Hi!
In the last few of months I have been experimenting with QuantConnect and creating some algorithms. Currently I have two algorithms that I think that could be interesting to publish in the alpha market. I was investigating what was neccesary to get the algorithms published.
One of the requirements is emitting insights, I could implement this but than it also seems nessecary to specify the 'insight period'. I could statistically approximate this however if I knew this I was a millionaire so I will always be wrong. My strategy is in essence a pair trade that lasts between 5 minutes and 5 days.
Questions:
- Why is this a required field?
- What is the impact if I put complete nonsense in this field?
- Will is my alpha still valuable if I cannot approximate insight period with some degree of accuracy?
best,
Dirk
Adam W
The Insight Period can be useful for funds (or your own algorithms) that systematically combine different Alpha signals. If your Insight predicts a symbol will go up with a certain magnitude, you can quite easily create a decent entry/exit-based trading logic around it. However as funds can't see the source code and can only work with what is given, there's not really a way for them to differentiate between long-term and high frequency signals.
Would strongly caution about arbitrarily setting the period. Depending on the portfolio construction model, when insights "expire" (at the end of the period) the portfolio rebalances and essentially liquidates the symbol. If your intention is to hold the position until some condition happens, this can cause it to liquidate when it shouldn't and throw everything off. There's a Portfolio implementation by Jeffrey here that you can use as a template for preventing this (i.e. insights acted upon only once). It can also mess up your Alpha scores.
Of course, but I think you should try your best to think about the underlying logic and figure out some way to roughly approximate it. One of my recent strategies is based on a similar entry/exit signal-type logic, and I store the durations for previous entry/exit signals and do something with them. As more data comes in, the Period converges in accuracy.
Dirk bothof
Cool! Thanks for the insights, it seems that I need to brew on this to make an initial model : )
Dirk bothof
The material on this website is provided for informational purposes only and does not constitute an offer to sell, a solicitation to buy, or a recommendation or endorsement for any security or strategy, nor does it constitute an offer to provide investment advisory services by QuantConnect. In addition, the material offers no opinion with respect to the suitability of any security or specific investment. QuantConnect makes no guarantees as to the accuracy or completeness of the views expressed in the website. The views are subject to change, and may have become unreliable for various reasons, including changes in market conditions or economic circumstances. All investments involve risk, including loss of principal. You should consult with an investment professional before making any investment decisions.
To unlock posting to the community forums please complete at least 30% of Boot Camp.
You can continue your Boot Camp training progress from the terminal. We hope to see you in the community soon!