One of my scenarios, with Specified Expenses Only, runs out of money in the last year. (Ends with Final Plan Savings of $-180,000).
I'm attempting to run a CSS analysis, expecting it to give me a negative number for the CSS column. Maybe I'm misunderstanding, but I thought this tool would either tell me that I could be spending more, or tell me that I should cut back on non-essential expenses. In this scenario, I have about $100k/yr in non-essential early, dropping to around $56k in the later year.
Instead, the analysis returns "This analysis took too long to complete and timed out. You may try again." (it shows it ran 9.6 seconds) and populates a figure of "-1,699,038,867,858,677" in the CSS box.
Feels like a divide-by-zero or buffer overrun error to me, but I'm not sure what's triggering it.
Thanks.
P.S. I checked the box to allow Pralana admins to access my data, if you want to check it out.
@tcbarney The Gold/Excel version would allow negative numbers in consumption smoothed spending, which was nice since it showed how much someone would need to "cut back" in order to get to 90% and retire on the date they've planned. I think there was some issue with that in the online version, perhaps it's not been rectified yet, I'm sure Stuart will look into it and respond here.
@tcbarney Hi Todd, we're looking at it. We've confirmed that the CSS analysis does not converge on a solution when the CSS amount goes negative. This is clearly undesired behavior and will be fixed.
Stuart
I have implemented a simple modification to the CSS analysis that allows the CSS amount to be negative. As you would expect, a negative CSS amount indicates the reduction in spending that a scenario needs to either hit $0 final savings (Deterministic) or 90% success rate (Monte Carlo and Historical).
Stuart and I will test this and it will be in the next release.
Thanks for the replies. I see now on the "Known Issues" page in Pralana Online that this was already flagged. Appreciate all y'all do.
Maybe you have this on your list also. When I run CSS, it seems to work on the deterministic mode, but not Monte Carlo or Historical modes. I expected the advanced algorithm to reach somewhere around 90% success with the MC mode, but it only achieves a 50% success rate.
@patton525 It should return a 90% solution, so we'll have to investigate and get back to you.
Stuart