How do we avoid systematic bias in the construction of the optimal risk budget allocations? about investment-funnel

[mikkelbechmogensen]

Should we try to get more data, and then make a sampling/estimating split of the data, where we sample on 70% of the data and estimate parameters on 30% of the data? In order to make it robust we could try to sample on the 30% of the data with the worst performance.

Look into FHS method of generating scenarios with a GARCH volatility model.

[VanekPetr]

I don't think we need more data at this point. Let's try/discuss the split on the 10y of data we already have.

[VanekPetr]

@mikkelbechmogensen Let's talk with Kourosh about it next week.