Seeking advice on whether trading is appropriate for my situation

Quote from twosigma:

I work as a programmer at an investment bank. I build systems for the credit trading team which receive data from disparate sources, integrates them, and offers analytical views of the data all in real-time.

My strong areas are data management, distributed systems, and functional programming. I'm well-versed in popular algorithms and data structures, and spend about 90 mins each night learning new tech (languages, tools, etc). I also know market microstructure.

I want to start something to generate revenue. Before you roll your eyes, I have no dream of creating a magical trading system which'll flood me with cash. Please read on.

I'm looking at three options for additional revenue:
1) Consult in my areas of expertise. So go out and build data warehouses, business intelligence systems, etc.
2) Build a software product and sell it.
3) Create a semi-automated trading system. Research and test strategies, deploy, and monitor them. This won't be intraday but likely swing.

The primary constraint is that I do not want to leave my job until I'm at least making more in one of these endeavours. I am patient and willing to follow one of these paths for years if that's what it takes. I just don't know which one. They all have their pros and cons.

So the reason why I'm asking this question here is: is option 3 the least likely to succeed even if I'm willing to be very controlled on risk, don't use irresponsible amounts of leverage, and willing to let it take years to reach fruition?
More...

Don't even think about quiting your day job. Even if you make more in one of these options than your real job it won't last until you get a good understanding of price action as Handle123 stated plus all the other advice given.

Psychology of trading destroys. The pressure of paying bills from money you are trying to make in the market is like oil and water, they DO NOT mix.
 
Quote from twosigma:

Ok, the general feedback is that I need to be able to trade profitably manually before considering anything automated. I understand the reasoning. Without the right context, crunching data won't help.

But isn't the scientific method applicable here? My plan was to get historical data and just follow the same process you would with an experiment.

1) Come up with a hypothesis
2) Identify all the variables involved
3) Backtest to see if the hypothesis holds (does x and y cause z?)
4) If it doesn't, start again. And if it does, then ensure there's no curve-fitting and start trading with a small amount of money.

If the signals are generated automatically, then the only manual trading I'm doing is following the system and entering and exiting.
More...

On the one hand, you're falling prey to the same things that we all falll for entering the business - searching for the Holy Grail. And using computers to do so. And thinking you're the first one to come up with this grand idea...
  • Can you backtest your psychology?
  • Can you backtest your emotional response to market activity? The highs... The lows.... The churn... To the fact that your backtest results and reality are magnitudes apart?
  • Can your backtesting tell the difference between real and spurious correlations?
  • Can your backtesting tell the difference between real and temporary "patterns"?
  • Can you handle the inevitable shortening of your timeframe as you silently evolve from reacting to trying to predict price changes?
 
Quote from twosigma:

Ok, the general feedback is that I need to be able to trade profitably manually before considering anything automated. I understand the reasoning. Without the right context, crunching data won't help.

But isn't the scientific method applicable here? My plan was to get historical data and just follow the same process you would with an experiment.

1) Come up with a hypothesis
2) Identify all the variables involved
3) Backtest to see if the hypothesis holds (does x and y cause z?)
4) If it doesn't, start again. And if it does, then ensure there's no curve-fitting and start trading with a small amount of money.

If the signals are generated automatically, then the only manual trading I'm doing is following the system and entering and exiting.
More...

Easy to say hard to do. A backtested strategy does not guaratee future results. It's not thatcut and dry.
 
Check out Dom993 , Acary, Kut2k and others over in the strategy design area if you care to
spend a little time reading on this site.

Good luck.
 
Quote from ofthomas:

only one responded rambled on... "the greatest trader this board has ever met" (according to another thread).... but wait, he will say that all his words were full of wisdom and that I just lacked the intellect to understand them... which btw, I did, I just didnt see how they were relevant to the OP... anyhow, just talking about jack's post.... I placed him on ignore now, that way i dont get distracted with the rambling...
More...

Oh Jack. He's holds two Masters degrees.

Masters in Obfuscation and Masters in Bullsh*t.

I seriously don't get why a person would spend so much time fabricating nonsense. It's a waste of time.
 
Quote from mastacoli71:

Easy to say hard to do. A backtested strategy does not guaratee future results. It's not thatcut and dry.
More...

Yes, there is a different set of questions I need to answer for which I haven't found any information. Questions such as How do I know whether this model, which is losing, is simply undergoing expected losses at the moment or suffering from systemic changes in the market, rendering it obsolete?
 
you seem numerically bent. my advice - come up with strategies based on stats, technicals etc. backtest them and trade at least 5 strats together in a portfolio. You will be better than 90% of people on this website, might still not make money however if all your 5+ strategies suffer from overfitting - however just diversifying over more strategies is a superior way than trading or trying to perfect any one strategy.

The whole process will take around 1.5 years of hard work - around 8 hrs per day - includes software selection, getting acquainted with software, getting acquanited with markets and proper market selection, proper strategy idea selection (like MR, momentum, pairs etc), designing tests on these strategies and finally making portfolio....I did all this took me roughly 1.5-2 years to sort everything out. Then I stumbled because I didn't execute my strategies as I designed them. Took me 10 more months to build proper systems so that I can execute flawlessly, without fail.

Its a long process my friend. Cheers :)

Edit: My experience building semi-automated/fully-automated intraday systems for futures
 
Quote from twosigma:

Ok, the general feedback is that I need to be able to trade profitably manually before considering anything automated. I understand the reasoning. Without the right context, crunching data won't help.

But isn't the scientific method applicable here? My plan was to get historical data and just follow the same process you would with an experiment.

1) Come up with a hypothesis
2) Identify all the variables involved
3) Backtest to see if the hypothesis holds (does x and y cause z?)
4) If it doesn't, start again. And if it does, then ensure there's no curve-fitting and start trading with a small amount of money.

If the signals are generated automatically, then the only manual trading I'm doing is following the system and entering and exiting.
More...
if the hypothesis worked ,you being right next to it , would have seen it, you need to learn to trade first, it might be a pyschological variable that isn't in your computer
 
Quote from ammo:

if the hypothesis worked ,you being right next to it , would have seen it, you need to learn to trade first ,you could get a lot of it trading in sim, it might be a pyschological variable that isn't in your computer
More...
 
You must log in or register to reply here.
Back
Top