21

Inspired by this question, and I don't know much about insider trading beyond it.

The title pretty much says it all. Let's say I make software that picks up on patterns of insider trading. (For example, short sales that happen just before big news comes out that causes people to trade). I use this software to identify insider trades and make the same trades. Have I committed a crime?

Rodrigo de Azevedo
  • 1
  • 8
  • 22
progressiveCavemen
  • 313
  • 2
  • 6
  • 12
    Legal advice from strangers on the internet is a baaaaad idea. Ask a lawyer. – ceejayoz Oct 30 '15 at 17:57
  • 9
    Oh trust me, if I was really planning on doing this I wouldn't be using stackexchange as legal counsel...As fantastic as this site is. – progressiveCavemen Oct 30 '15 at 18:04
  • 3
    Good luck. Don't think you're the first to try, and the competition is better funded... but if you think you can square this circle quickly and reliably enough to do yourself any good, go for it. – keshlam Oct 30 '15 at 19:01
  • 4
    Just a thought before you attempt this: You might be better off running a simulation of your software's picks for a few years to see if the strategy actually works to create a better return than the market indexes. – JohnFx Oct 30 '15 at 19:32
  • 11
    Anyway, if you can reliably identify "short sales that happen just before big news comes out" before the news actually does come out, I suspect you'll find plenty of ways to get rich on that trick alone. :-) – Ilmari Karonen Oct 30 '15 at 22:20
  • 2
    @JohnFx You can do a similar experiment without having to wait a few years for the results, as long as you're able to simulate the algorithm on historical data. When developing and tweaking the algorithm, only use data for a portion of the possible stocks (say, half), chosen randomly. Once you're pretty sure it's good, try it out on the other half. (If it fails and you have to start over, make sure you randomly split them in half again, rather than re-use the old halves.) –  Oct 30 '15 at 23:08
  • 2
    @IlmariKaronen I think the idea is to look for news items like "big announcement next Monday!," or an anticipated quarterly sales data release, etc., along with an insider shorting a whole bunch. You use that info to conclude that the "big announcement" is probably something negative. –  Oct 30 '15 at 23:09
  • Let's say the OP identified specific Twitter/Facebook/LinkedIn accounts of some public company's officers or employees and data mined those. General conclusions from the ratio of positive to negative words or phrases: insider trading or not? – user662852 Oct 31 '15 at 01:32
  • 1
    @user662852: Probably not. By definition, "insider trading" relies on information not available to the general public. If you can get reliable company information from a social media account, so can everyone else. A public tweet or Facebook post is, by definition, not insider information. On the other hand, if these are private tweets or posts--and the company is using a higher ratio of positive words in public statements--then your data mining might be considered insider trading. – Bruce Alderman Oct 31 '15 at 02:25
  • @BruceAlderman: placing trades on the basis of whether your CEO golf buddies are chewing their nails or not this week. Insider trading or not? ;-) – Steve Jessop Oct 31 '15 at 12:47
  • 2
    "Beating the market" like this is really hard. – Kevin Oct 31 '15 at 14:42
  • @SteveJessop Depends on their golf score. – Bruce Alderman Oct 31 '15 at 14:58
  • @BruceAlderman: It can't be quite that cut and dried. In an extreme case, consider an insider encoding market orders to his broker (or friends or family) in public social media posts using steganography. In such a case, the information wouldn't be public even though the posts containing it are. – Ben Voigt Feb 23 '19 at 23:22
  • @BenVoigt but that's a completely different scenario from either what user662852 described above, or what the OP is asking about. You're not going to be able to data mine information encoded via steganography. And again, it only matters if the encoded information contradicts publicly available information, in which case the insider wouldn't want to post the evidence in a public place. – Bruce Alderman Feb 25 '19 at 19:29

3 Answers3

37

A cautious "no." Public information is just that, public. If you are able to data mine and use that information to create a buy signal, you are legally able to trade on that signal. Since nothing is ever 100%, your signals would likely just exhibit a very high correlation to stocks being traded on inside information. Regardless of why those trades happened, the fact that they did is public, milliseconds after the trade.

Cautious, because you might still be investigated. As long as you have no ties to the real insiders at these companies, and are genuinely running software to create these signals, you should be fine.

JTP - Apologise to Monica
  • 172,273
  • 34
  • 296
  • 560
  • 5
    Obligatory remark that the software you're describing boils down to predicting the future and would be essentially impossible to create. – Eric Oct 31 '15 at 01:31
  • 7
    @Eric: I fail to see how this is "predicting the future". OP's hypothetical software would be looking for patterns in historical data even though that "historical" data might be only milliseconds old. No "prediction" here - just pattern analysis. From my point, however, I think it's difficult because to know that it's "insider trading" you'd have to know what was in the mind of the person doing the trading, which could prove difficult. Of course, if you track the activities of potential insiders and their contacts you might be able to pick up something. Best of luck. – Bob Jarvis - Слава Україні Oct 31 '15 at 02:37
  • It sounds to me like a program that basically copies what insiders do if those insiders have previously shown a pattern of trades that predate relevant news. I doubt the insiders would be stupid enough to leave enough of a pattern that you could find it! – Loren Pechtel Oct 31 '15 at 04:16
  • 3
    Phrasing is essential. If you say you are "predicting insider trading," your words could be more trouble than if you instead say, "detecting behaviors comparable with those of insider trading," the words no longer suggest that, at the end of the day, you are aware of insider trading, merely behaviors that line up well with what insider traders do. – Cort Ammon Oct 31 '15 at 10:21
  • 1
    @LorenPechtel: but insider traders don't know what patterns the OP can or can't detect with "sufficiently high confidence" to make a profit on. They aren't stupid enough to leave patterns the SEC can detect with sufficiently high confidence to investigate (or, if they are, they won't remain insider traders for long). But the whole fun of "big data" is that it's incredibly difficult for the person whom the data concerns, to know what you'll be able to deduce from it ;-) – Steve Jessop Oct 31 '15 at 12:52
  • @SteveJessop If you can detect it so can the SEC. – Loren Pechtel Oct 31 '15 at 23:25
  • The real question is, if I hire Professor X to tap into the minds of people to help me get data for trading, is that insider trading? – yuritsuki Oct 31 '15 at 23:46
  • 1
    @LorenPechtel: with different required levels of confidence to do anything with those people detected as potentially doing it. There are plenty of areas other than market trading where researchers have identified groups of people who are more likely than average to be committing a crime, but where the authorities can't or don't act despite being in principle able to gather (or being given) the same data. OP doesn't need definite insider traders for this hypothetical scheme, just a smallish group some of whom are at it. – Steve Jessop Nov 01 '15 at 02:26
4

Nothing at all wrong about that. Be careful that especially before big news, it's rather common that there are huge swings because (a) the market makers pull/widen their quotes, (b) others already do what you are describing and (c) other actors try to exploit the situation by moving the market in one direction chosen at random and taking the profit before the actual news hits, so it doesn't even matter to them which side they bet.

If you spot your pattern you can't know for certain if the cause is actual insider trading or (c) amplified by (b). You don't know, so you take an educated guess and make a bet. That's trading, not insider trading.

Edit: It would possibly be a slightly different case if you could write software that identifies insider trades with a 100% guarantee, which nobody else has been able to do so far. In that case sell it to market regulations or the DOJ for a few millions and retire early and risk free.

Peter
  • 2,113
  • 1
  • 10
  • 11
0

I think the question hangs on "am I insider trading", because the answer depends on who you ask. To a human jury of your peers, the answer is probably "no". You could make a strong case that mimicry is not the same as insider-trading. But to any program that detects insider trading, including your own, the answer would obviously be "yes" because that is essentially what you are trying to do. Mimic insider trading.

And I can foresee this going horribly wrong...

Any programmer worth their salt will make sure to blacklist themselves from insider-trading-detection, else you will create a positive-feedback loop of faux-insider-trading. But unfortunately you cannot blacklist other mimics - after all they look and act just like the real thing. Then a human on any of those bot's lists of insider-traders could control all the bots in a very predictable way.

J.J
  • 101