NOT-ADVICE!!
like - really.
I've been thinking about spread width and position sizing, and I've got some examples that I want to work using the option chain at this moment, with the share price at $1047. For some reason (it was a good one at the time) I'll pretend that we've sold a range of different positions all keyed on the 1100 strike put ($50 ITM right now); ugh! Let's also assume a $1M cash pile so this works as well in a retirement account as a brokerage. The specific purpose of this chosen strike is to mimic a really bad / deep ITM circumstance. Working these same numbers using a 1050 or 1000 strike put will be a LOT friendlier (so roll early!?!)
Cash Secured Put.
The 1100 strike will tie up $110k per contract, so we can sell 9 of these. We did sell 9 of these and they are now trading at 57.75 (this week) or 71.00 (next week). For this week expiration we're lined up for a 5775*9 = ~$54k loss - about 5%
Clearly this is bad - we can roll this week CSP to 1090 next week for $65 for a net credit of 7.35 and a $10 improvement in the strike. $50 ITM and we can add 1 week while improving the strike $10 (pretty awesome really).
As a bonus we retain the other management choices (the $200 wide, $100 wide, and $50 wide put spreads discussed next).
$200 wide put spread
As a starting point, we have the 900/1100 put spread expiring this week. That spread is trading at 57.75 - 1.20 = 56.55. At $20k per spread we could have sold 50 of these and we're looking at a 5665*50 = $56k*5 = $280k or so loss or around 28%.
To roll to next week we need a 56.55 credit. Next week's 900/1100 is 71 - 8.30 = 63.30. The 890/1090 is 63 - 7.70 = 55.30. A small net debit for the $10 strike improvement, or a reasonably large net credit while keeping the strikes unchanged.
Even on a $200 wide spread, getting 1/4th of the way ITM and our roll options are already getting bad, but aren't really bad. There is a credit or a strike improvement on offer.
$100 put spread
As a starting point we have the 1000/1100 put spread expiring this week. This sounds bad already! It is trading at 57.75 - 5.10 = 52.65. At $10k per spread we could have sold 100 of them (and we did). We have a $5265 * 100 = $526k loss we're looking at.
EDIT: I did this roll logic badly / wrongly. I've edited to distinguish between incremental credits and unrealized losses.
We can roll this spread to next week. The 1000/1100 is trading at 71 - 20.60 = 50.40 and the current week position is trading at 52.65. We sold 100 of these so we're looking at a 2.40 or $24k realized loss in order to delay the $500k loss that is unrealized as yet. We paid $24k for some incremental time and hoping for the share price to go up - preferably above that $1100 strike we sold.
We can get a net credit with the 1010/1110 = 79 - 23.40 = 55.60. We pick up a $3 credit or $30k (yay!) while rolling our $500k unrealized loss out a week. The extra $30k is nice, but we also worsened our position by $10. Now we need the 1110 strike to go OTM. We're also $10 further ITM with this roll. The point is that we're either paying a net debit or we're rolling to a worse strike - the position is at the midpoint, and will be getting worse using a 1 week roll. It might be that a 2 week roll will get us into the small net credit, but I suspect that it won't make a difference.
$50 put spread.
I'm not going to do the math - this would be the 1050/1100 and is pretty much all ITM, so about a $50 loss on a $50 wide spread; around $1M loss on $1M at risk. The roll option is going to be particularly bad and with the 'narrow' spread width there won't be more capital to add to improve the position. In this case you'd definitely want to have used a fraction of your margin (25%?) in order to have management choices from here. Including take the big loss (hopefully sooner than the long put going ITM) as you reduced the 100% loss to $250k by only using 1/4th of your available cash. You'll be able to run the same size position next week.
BUT. That $200 wide spread has an additional management approach available. We could cut that spread size in half and double the contracts ($100 wide spread with 100 contracts). We'll need to beat 56.55 but we have 2x the contracts to do so. So the 940/1040 is around $23 ($46). The 950/1050 is around 39-12 = $27 ($54). Almost enough - so the 960/1060 = 44-13 = 31, or $62. About a $6 credit and a $40 improvement in the strike.
Similar idea but moving down to a $50 wide spread (4x the contracts at 1/4th of the spread size).
25% loss at this moment (56.55/200). Closing right now will cost us $56.56 (I didn't do the math on this one - I'd expect a further strike improvement - maybe to the 1010 range, or about 1/2 of the first improvement).
The $100 wide spread also has this 1/2 strike / double contracts approach available but it won't be as good as we're already starting 50% ITM.
The $50 wide spread doesn't have this option available (you could drop to a $25 wide spread with another 2x the contracts...).
This is also about position sizing. How much of that $1M do we put into these positions? I did the math here assuming its all in the position. That doesn't strike me as awfully risky with the cash secured puts. They can be rolled ~forever (and I have rolled a deep ITM one of these for 5 months). None of the spreads can be rolled ~forever unless you can roll for at least a net credit, and preferably a net credit plus a strike improvement. As long as there is a strike improvement and a net credit (with no reduction in spread size), then they actually can be rolled forever The problem is that the range in which the spread can be rolled effectively is a LOT smaller than the range in which the CSP can be rolled effectively ($50 ITM vs $200 ITM from past experience with deep ITM cash secured puts - it's highly dependent on the share price and on IV).
Do we use 1/4th of the available margin for that $50 wide spread? That's 50 contracts instead of 200 - I don't know where that'll fall relative to a cash secured put, and is something to evaluate. Do we use 1/2 of the margin for the $200 wide spread? I don't know what a somewhat aggressive margin usage will be, and I know that there is all sorts of position sizing stuff that's been written. Maybe somebody has a good book for us to read on this topic
This also points to the closest to ADVICE I have here - if you're rolling a spread for time, take the maximum strike improvement you can get (subject to the net credit, or at least a really small net debit). A week or a month of no income is trivial next to the sorts of large losses that are available via spreads.
And think about what a 50% or 100% loss on these spread positions will do to your account.