Back to Community
How would we know if Quantopian stole someone's algorithms?

I understand that Quantopian needs our source code in order to run it on their servers and I trust that they're honest about never looking at anyone's source code. My concern is, however, that if they did steal an algorithm, how would we know?

3 responses

Since we're speaking philosophically, if you can't know, why do you care?

Presumably, the hypothesis here is that your algorithm actually works, and that it would stop working if someone else started running it, at scale. The profitability could be "arbitraged away" or at least diluted. If it actually happened, and you suffered actual losses, and it would be worth the headache, you'd lawyer-up and go after Quantopian, get the authorities involved, etc. One can imagine Quantopian being audited, turning over records, etc. In the end, a "smoking gun" might turn up, showing that indeed your precious code had been stolen, and that you had suffered losses as a consequence (or at least that Quantopian had violated their terms of service...but if no actual damage was done, then the lawyers win, since presumably you'd need to pay them up front).

Or maybe the concern is that Quantopian's business model is not to pay users for their work, but to steal it. It is a systematic scheme to fleece the masses. I kinda doubt that's what's at play here. Quantopian needs diversification of ideas and talent. They need word to get out that Joe Sixpack (well, his equivalent with an advanced technical degree, perhaps) bought a sports car with his first check from Quantopian. They need to make full time (or extended contract) job offers to "managers" who have generated profits (my working theory, not something Quantopian has said, to my knowledge). Etc. My read could be wrong, but they are not looking to grab a handful of super-algos (which probably don't exist anyway, at scale) and disappear into their newly acquired mansions. Point72 needs someplace to park its $11B (or whatever the colossal sum is), and a few algos won't cut it.

There seems to be an alchemist mindset in this field, that I and only I have found the philosopher's stone. That I know something that nobody else has managed to discover (aside from insider information, which is different case altogether). In the end, it probably ain't true. Something else is at play. Either I'm deluded or the strategy is not scalable to a level that would be interesting to big-time institutional investors.

That is a great write up. However, where can I find an example of Mr. Joe Six Pack actually earning anything from Quantopian? Where is the list of successful authors who are making money from their algorithms. I just created one such algorithm, which turns 1 million into 26 million back tested, and even thrived during the market crashes, actually making money. The stats are below. How do I ensure, other than just applying for a copyright that Quantopian doesn't actually steal what I wrote and sell it to some slob on Wall Street? How does one go about actually receiving an allocation? Surely, we don't know one way or another, because there's no data to back this up. Can you point us to some examples of individuals who have successfully capitalized on their coding talent?

Total Returns
2578.89% - 26 million

Benchmark Returns
131.02%

Alpha
0.54

Beta
-0.24

Sharpe
2.47

Sortino
4.30

Volatility
0.21

Max Drawdown
-21.6%