Monroe Trout, of New Market Wizards fame said in his interview that either a position/signal is valid and should be given full size, or not at all. How can we apply this to our own portfolio, if at all?
10 positions x 10% (or fixed number)
When I first started building a (concentrated) portfolio, I followed this advice to find the top ten stocks I liked and buy them. Bought such disparate stocks as NEU and BIDU, and sizing was easy (especially with a small account).
The beauty of this approach was its simplicity - buy 10% of the total equity and focus on the research behind the position. Maverick Capital (a Tiger Cub) and Kynikos/Chanos (short seller) both limits to 5% of portfolio.
However, its disadvantages are multiple: 1) lack of accounting for volatility. BIDU moves 5% a day often, while bond (etfs) moves are quoted often in 1/100 of a 1% (i.e. one basis point). 2) arbitrary portfolio sizing. Why 10% or 5%? 3) no relation to conviction/forcing poor positions. If you have only 5 good ideas (for 10x10), you either have to be happy with 50% exposure or buy positions you don't want
A more dynamic (and tougher approach): 5-10-20%
Now, I typically start with 5%, and increase in the next week or two in accordance with conviction and volatility. This is a scale-in/out strategy that aims to smooth out new/old positions, but plenty more to think about.
But this is a tough issue indeed, and especially for large portfolios ($100mm+).
Could you scale sizing by volatility? So you want each position to be 10% of the "risk", however you define it, of your portfolio. This could be simply beta adjusted delta or something more complicated.
ReplyDeleteYou could put more "risk" on your higher conviction trades like you mention and still scale into positions.
Yes, totally agree with that approach - that's the direction I've been heading in and have scaled into a volatility-adjusted as well as a risk-to-thesis position size.
ReplyDeleteThe latter tries to estimate how much price movement there could be if I am wrong on the thesis or risks materialize - (e.g. if aig's core insurance business falters on eps miss and the discount to book is justified, as happened last week). From earlier this year, we saw aig fall from 35 to 27 and a 20-25% drop from high is possible on such a miss.
For a 10% position, that'd mean a 2.5% fund loss, even if the year-long thesis is intact. Can then work backwards to say if I only want to risk 1% drawdown on such a risk, would only want a 4% = 1/2.5*10 position?
This is highly subjective and prone to recency-bias etc, but has been good mental exercise for me so far even in the end push the size to higher/lower level.