366 reputation
21
bio website
location
age
visits member for 3 years
seen Oct 19 at 16:55

Oct
19
answered If I am very fast (less than 10 microseconds latency) what would be the first strategy to execute?
Nov
25
awarded  Yearling
Jan
3
awarded  Necromancer
Dec
3
answered Can the concept of entropy be applied to financial time series?
Dec
3
comment Algorithms for predicting a couple points in the future
You're working on the Kaggle competition, right? Try SVM on the whole series before the spike (so each observation is just a feature)
Dec
1
answered What exactly is meant by “microstructure noise”?
Nov
30
answered References for developing an automated trading system?
Nov
30
comment How to estimate the probability of drawdown / ruin?
I am interested, please do tell :)
Nov
30
comment How are risk management practices applied to ML/AI-based automated trading systems
I think you are both right, the difference is one of degree. Anything you do involving any historical does introduce bias ... even a model you decide not to use because it performs horribly on out of sample backtests has introduced a bias by removing one bad model from your search space. However, the assumption underlying an ML based strategy is that you can become biased toward a better model. Lirik's question is about controlling the risk this introduces, rather than eliminating it totally.
Nov
28
awarded  Teacher
Nov
27
answered How are risk management practices applied to ML/AI-based automated trading systems