When analyzing potential spread trades I like to use Online Options Profit Calculator http://www.optionsprofitcalculator.com/calculator/iron-condor.html as my broker's calculator is not very good, but I've noticed problems with it. E.g., theta (time decay) on weekend days are shown as if they are equivalent to weekdays; and while there should be more decay over a weekend than overnight, weekend gaps versus daily overnight gaps are not so much greater as to justify two full days and night's worth of theta - I wouldn't think it's equivalent to even a single day's worth of decay, and my experience seems to bear this out. The example in the link shows an enormous time decay of $158 over the coming 3 day weekend. Does anyone here have any rule of thumb for estimating what it really should be and how to adjust the expected decay for the subsequent days that are also affected by this initial distortion? Are there better calculators which adjust for this difference in theta for times when the market is closed?
If it's calculating theta the way you described it's an incredibly crappy model and definitely should not be used. It's okay to ascribe no theta to a weekend. It's better to ascribe some. But it's not okay to ascribe a full trading day.
I don't know if you looked at the example in the link, but it is not distinguishing days when the market is closed from when it's open. Do you know of any that do it right?
Watch "reality" first, then work the model that best represents what you see for your underlying. Much of finance is captured by Black-Scholes-Merton. BSM uses TO THE SECOND, calendar days. (I think -- I forget now. It's not relevant to remember.) This matches commercial calculation of interest rates, which follows in r and rho. But if you survey the 'net, you'll find a grand 50|50 mix of TIME= calculations where a year has 365 days, or 252. There's your weekends. To call one "wrong" is to call the other "correct." But it's the market in front of you that will decide whether those weekend days count. I have not seen it. I go, to the second, with 252 days, AND I take out holidays (like the upcoming U.S. "MLK" birthday. Go wit' de market. (My vote.)
Is there no hybrid of the 252 and 365 based calculators? Weekends and overnights should count, as evinced by opening gaps, but to a far lesser extent than daytime trading hours. If I was forced to choose, I too believe the 252's will be less distorted from reality as the weekend decay shown in the 365 in the link is, in my observation, grossly exaggerated. Do you have any recommendations?
"...But if you survey the 'net, you'll find a grand 50|50 mix of TIME= calculations... To be clear, I use 365 for anything related to interest rates, and 252 for the rest. Why? I work the short schedule (weeklys; subweeklys) harder than most. And as you're aware, the difference between the 365 and 252 year increases dramatically as time shortens. BUT FOR YOU, again: interrogate your underlying and the predominant timeframe you trade. (A few times, even!) It sounds like in what you've seen, you've seen 252 work better for those realities. Use it, but remember that it's the (collective) market action that steered your decision, rather than something else.
It varies from stock to stock. Sometimes in a big way. PM and MO provide good examples of this. Almost all of MO's decay will occur during trading hours because the domestic regulatory and litigation risks pretty much line up with trading hours. PM on the other hand has a very smooth decay because its risks are more market based than regulatory. So, you'd expect PM to decay on a 365 day model and MO to decay on a 252 day. And just to add some more confusion to the mix, this doesn't always play out that way. The model is only accurate to the extent that it lines up with market participant behavior and those market participants following the model. If you can figure out which model most accurately applies to which stock, you'll have one heck of an edge.
I don't know if there are any retail calculators that will calculate weekend theta. If you need that kind of precision, you should probably write your own. FWIW - I don't worry about weekend theta and I use 256 day count convention.