An automated trading system is only as good as it's developers. Bad programming and analysis creates bad returns with increased execution costs, with no guarantees ... If you can't trade profitably using what's infront of you, you won't be able to do it with programming. You'll just increase your losses and spend a lot of time and expense programming. Garbage in, garbage out... Hey how many programmers does it take to screw in a light bulb?
Nope. Except for investment banks like Goldman Sachs where a programmer can earn $400K+ per year, most compsci majors from the best schools would not dream of working for a bank. For all the complicated systems (ATMs, Teller stations, check processing, etc.), the bank simply buys a system from a software vendor. The in-house work is mainly custom reports (extremely boring), and maintaining web sites.
GS built most of their systems in-house. Depending on the systems, they sometimes buy from outside vendors. Other banks might do it differently. And yes, good GS quants can get $450K. So do good quants/traders at prop shops and hfs. I-banks regularly recruit at those schools. Compsci/EE/math/physics/stats/sciences majors regularly get recruited by banks, hfs, prop shops, etc. Not all of it for trading roles. Many for quant roles to start. You wrote, "For all the complicated systems (ATMs, Teller stations, check processing, etc.), the bank simply buys a system from a software vendor. The in-house work is mainly custom reports (extremely boring), and maintaining web sites." ATMs? What are you talking about? That's commercial banking. I'm talking about i-banks and trading desks. I don't know what commercial retail banking hiring practices are. They might do boring custom reports and maintaining websites. haha. But most ibanks that come on campus to recruit at the schools mentioned above do not any ATMs or websites. lol. At least not at one of the schools mentioned above...
This is quite true. A level of user input will always be needed. No perfect black boxes here. You’ll find those in fantasyland. In regards to building, my live production systems spent nearly two years in a quantitative research phase using statistical software, before converting the logic to compiled executable code, which itself took nearly a year to smooth out and get in line with the statistical software. If you don’t do your homework, you’ll end up like this poor fish:
Exactly so ... specifically, you shouldn't substitute another horse half-way through a race on the track for which one horse had originally been selected because it was better suited there, even if it isn't doing as well as expected/hoped (or use one timeframe's risk-management parameter in another timeframe's trade). No need to win every race, if the losses are cheap. This is yet another example of "trying to make your win-rate as high as possible" not always being the greatest of strategies.
Xela, You said it best. I should NOT use "timeframe's risk-management parameter in another timeframe's trade"! Win rate is a useless metrics. I'm not sure which timeframe I should stick with...
It's brokerage houses that hire most of the programmers really to design trading systems, software, quant analysis applications, risk management models, pricing models and etc. Programming for banks are quite limited to mostly online interfacing applications as more and more transactions are done online these days.
I said i-banks. When I say banks I meant i-banks. Sorry for the confusion. Commercial banking was never on my mind. Anyhow, what you said is true.
high win rate is usually an indication of holding positions past appropriate stop loss levels. that practice always ends in one trade wiping out many smaller gains.