we developed trading software for a customer that implemented an extremely simple algorithm to decide between buying/selling stocks. Each calculation took less than a millisecond.
Of course this wasn't very impressive. Something that calculates so fast can't be worth anything.
So about half of the project's development time was spent on creating a realistic (that is: unpredictable, jerky) progress bar that created the illusion of a complex analysis taking 30 seconds or so.:-)
This is an interesting proposal. We are using something similar at my company, which offers one of the most detailed databases for retail structured products.
What some people posting here misunderstand is that the proposal isn't about making available magic formulas or explaining, why a call option retails at $12,34 at a given moment. It is simply about calculating the resulting cash flow that the buyer receives from a product in a given scenario. So it let's you try out, what you receive when you assume that at maturity the underlying stock is at $15.
It does not automatically tell you if the option should be worth $12,34 or $2,34 right now. The current fair value is part of the magic sauce: It can only be calculated with the right assumptions, eg. the assumed volatility, the current interest rate structure,... Different market participants will usually make different assumptions here, but with these assumptions plus a numerical pricing model plus the program supplied to you under this scheme you can also calculate your fair value for the product.
In my opinion a general purpose language like Python is not a good fit for this use. In any case, very strict interfaces would need to be defined. If there were multiple ways to input the assumed stock price (which is just one of a very large number of parameters), then this would not be much more useful, than the current written documentation of the products. With a general purpose language there might also be hundreds of possibilities to program even simple structured products, which makes reading/debugging these programs needlessly hard.
We have developed a special purpose language internally, that can be used to easily program variable cash flows that depend on other market data. The good thing about such a special purpose language is that most standard product have a generic representation, that is short enough to be easily read, understand and (if needed) debugged. We supply "code" for the more than 400.000 active retail structured products in our database to our customers and they can for example try out, which product works best for then in a given investment scenario.
You might have a similar effect by using a generic language + strict interface definition + enforced use of a standard library. But a compiler/interpreter for a simple special purpose language can easily be implemented in most environments while many possible users of this project might not like adding a full python-interpreter to their system.
Does anyone know a cheap external RAID system that connects via USB 2, supports 4 or 6 HDDs and is optimized for desktop usage (ie. power-saving and silent)?
Level 1 or 2 has got nothing to do with the delay of the market data (realtime or delayed).
With Level 1 access you can only see the best bid and the best ask in the market. Level 2 market data includes market depth. It gives you the 10 (depends from exchange to exchange) best bids and asks.
This way you can see, whether a good bid is just a single trader trying desperately to buy a few shares or if it's just the tip of an iceberg.
Delayed level 2 market data is quite common. It just doesn't make too much sense.
At work I look after a number of databases for financial instruments: options, certificates, bonds, funds,...
When we started to collect data for our funds-database about two years ago I was shocked that at most funds-companies a central spreadsheet is used as the main data-source for all information. Companies that invest billions of Euros rely on an Excel-sheet lying on a central fileserver for most of their data-housekeeping.
Well, not all the companies work this way. There is even an organization that tries to standardize an xml-schema for the exchange of funds-data (http://www.funds-xml.org/) - but of course no one really uses it.
I have build something similar one or two years ago.
It's quite easy to do: Use Javascript to install an onclick handler that sets a cookie with the click-coordinates. On the server-side you need to check the cookie and the referrer and store url and coordinates in a database. It's best to limit this to just one browser, ie. the much hated internet explorer 6 (because of slight differences in positioning algorithms). Also, if you have a high traffic site, you might only want to store a small percentage of these tuples.
To diplay the click-information for a web page simply cumulate all click in a square of a certain size (ie. 20) and draw a small rectangle filled with a colour corresponding to the number of clicks.
It really helped us to identify some unused/crappy navigation on our website.
Any free in-memory-databases that you are aware of? TimesTen and Polyhedra are quite expensive and I couldn't find the license info for datablitz, but sinde the website announces "...is now a product!" I guess that it's not free either.:-(
If it doesn't, bad luck. One size doesn't fit all. Use something else. But I'm pretty sure, that most databases are much smaller than half a gigabyte.
I'm currently planning the development of a quote-server. The underlying market-data feed can contain several thousand updates per second. If someone could point out another way of doing this, please respond.
Can anyone recommend anything else for a project like this (thousands of updates per second)?
I'll probably end up using a ready-made in-memory db, or role my own, since the application area is very specific and no general routines (for indexes,...) are needed.
This is not a question for the interview (just in case you didn't notice).
For our company we need to set up a solution that streams audio-files (interviews). This is not supposed to be a continous stream like a radio broadcast for which something like icecast would be great, but a single interview that is supposed to be streamed to the user when he clicks on a link on out webpage.
Most modern players support streaming MP3 when you send a.M3U (Winamp Playlist) to the web-browser that contains hyperlinks to your files. Unfortunately there is no way to seek in the audio to skip boring parts, you have to listen from the beginning to the end.
Commercial alternatives like Real or Microsoft support seeking because the implement their own protocol and don't "abuse" HTTP. Is there a free implementation of a similar (or compatible) server? Can Apple's Darwin Streaming Server seek in MP3s (the website is not clear on this subject)?
No, we don't have an open relay. We have everything properly configured and don't allow relaying. But some %'&$"#!-spammer decided that using michael@ourdomain.de in the "From:" line would be a good idea when sending out spam.
I get several hundred emails per day, either automated replies that tell my, that "your message to iojrf323@yahoo.com could not be delivered" or angry users that accuse us of spamming.
I try to contact the admins of the abused systems and enter their servers into an open-relay list, but that hasn't slowed down the rate of incoming emails.
IMHO the important part about WAP isn't the protocol to deliver web pages to your phone, but the well-formed XML (WML) used to describe them.
Fast wireless networks are cool, but they don't enlarge your phone's display. No matter, how fast your connection is, reading even the stripped down HTML version of slashdot on an phone will always suck. Enter WML, specifically designed for small screens.
If only Nokia, phone.com, symbian and the others would implement the standards properly... You hate incompatibilities between Internet Explorer and Netscape? You will love them, after designing pages for WAP-phones.:-)
You must be kidding! Futurama is my favourite Cartoon right now. I need more than 2 seasons. I hope this is just a stupid unjustified rumour. bye, Paul.
Please, you must be kidding! Futurama is my favourite Cartoon right now. I need more than 2 seasons. I hope this is just a stupid unjustified rumour. bye, Paul.
Slashdot Mirror
we developed trading software for a customer that implemented an extremely simple algorithm to decide between buying/selling stocks. Each calculation took less than a millisecond.
Of course this wasn't very impressive. Something that calculates so fast can't be worth anything.
So about half of the project's development time was spent on creating a realistic (that is: unpredictable, jerky) progress bar that created the illusion of a complex analysis taking 30 seconds or so. :-)
This is an interesting proposal. We are using something similar at my company, which offers one of the most detailed databases for retail structured products.
What some people posting here misunderstand is that the proposal isn't about making available magic formulas or explaining, why a call option retails at $12,34 at a given moment. It is simply about calculating the resulting cash flow that the buyer receives from a product in a given scenario. So it let's you try out, what you receive when you assume that at maturity the underlying stock is at $15.
It does not automatically tell you if the option should be worth $12,34 or $2,34 right now. The current fair value is part of the magic sauce: It can only be calculated with the right assumptions, eg. the assumed volatility, the current interest rate structure, ... Different market participants will usually make different assumptions here, but with these assumptions plus a numerical pricing model plus the program supplied to you under this scheme you can also calculate your fair value for the product.
In my opinion a general purpose language like Python is not a good fit for this use. In any case, very strict interfaces would need to be defined. If there were multiple ways to input the assumed stock price (which is just one of a very large number of parameters), then this would not be much more useful, than the current written documentation of the products. With a general purpose language there might also be hundreds of possibilities to program even simple structured products, which makes reading/debugging these programs needlessly hard.
We have developed a special purpose language internally, that can be used to easily program variable cash flows that depend on other market data. The good thing about such a special purpose language is that most standard product have a generic representation, that is short enough to be easily read, understand and (if needed) debugged. We supply "code" for the more than 400.000 active retail structured products in our database to our customers and they can for example try out, which product works best for then in a given investment scenario.
You might have a similar effect by using a generic language + strict interface definition + enforced use of a standard library. But a compiler/interpreter for a simple special purpose language can easily be implemented in most environments while many possible users of this project might not like adding a full python-interpreter to their system.
bye, Paul.
That's not a wind turbine, but a joke. That thing hardly produces as much energy as a hamster running around in his little wheel.
Oh great, so you want a guy who denies that the holocaust happened as your leader?
Does anyone know a cheap external RAID system that connects via USB 2, supports 4 or 6 HDDs and is optimized for desktop usage (ie. power-saving and silent)?
Level 1 or 2 has got nothing to do with the delay of the market data (realtime or delayed).
With Level 1 access you can only see the best bid and the best ask in the market. Level 2 market data includes market depth. It gives you the 10 (depends from exchange to exchange) best bids and asks.
This way you can see, whether a good bid is just a single trader trying desperately to buy a few shares or if it's just the tip of an iceberg.
Delayed level 2 market data is quite common. It just doesn't make too much sense.
bye, Paul.
At work I look after a number of databases for financial instruments: options, certificates, bonds, funds, ...
When we started to collect data for our funds-database about two years ago I was shocked that at most funds-companies a central spreadsheet is used as the main data-source for all information. Companies that invest billions of Euros rely on an Excel-sheet lying on a central fileserver for most of their data-housekeeping.
Well, not all the companies work this way. There is even an organization that tries to standardize an xml-schema for the exchange of funds-data (http://www.funds-xml.org/) - but of course no one really uses it.
I have build something similar one or two years ago.
It's quite easy to do: Use Javascript to install an onclick handler that sets a cookie with the click-coordinates. On the server-side you need to check the cookie and the referrer and store url and coordinates in a database. It's best to limit this to just one browser, ie. the much hated internet explorer 6 (because of slight differences in positioning algorithms). Also, if you have a high traffic site, you might only want to store a small percentage of these tuples.
To diplay the click-information for a web page simply cumulate all click in a square of a certain size (ie. 20) and draw a small rectangle filled with a colour corresponding to the number of clicks.
It really helped us to identify some unused/crappy navigation on our website.
I didn't want to say that it was hard. Only that a traditional database would probably break down under a naive implementation. :-)
Do you store the intraday-tick-data, or only the last values? How do cope with persistance, do you dump all data to disk every few minutes?
Any free in-memory-databases that you are aware of? TimesTen and Polyhedra are quite expensive and I couldn't find the license info for datablitz, but sinde the website announces "...is now a product!" I guess that it's not free either. :-(
can be great, if your data fits into main-memory.
...) are needed.
If it doesn't, bad luck. One size doesn't fit all. Use something else. But I'm pretty sure, that most databases are much smaller than half a gigabyte.
I'm currently planning the development of a quote-server. The underlying market-data feed can contain several thousand updates per second. If someone could point out another way of doing this, please respond.
a nice free in-memory db is: FastDB
Can anyone recommend anything else for a project like this (thousands of updates per second)?
I'll probably end up using a ready-made in-memory db, or role my own, since the application area is very specific and no general routines (for indexes,
This is not a question for the interview (just in case you didn't notice).
.M3U (Winamp Playlist) to the web-browser that contains hyperlinks to your files. Unfortunately there is no way to seek in the audio to skip boring parts, you have to listen from the beginning to the end.
For our company we need to set up a solution that streams audio-files (interviews). This is not supposed to be a continous stream like a radio broadcast for which something like icecast would be great, but a single interview that is supposed to be streamed to the user when he clicks on a link on out webpage.
Most modern players support streaming MP3 when you send a
Commercial alternatives like Real or Microsoft support seeking because the implement their own protocol and don't "abuse" HTTP. Is there a free implementation of a similar (or compatible) server? Can Apple's Darwin Streaming Server seek in MP3s (the website is not clear on this subject)?
bye, toolbar.
No, we don't have an open relay. We have everything properly configured and don't allow relaying. But some %'&$"#!-spammer decided that using michael@ourdomain.de in the "From:" line would be a good idea when sending out spam.
I get several hundred emails per day, either automated replies that tell my, that "your message to iojrf323@yahoo.com could not be delivered" or angry users that accuse us of spamming.
I try to contact the admins of the abused systems and enter their servers into an open-relay list, but that hasn't slowed down the rate of incoming emails.
Any ideas?
IMHO the important part about WAP isn't the protocol to deliver web pages to your phone, but the well-formed XML (WML) used to describe them.
:-)
Fast wireless networks are cool, but they don't enlarge your phone's display. No matter, how fast your connection is, reading even the stripped down HTML version of slashdot on an phone will always suck. Enter WML, specifically designed for small screens.
If only Nokia, phone.com, symbian and the others would implement the standards properly... You hate incompatibilities between Internet Explorer and Netscape? You will love them, after designing pages for WAP-phones.
You must be kidding! Futurama is my favourite Cartoon right now. I need more than 2 seasons. I hope this is just a stupid unjustified rumour. bye, Paul.
Please, you must be kidding! Futurama is my favourite Cartoon right now. I need more than 2 seasons. I hope this is just a stupid unjustified rumour. bye, Paul.