Holding a trade for months seems like a long time to me. But I suppose if you're trading in size, then you need an ultra-stationary spread that is worth waiting for to "ripen". I'm glad you used ED in your example chart. I think the ED is the most flexible contract for spread trading. You have about 15 contracts in the first five years with decent volume. You can dial up any combination of directional volatility and convexity flex that suits you.
I suppose it's somewhat less likely in a struggling stock, but what about a takeover? It's the short side of these strats that scares me (granted by trading a basket you are mitigating the damage via smaller single name position size). Not really an issue for equity index vs. equity index, but I suspect there's not nearly as much juice there. Never studied what the price shocks look like in commodities, but I would guess they're somewhere in between.
Always that risk, too. My feeling is that the single-stock event risk is not symmetrical, with the downside scaring me more. It's a judgement call, not to be defended but simply explained.
In case it's interesting to anyone, I ran a quick study on overnight equity moves > 25% (survivorship bias free database) for stocks that were trading > $5 and had a 10 day moving average for volume > 500,000 before the large move...here's what the bins look like 1985 - Present: Bin : Count +200% to +300% : 1 +100% to +200% : 17 +75% to +100% : 19 +50% to +75% : 68 +25% to +50% : 672 -25% to -50% : 955 -50% to -75% : 112 -75% to -100% : 18 Hope that's not too hard to read...also, it wouldn't shock me if a couple of these were splits that the data provider missed, but they shouldn't be. And actually the counts look a bit low to me, but who knows....regardless, it should give some idea of relative frequency.
Thanks for running these numbers. Results are consistent with the put/call skew often seen in equity options.
Months .. then transaction costs may not be an issue.. but carry will be .. plus or minus. Months .. hey Martinghoul -- what do you do between trades? Other than posting on message boards Months .. along the same lines, "general" color on exit criteria could be interesting. I assume you're not a mechanical bounce-off / return-to kind a guy.
Looking at my statements for this year, I can see that my average holding timeframe for Eurodollar Calendar Spreads (Z1-Z2 particularly) was 14 trading days. A Nymex or ICE crack spread might be 30 minutes up to three hours. As of this morning, the August RBOB Crack had a 202 tic average twenty day trading range and a twenty day hisorical volatility of 93 %. Obviously you want to trade that beast differently than a Eurodollar Calendar that has a seven tic daily trading range. For me, at least, the volatility and trading range of the particular spread determines the holding period. We set profit targets and stop-loss levels at the time of entry. As an example, I bought the U1-Z1 Heating Oil Calendar Spread last Wednesday morning on July 13 and it is marking up nicely but we are still a fair bit away from the target levels.
Hahaha... I got fingers in lots of pies, mate, so there's stuff to keep me occupied (other than ET, that is). Also, months, I'd say, is the upper bound and it could be days or weeks. As to exits, I am generally a very RV oriented person. So I don't have stops (unless I judge that there's been a regime change) and I normally add to trades if they go against me. Targets is where it gets interesting. When I look at trades to decide whether to put them on or not, there's one measure that encapsulates everything, the ex-ante Sharpe. Since the expected return goes into the calculation, my targets are defined at the time I get into the trade. I can tell you more about the calculation of the ex-ante SR, if you're interested, but it's not exactly 100% rigorous, mechanical and systematic (it's a somewhat more formalized way of doing what bone mentions above). However, it has served me well, with the caveat that I believe that my mkt lends itself better to mean-reversion types of trades than others (constrained distributions and all that jazz).