80 reputation
6
bio website
location
age 36
visits member for 3 years, 5 months
seen Oct 13 at 23:12

Oct
16
comment Statistical significance of trading systems that use indicators with long lookbacks
I don't think he's talking about quantity of backtesting data, but instead the timespan over which each value of the indicator is taken (the 'lookback'). If this lookback timespan is comparable in size to the backtesting timespan, then there will be strong serial correlation in the indicator.
Sep
7
comment Computing FX forward delivery dates
Great, thanks Phil
Sep
6
comment Computing FX forward delivery dates
Thanks Phil, I agree there are a lot of factors to consider. However, I was trying to ask: calendar issues and such-like aside, are delivery dates for short dated outrights (say, 1W) computed by 1) adding a week to today, and then adding T+2 business days, or, 2) adding a week to the spot value date?
Jul
1
comment Optimality of Kelly criterion in non-normal environment
That's not an issue, it generalizes easily to a continuous distribution. The danger lies instead in not knowing the precise distribution of your returns. There's an excellent summary of the good and bad properties of the Kelly Criterion here: edwardothorp.com/sitebuildercontent/sitebuilderfiles/…