Amazon Finspace - Security and Trust

kmiklas

kmiklas

I've been looking at this as a platform for Data Analysis, but I have some doubts about sending my data sets into Amazon.

Data sets are the bedrock of ML, AI, and Data Science/Analytics in general. Data sets reveal much about a strat; just knowing what they are gives much away. Exponentially worse is actually _providing_ the data sets.

If you had taps set up all over the markets, on satellites, on the dark side of the moon; or you had proprietary data, like card usage info, bank account history, trade records, etc., would you happily pump these data into AWS? Let's not forget that when push came to shove, AWS shut down Parler's servers.

The tool looks great, but said data sets are both sensitive and expensive. I'm not in a rush to hand them over to Amazon.

Any thoughts? Do I have a tinfoil hat on, or is it a legitimate concern?

https://aws.amazon.com/finspace/

 
Last edited:
mukoh said:
Seriously doubt you have access to data that AMZN already doesn't have.
More...
Market and public data, fully agree. Non-market, proprietary, and other "weird" data, maybe not. Do you think they have, say, transaction data for all of American Express's customers? They do have data for all Amex purchases through Amazon.. but there it ends, no?

..but most importantly, it's not just the data, it's knowing the set of data sets that can be crunched to produce indicators. That's more then half the battle. They have a deluge of data, but finding the right sets to analyze to produce profitable indicators and information is tricky! That's where the creativity, inspiration, curiosity, and art of the Data Scientist come into play.

If I told you, "I use these ten data sets to produce profitable indicators," you can put those data sets under a microscope, find that edge, and think along the same lines that I think... and this is what you're doing if you piping a set of data sets into Amazon Finspace. Actually even worse, because not only have you given away your target data sets, but your methods, conclusions, and inferences based on the datasets are now available and visible within Amazon.

It gives me cold feet... I'd prefer to keep this stuff in-house. If I were analyzing more mundane data sets, like traffic data, bus schedules, weather data, or the like, then it wouldn't matter as much. Financial data are just too sensitive.

I think that's why they set up an entirely different set of servers--they're trying to show a concern for privacy. As the video says, Finspace is not accessed through the AWS console; it's a separate front-end access point. Not sure if it's just a façade for the AWS cloud service, or if they actually provision separate hardware. I can see why fintech companies are setting up their own clouds.

It's setting off my Spidey-sense.
 
Last edited:
Yes, you have a tinfoil hat on.

Amazon is a trillion dollar company that functions as a quasi-state, complete with its own bureaucracy and court system. Its market cap is greater than the GDP of 95% of the countries in the world. Your data set means nothing to Amazon.
 
Retief said:
Yes, you have a tinfoil hat on.

Amazon is a trillion dollar company that functions as a quasi-state, complete with its own bureaucracy and court system. Its market cap is greater than the GDP of 95% of the countries in the world. Your data set means nothing to Amazon.
More...
But what about the data set of a hedge-fund client?

If you were working for Citadel, would you recommend Amazon Finspace?
 
kmiklas said:
Market and public data, fully agree. Non-market, proprietary, and other "weird" data, maybe not. Do you think they have, say, transaction data for all of American Express's customers? They do have data for all Amex purchases through Amazon.. but there it ends, no?

..but most importantly, it's not just the data, it's knowing the set of data sets that can be crunched to produce indicators. That's more then half the battle. They have a deluge of data, but finding the right sets to analyze to produce profitable indicators and information is tricky! That's where the creativity, inspiration, curiosity, and art of the Data Scientist come into play.

If I told you, "I use these ten data sets to produce profitable indicators," you can put those data sets under a microscope, find that edge, and think along the same lines that I think... and this is what you're doing if you piping a set of data sets into Amazon Finspace. Actually even worse, because not only have you given away your target data sets, but your methods, conclusions, and inferences based on the datasets are now available and visible within Amazon.

It gives me cold feet... I'd prefer to keep this stuff in-house. If I were analyzing more mundane data sets, like traffic data, bus schedules, weather data, or the like, then it wouldn't matter as much. Financial data are just too sensitive.

I think that's why they set up an entirely different set of servers--they're trying to show a concern for privacy. As the video says, Finspace is not accessed through the AWS console; it's a separate front-end access point. Not sure if it's just a façade for the AWS cloud service, or if they actually provision separate hardware. I can see why fintech companies are setting up their own clouds.

It's setting off my Spidey-sense.
More...

As long as the data sets are from the financial market and about the financial market and do not involve any personal and private information, I am fine with that. After all this is what every single trading software, trading platform, data software already have access to. The only difference is Amazon is allowing for a broader type and spectrum of data and is providing analysis from a data point of view and not from a chart or graphic point of view.
 
Retief said:
Yes, you have a tinfoil hat on.

Amazon is a trillion dollar company that functions as a quasi-state, complete with its own bureaucracy and court system. Its market cap is greater than the GDP of 95% of the countries in the world. Your data set means nothing to Amazon.
More...

Amazon has abolished the arbitration system btw. Now if you have a dispute, you have to sue the other party using the regular judiciary system.
 
JSOP said:
As long as the data sets are from the financial market and about the financial market and do not involve any personal and private information, I am fine with that. After all this is what every single trading software, trading platform, data software already have access to. The only difference is Amazon is allowing for a broader type and spectrum of data and is providing analysis from a data point of view and not from a chart or graphic point of view.
More...
But that's the point... some of the best indicative data sets _about_ financial markets come from _outside_ financial markets.

Most, if not all, arbitrage contained _within_ the set of market data is gobbled up by the hedge funds in microsecond time before it even hits most screens.

How do you think Citadel pulled in a record 6.7B in revenue last year? [1] They're the first to market on all the COVID-related volatility.

idk... I just don't feel comfortable piping proprietary data sets into Amazon.

1. https://www.bloomberg.com/news/arti...s-reaps-record-6-7-billion-year-on-volatility
 
kmiklas said:
But that's the point... some of the best indicative data sets _about_ financial markets come from _outside_ financial markets.

Most, if not all, arbitrage contained _within_ the set of market data is gobbled up by the hedge funds in microsecond time before it even hits most screens.

idk... I just don't feel comfortable piping proprietary data sets into Amazon.
More...

Yeah but if Amazon is not in the business of financial trading or management, then it's not that bad but if Amazon one day wants to open its own brokerage or financial management firm, then yeah I would have problems with it. Right now it looks like it's just trying to be a data software or trading software company like MultiCharts or eSignal. We have no problem with these companies having access to financial data and MultiCharts is owned by Russians. LOL
 
You must log in or register to reply here.
Back
Top