7 ms·

What a lame way to insider trade. I thought that instance of the Capital One analysts using internal database queries to understand which retailers were having

by extr 5y ago

What a lame way to insider trade. I thought that instance of the Capital One analysts using internal database queries to understand which retailers were having good quarters was far more interesting and insidious. Similarly you could imagine a lot more subtle ways to leverage insider Twilio information (count of verification texts sent by client?) to make far more obscure bets. The SEC definitely takes a look at anyone who goes from zero trading -> making improbably great trades worth millions though, so better have a good patsy lined up! Edit: Context for the traders I was talking about. What they did was much cooler and honestly kind of impressive: https://www.bloomberg.com/view/articles/2015-01-23/capital-one-fraud-researchers-may-also-have-done-some-fraud https://www.bloomberg.com/view/articles/2015-01-23/capital-o...

RC_ITR 5y ago
>count of verification texts sent by client This is what I thought it would be based on the headline and was frankly disappointed that the truth was so quotidian. Certain Twilio engineers could very likely keep an index of major gig-economy/delivery services and track both user and driver sign-ups in a way that could be very valuable NMPI and have the advantage of being at a broad range of companies that are only tangentially related to their employer. In fact, @SEC if you're reading this, you might be missing some bigger fish...
kyleblarson 5y ago
Add to that they apparently chatted about it in some tool that had an auditable history and one can surmise that these people weren't the most sophisticated bad actors.
- MengerSponge 5y ago
  "The stock will go up for sure." Swiper, no swiping!
ramesh31 5y ago
>I thought that instance of the Capital One analysts using internal database queries to understand which retailers were having good quarters was far more interesting and insidious. How is this seriously different than what any one of a million other "AI analytics" companies are doing with customer data right now?
- tptacek 5y ago
  Because it was Capital One's internal data, which belongs to the shareholders. US insider trading laws are about theft, not about fairness (the markets require "unfairness" in this sense in order to perform price discovery).
- bilbo0s 5y ago
  The "AI analytics" companies are all communicating expressly their intent when they purchase the data at the outset. This makes it legal. Full disclosure, I'm a privacy advocate and believe this sort of selling of customer data should not even be allowed. That said, what the companies you allude to do is totally legal by law, and significantly different in practice than taking data without informing anyone of your intent. I guess the short answer to your question is that this is the United States. Everything is legal here, provided you do the paperwork first.
- maxerickson 5y ago
  It's discussed in the link they added. The retailers shared the data with Capital One for specific business purposes, not for stock market analysis.
- jeromegv 5y ago
  It was an internal database (from Capital One credit cards), they weren't allowed to use that data for analysis. This wasn't publicly accessible information. See article: https://www.bloomberg.com/opinion/articles/2015-01-23/capital-one-fraud-researchers-may-also-have-done-some-fraud https://www.bloomberg.com/opinion/articles/2015-01-23/capita...
- FateOfNations 5y ago
  The difference is public vs. non-public information. Doing something like counting cars in retailers parking lots is something any investor could do (or pay someone else to do). Trading using trends from non-public credit card transaction data isn't something that any "general public" investor could do on their own. Also note this is about buying and selling securities, not selling consumer data for marketing purposes.
jliptzin 5y ago
Yea, now just imagine how many smarter people are doing it and getting away with it
- kasey_junk 5y ago
  Not many? Describe the trade steps necessary to insider trade with high leverage that isn’t trivial for a data analytics program to flag with high confidence?
  - extr 5y ago
    I don't find it hard to imagine. 1. Work for a company that provides signal-giving infrastructure for a variety of companies across disparate industries. Twilio is actually a perfect example. 2. Find someone to work with that already has or is willing to build a plausible history of so-so trades. 3. Occasionally tip this person off that company Y in industry X is about to have a good quarter. Never trade the same company twice. Parallel construct a plausible rationale for the trade based on public market signals if you want to sleep well at night. 4. Don't communicate over easily subpoenaed channels, obviously. The SEC isn't the NSA, basic OPSEC is fine. 5. Profit? That's why I'm so disappointed by the OP! It's so easy to dream up a situation where they used their access in an effectively untraceable way.
    - kasey_junk 5y ago
      That provides extremely low edge for both parties involved. Your participant with a history of trades could just as easily do research using public information (company press release aggregation is a classic version) to get similar edge without entering a criminal conspiracy. Notice I mention high leverage. There are 2 challenges (as I see it) to committing insider trading. The first is having information ahead of time that actually predicts how the market will move based on the information. It’s not enough to know a company will have a good quarter you need to know it will be better than expected. Second you need to be able to make outsized returns on that information. You have to beat the market returns by themselves. This compounds with the other issue above. The less obviously market moving data you have the more leverage you need. This narrows the search space pretty considerably for an analytics based approach. You look at insiders with lots of access or anyone making high leverage bets. Finally, do a mental exercise on your step 2. What does that even look like? An insider randomly approaching hedge funds? A random encounter at a bar? A frat reunion where they share each other’s trading history like term papers?
denimnerd42 5y ago
sounds like they went back to China and got new jobs. https://www.crunchbase.com/person/bonan-huang https://www.crunchbase.com/person/bonan-huang
- trhway 5y ago
  absolutely honest :) "He used to work for United States JP Morgan Chase, Capital One, Jimu.com (as Strategy Analytics Director). Bonan mainly focusing on data mining, machine learning and programming large data quantification in analysing stock."

by extr 5y ago

RC_ITR 5y ago

>count of verification texts sent by client This is what I thought it would be based on the headline and was frankly disappointed that the truth was so quotidian. Certain Twilio engineers could very likely keep an index of major gig-economy/delivery services and track both user and driver sign-ups in a way that could be very valuable NMPI and have the advantage of being at a broad range of companies that are only tangentially related to their employer. In fact, @SEC if you're reading this, you might be missing some bigger fish...

kyleblarson 5y ago

Add to that they apparently chatted about it in some tool that had an auditable history and one can surmise that these people weren't the most sophisticated bad actors.

MengerSponge 5y ago
"The stock will go up for sure." Swiper, no swiping!

ramesh31 5y ago

>I thought that instance of the Capital One analysts using internal database queries to understand which retailers were having good quarters was far more interesting and insidious. How is this seriously different than what any one of a million other "AI analytics" companies are doing with customer data right now?

tptacek 5y ago
Because it was Capital One's internal data, which belongs to the shareholders. US insider trading laws are about theft, not about fairness (the markets require "unfairness" in this sense in order to perform price discovery).
bilbo0s 5y ago
The "AI analytics" companies are all communicating expressly their intent when they purchase the data at the outset. This makes it legal. Full disclosure, I'm a privacy advocate and believe this sort of selling of customer data should not even be allowed. That said, what the companies you allude to do is totally legal by law, and significantly different in practice than taking data without informing anyone of your intent. I guess the short answer to your question is that this is the United States. Everything is legal here, provided you do the paperwork first.
maxerickson 5y ago
It's discussed in the link they added. The retailers shared the data with Capital One for specific business purposes, not for stock market analysis.
jeromegv 5y ago
It was an internal database (from Capital One credit cards), they weren't allowed to use that data for analysis. This wasn't publicly accessible information. See article: https://www.bloomberg.com/opinion/articles/2015-01-23/capital-one-fraud-researchers-may-also-have-done-some-fraud https://www.bloomberg.com/opinion/articles/2015-01-23/capita...
FateOfNations 5y ago
The difference is public vs. non-public information. Doing something like counting cars in retailers parking lots is something any investor could do (or pay someone else to do). Trading using trends from non-public credit card transaction data isn't something that any "general public" investor could do on their own. Also note this is about buying and selling securities, not selling consumer data for marketing purposes.

jliptzin 5y ago

Yea, now just imagine how many smarter people are doing it and getting away with it

kasey_junk 5y ago
Not many? Describe the trade steps necessary to insider trade with high leverage that isn’t trivial for a data analytics program to flag with high confidence?
- extr 5y ago
  I don't find it hard to imagine. 1. Work for a company that provides signal-giving infrastructure for a variety of companies across disparate industries. Twilio is actually a perfect example. 2. Find someone to work with that already has or is willing to build a plausible history of so-so trades. 3. Occasionally tip this person off that company Y in industry X is about to have a good quarter. Never trade the same company twice. Parallel construct a plausible rationale for the trade based on public market signals if you want to sleep well at night. 4. Don't communicate over easily subpoenaed channels, obviously. The SEC isn't the NSA, basic OPSEC is fine. 5. Profit? That's why I'm so disappointed by the OP! It's so easy to dream up a situation where they used their access in an effectively untraceable way.
  - kasey_junk 5y ago
    That provides extremely low edge for both parties involved. Your participant with a history of trades could just as easily do research using public information (company press release aggregation is a classic version) to get similar edge without entering a criminal conspiracy. Notice I mention high leverage. There are 2 challenges (as I see it) to committing insider trading. The first is having information ahead of time that actually predicts how the market will move based on the information. It’s not enough to know a company will have a good quarter you need to know it will be better than expected. Second you need to be able to make outsized returns on that information. You have to beat the market returns by themselves. This compounds with the other issue above. The less obviously market moving data you have the more leverage you need. This narrows the search space pretty considerably for an analytics based approach. You look at insiders with lots of access or anyone making high leverage bets. Finally, do a mental exercise on your step 2. What does that even look like? An insider randomly approaching hedge funds? A random encounter at a bar? A frat reunion where they share each other’s trading history like term papers?

denimnerd42 5y ago

sounds like they went back to China and got new jobs. https://www.crunchbase.com/person/bonan-huang https://www.crunchbase.com/person/bonan-huang

trhway 5y ago
absolutely honest :) "He used to work for United States JP Morgan Chase, Capital One, Jimu.com (as Strategy Analytics Director). Bonan mainly focusing on data mining, machine learning and programming large data quantification in analysing stock."