Thursday, May 18, 2017

From alpha to actual returns: Why your mileage will vary

One of the characteristics of a good financial modeler is to know his model's limitations. He know how and why they work, and under what circumstances they will fail.

I have been asked a number of times in the past to disclose the returns of my trading account, or the signal dates of my trading model, whose out-of-sample buy and sell signals are shown below. I have resisted those requests because the disclosure of the data leads to a false level of precision in return that readers cannot expect.

The process of turning a forecast alpha, or signal, into actual portfolio return is a tricky one. There are three decisions that have to be made:
  1. What and when do you buy and sell?
  2. How much do you buy and sell?
  3. How do you time the trade?
The "arrows" in the above chart only answer question 1.

The answer to the second question depends on the trader's personal circumstances. What is his risk tolerance, or pain threshold? What are the tax consequences of the trade? How much conviction does he have in the signal? Should he scale into a position? If the trade turns against him, how does he manage his risk, or should he average down? Conversely, should he take partial profits if the trade moves in his favor?

I know nothing about my readers. I know nothing about the answers to any of the above questions. If I was a portfolio manager of an actual fund that uses this strategy, then there would be disclosure documents specifying the kinds of risks an investor can expect, as well as the tax implications of investing in the fund.

The full post can be found at our new site here.

Our Sale in May event
Announcing our "Sale in May" event! Get $1 off plus an extra month free off the first year of an annual subscription*. Use the coupon code May2017 at checkout.

* Offer is only available to the first 100 to sign up and expires May 31, 2017. Subscription date extension will be made after order processing.

No comments: