3 ms·
Gabriel, With a few months of back data at Sand Hill Exchange, we've found our results often shockingly accurate. For example, even before we experimented wit
by ghall 12y ago
Gabriel,
With a few months of back data at Sand Hill Exchange, we've found our results often shockingly accurate.
For example, even before we experimented with a profit incentive, our users forecast big increases in valuations on Slack (http://blog.sandhill.exchange/post/101804376238/sand-hill-exchange-tuesday-nov-4 http://blog.sandhill.exchange/post/101804376238/sand-hill-ex... ), Beepi, Mattermark, and Reddit, all of which proved true.
Our community also outperformed journalists: http://blog.sandhill.exchange/post/101679447223/parsing-the-predictive-power-of-tech-journalists http://blog.sandhill.exchange/post/101679447223/parsing-the-...
I won't tell you what the smart money has been betting on in the past few months, but it will be great to see how this pans out. Here's the January predictions of relevance: http://blog.sandhill.exchange/post/109527449373/january-2015-ipo-valuations http://blog.sandhill.exchange/post/109527449373/january-2015...
- scottkduncan 12y agoSo...insider trading then?