Newbie question
- Thread starter pcll99
- Start date
my 2 cents:
1. Learn how to trade manually on one instrument... I prefer 24x7 futures or forex.... but equities should be equally fine; Focus on education and do not hesitate to spend money if necessary... the money you pay for education is peanuts compared to potential losses if you don't know what you are doing. See the other comments from reputable sources above...
2. You have to consistently toggle between being an engineer vs being a trader. Engg mindset makes your system reliable and robust... protects you from highly volatile markets, whereas your trading mindset generates the alpha.
3. It takes quite some time to develop a system; the main issue you will be confronted with is to try things out and see what works; try to avoid this urge; this is not a short term dev-ops project. Focus on the long term. Try to architect your system as follows:
(i) Data feed factory that stores all the raw feed that you get from the exchange
(ii) Order management system that can place & manage trades "reliably"... a small bug can cost a lot. Have unit tests that make sure that the broker's APIs are consistent with you have in place.
(iii) Compute engine that can input the raw data feed and generate the indicators, candles, footprint, volume profile... etc
(iv)- replaceable strategy engines that input the data from the compute engine and generate trading signals...
Try to make sure that (i,ii,iii) are develop once and forget (6-12 man month effort max)... iv is a completely different ball game... you need to think very differently; this is where your trading mindset can help.
5. Be careful on what open source components you plan on using... make sure they don't change too often. Have good config mgmt scripts in place right from day one... otherwise, OS version migration becomes a nightmare... I burnt my fingers once...
Always keep an eye on what you are doing... how do you know that you don't know something?
Most (not all) comments in these forums are deceiving... I suggest that if you are serious, approach a broker who has experience in catering to algo traders... your broker can give you good pointers because he has a vested interest in your success...
Anyway, good luck with your efforts....
Last edited:
I've had experience at Optiver in Amsterdam and they were really good imho.
Thanks. All the best in Europe are in Amsterdam. Is that random or geography/exchange related?
thanks for this interesting partic the system stuff.my 2 cents:
1. Learn how to trade manually on one instrument... I prefer 24x7 futures or forex.... but equities should be equally fine; Focus on education and do not hesitate to spend money if necessary... the money you pay for education is peanuts compared to potential losses if you don't know what you are doing. See the other comments from reputable sources above...
2. You have to consistently toggle between being an engineer vs being a trader. Engg mindset makes your system reliable and robust... protects you from highly volatile markets, whereas your trading mindset generates the alpha.
3. It takes quite some time to develop a system; the main issue you will be confronted with is to try things out and see what works; try to avoid this urge; this is not a short term dev-ops project. Focus on the long term. Try to architect your system as follows:
(i) Data feed factory that stores all the raw feed that you get from the exchange
(ii) Order management system that can place & manage trades "reliably"... a small bug can cost a lot. Have unit tests that make sure that the broker's APIs are consistent with you have in place.
(iii) Compute engine that can input the raw data feed and generate the indicators, candles, footprint, volume profile... etc
(iv)- replaceable strategy engines that input the data from the compute engine and generate trading signals...
Try to make sure that (i,ii,iii) are develop once and forget (6-12 man month effort max)... iv is a completely different ball game... you need to think very differently; this is where your trading mindset can help.
5. Be careful on what open source components you plan on using... make sure they don't change too often. Have good config mgmt scripts in place right from day one... otherwise, OS version migration becomes a nightmare... I burnt my fingers once...
Always keep an eye on what you are doing... how do you know that you don't know something?
Most (not all) comments in these forums are deceiving... I suggest that if you are serious, approach a broker who has experience in catering to algo traders... your broker can give you good pointers because he has a vested interest in your success...
Anyway, good luck with your efforts....
Not exactly sure but the Dutch did open the first stock exchange in the world.
the stock 'exchanges' were like open auction exchanges much like cattle auction, tulip auction and stocks were sold like that ...unlike today. it's rigged prices. and fixed. todays' stock exhanges are much like flea market table booths. each dealer putting their price or OTC markets. over the counter market. there is no longer an 'open market' aucdtion. even the T-bill markets fixed same with forex. that is change in the electronic 'markets'. prices are controlled and manipulatd..it's no 'free market'
the entire market is now more like the penny stock exchanges of the 70's or like the over the counter markets. and more suspectible for insider trading and market manipulation etc.
the entire market is now more like the penny stock exchanges of the 70's or like the over the counter markets. and more suspectible for insider trading and market manipulation etc.
Start by looking for a successful trader that is willing to exchange skill set and knowledge.
I think that is the big misconception about the industry. It is quite easy to get into a rut in this profession just as it is in others. If a programmer has been in the same company doing the same thing over and over he/she will never be "nuclear".
I bought 2 of echan's books. All the coding examples are in Matlab. Since Matlab is quite expensive I tried converting them to python. I couldn't quite duplicate his results.