware etc.) - then it seems you will definitely have a hard time competing to true HFT shops.

there are a new breed of hft i define as "vampire traders" they have even more sophisticated hardware, data acquisition, algorithms, order entry style etc. than all the high frequency traders and they are able to basically prey on hft.
 
  • Like
Reactions: 931
i coloc to exchange...but not connected in fpga despite tcp accelerated.
i agree its impossible to compete with hft. top of them are below 20 micro tick to trade, but even today if you want to trade a stock simply one time a day. you have no choice to try to set up a kind of antigaming framework (orderflow, hidden liquidity detection, hft activity)...because every order you send from the point their is moore than one tick size of spread IS GAMED. this is what i noticed.

even people who think they escape hft by trading end of day...its a dream. it will work on back test. but as soon as you send your order for auction, you need to handle imbalance process (intraday) otherwise you will enforce your back test...
 
  • Like
Reactions: 931
Back
Top