Thank you for this clarification. In most cases have the contracts which I'm trading only a one tick bid/ask spread. So there is not much to be gained.
Actually, spreads are likely to be even tighter. For example, VIX futures tick size is 0.05 while calendar spreads trade with a tick size of 0.01.
Yes, bid-ask is what I meant. On fair value: even if you think that the current contract you’re exiting is ‘too cheap’, the next contract you’ll be entering into is likely to be be ‘cheap’ too, isn’t it? If this works reliably, it is an independent signal I’d argue.
No, the spread can be trading either rich or cheap to fair value. For example, S&P 500 futures roll has a certain fair value based on the dividend amounts/dates and the interest rates. Imagine that when it comes the time to roll, the actual calendar spread is trading at a different level due to the markets pressure. Let's say this time around everyone is rolling short so the spread is trading cheap to fair. If you have to roll short as well, you can decide to forego the roll, let it expire and re-establish the position once the rolling pressure has passed.
Interesting phenomenon, tx for explaining, although I’m surprised by the fact that ‘everyone is rolling short’ as futures are in zero net supply. I’d also imagine that this is an ‘enforceable arbitrage’, since dividend futures and interest rates can be traded to lock in any mispricing. Is this big enough in your experience to base your decision on whehter to roll or not on? And where do you get data on fair value?
Not everyone rolls. So the spread movement depends on the balance of supply/demand between rollers and non rollers. GAT
Hi all, when evaluating trading strategies, I typically look at mean, variance, skew, kurtosis, among other things. When working with daily returns, it is easy to annualise these (*252). Same goes for standard deviation (*sqrt(252)). How about skew? Does anyone have any thoughts on annualising skew?
When calculated correctly skew is a standardised measure so doesn't need annualising. γ = E([{rt – m}/ s]^3) where y is skew, E is the expectations operator, rt is the return, m is the mean, s is the standard deviation (sorry it looks better in greek) GAT
@AvantGarde this is a nice probability review with market focus https://faculty.washington.edu/ezivot/econ424/probreview.pdf, skew formula above pretty printed on page 25