New answers tagged

2

If you are asking these kinds of questions, it is very unlikely you will be able to successfully train any kind of ML or AI model on stock data with your current level of skill. In fact, each of the questions you ask usually involves weeks, months and -- in some cases -- years of experimentation to discover the right mix of potential features. In ...


3

Maybe this is too simple, but here’s what I think of when you ask for option strategies given a view on forecasted price densities: -Think returns are going to be higher than expected? Buy a call. -Think returns are going to be lower than expected? Buy a put -Think scale is going to be higher than than expected? Long straddle or long strangle. -Think scale ...


5

I hope I understood you correctly and that the following thoughts help you a bit. Reference point: Univariate curve fitting using splines With a univariate function $f(x)$ you can perform 1D spline interpolation and require for each (inner) $x_i$-node that: $$ \begin{align} \left.f_{i-1}(x)\right|_{x=x_i}&=\left.f_i(x)\right|_{x=x_i} \quad \mathrm{...


5

Take a look at compilations such as 151 Trading Strategies. I wouldn't expect this information to be widely disclosed. After all, a non-profitable strategy is a supermartingale which means there is an opposing set of algos that is profitable as we speak. Secondly, many strategies are conditional upon a market regime, and could become profitable should the ...


11

Is there a typical "half-life" of a strategy? This is a really subjective question, and I don't think any singular answer will generalize well. That being said, I will give some examples from personal experience. I have made hundreds of trading models in my career. I have only deployed 9 into live trading in the last ~25 years. Of those 9, 2 of ...


5

6 months is a reasonable estimate. Typically such strategies do not decay as in half-time but rather stop working in a discrete manner. Parameter tweaking can help, but for a limited period of time. To add to your list which is valid in itself: Infrastructure changes at the trading venue, in particular latency-related changes Market-making program changes ...


0

For true cross-validation you need $N$ separate calibration sets, for example one data set per year. You can then test performance on, say, the month of data following each calibration set. In feature engineering (FE), the essential problem is that if you systematize FE to "try many things", it becomes much harder to avoid the pernicious issues of ...


Top 50 recent answers are included