Quest for market data feed
- Thread starter kanonka
- Start date
Your basic problem is LOW BUDGET...
So you are mucking around with "retail" crapware.
There is no serious trading firm on earth using these products...
They are all in the enterprise space...
With Thomson Reuters and their competitors.
You DO NOT save money on doctors, lawyers, accountants, and market data...
That's always penny wise and pound foolish.
Here is the entire enterprise market data space:
http://www.bobsguide.com/
Au contraire:
An expensive feed may be the first step to ruin.
Assuming that it takes 2 years to get profitable and you pay 120 per month (for the feeds you are talking about probably much more) traders are paying 3k over the time just for the feed.
Therefore people will be heavily tempted to start trading at a point where they surely will get eaten because they did not develop an edge yet.
Another point: There are many expensive feeds out there that give much less quality than feeds that are cheap.
Speaking of start-up: what are the costs involved?
- Renting data storage servers
OpenTick boasted 10TB of storage, thats about 10 servers.
Add another 10 servers for redundancy and overhead.
That's 20 servers at $200/month = $4000/month.
- An average consumer pulls ~0.5 MBit/sec during market hours,
let's say 100 clients per 100Mbit line. One unmetered 100Mbit line goes for about $1000/month, thats $10/month/client.
- One American guy at $4000/month, for monitoring the servers. One Ukrainian hard-core programmer $1000/month (may we have 2 at that price?). Additional expenses $1000/month.
- How much do you pay to exchanges for the data feed and rights to redistribute the data? Don't really know. Let's say $10000/month + $10/month/client (uneducated guess).
That sums up to $20k/month + $20/month/client.
So the breakeven point would be 250 clients paying $100 each monthly.
Does it look realistic?
- Renting data storage servers
OpenTick boasted 10TB of storage, thats about 10 servers.
Add another 10 servers for redundancy and overhead.
That's 20 servers at $200/month = $4000/month.
- An average consumer pulls ~0.5 MBit/sec during market hours,
let's say 100 clients per 100Mbit line. One unmetered 100Mbit line goes for about $1000/month, thats $10/month/client.
- One American guy at $4000/month, for monitoring the servers. One Ukrainian hard-core programmer $1000/month (may we have 2 at that price?). Additional expenses $1000/month.
- How much do you pay to exchanges for the data feed and rights to redistribute the data? Don't really know. Let's say $10000/month + $10/month/client (uneducated guess).
That sums up to $20k/month + $20/month/client.
So the breakeven point would be 250 clients paying $100 each monthly.
Does it look realistic?
Questions:
1. Do you want to follow 1000 or less stocks... ??
2. Big Question: Do you need EVERY single tick of every stock in real time to make your trade based decisions or would a snap shot value every second on every one of the 1000 or less stocks be sufficient... ??
If the answer to the two questions above is yes.. than you can pull bid ask quote or last trade data, quite inexpensively and every half second, from a DDE feed off an excel sheet into text files or a database...
using Reuters Lite DDE feed from their Equis division.
all for less than $190 a month...
just a thought...
You overestimated expenses for the feed price and servers rent, and underestimated the salary, but overall yes, 200-300 clients is a breakeven point. And I know that at least 1000 former clients of Opentick will jump on this deal, provided that it will give all (or more) goodies that OT had.
1. I would happily follow more than 1000 stocks, but there are not many stocks that trade frequently (one trade in 5 sec or more frequent). For my logic high frequency is essential.
2. Snapshot for every second would be sufficient (for now), but why limit yourself? From the data i collected, tick-by-tick is just ~30% more data than second-by-second (for heavily traded stocks tick-by-tick about is 10 times more data, but for others it's one trade in 5 sec, so ~30% is average number when spread over 1000 stocks).
3. DDE is a joke. Sorry, it's like driving skateboard to work after flying there in helicopter. Please, don't even mention DDE or excel to any programmer worth it's penny...
Did you look at gndt.com? They seem to offer unlimited feed + 2 days backfill for $200
1. You asked "Why Limit Yourself". You wanted to be cost effective that is a self implied limitation... otherwise there was no reason to post a possible solution to try to be helpful to you..
2. DDE is NOT a joke... only your opinion is a joke... if you want to get snapshot data very cheaply for 2000 or less stocks then DDE is a very inexpensive way to go... we have done it... it works just fine.. as long as we do not need real time tick by tick data to form a trading decision... Most DDE feeds are pure sh***t... agreed... not the Reuters Lite that is why i mentioned it...
We have a setup where we gather data every second or so on 2000 stocks using DDE for snapshot data decision... and we use less than 30% cpu...
again as i have mentioned more than twice... This is snapshot data every second that is very cost effective... otherwise, if you want real time data streams... then pay for it.