3 ms·
As a former quant fund manager Netflix presents an interesting problem. If you read about the quant space, a whole load of people reckon they can use ML to make
by lordnacho 7y ago
As a former quant fund manager Netflix presents an interesting problem. If you read about the quant space, a whole load of people reckon they can use ML to make money.
The problem for them is much like for Netflix.
Having a way to predict the future is not the whole story of how to decide what to do.
Netflix having a whole load of data about what people watched, for how long, their demographics, etc, is probably very useful. But it is not an obvious step to go from knowing what people saw, even predicting it, to deciding what to produce and what to present to them.
Add to that the likely fact that people often want something fresh and innovative, and you have quite some stretched assumptions in statistical terms.
Not saying this is the end, I don't know enough about the business. Just that it's an interesting jump that I rarely hear anyone talk about.
- sgt101 7y agoWell, driving by using the rear view mirror isn't a wildly successful strategy. Many people are surprised by the difference between testing strategies for ML working to predict physical phenoma and complex systems like the markets; you can backtest market strategies to bits and it matters not even a bit, it means nothing.