3 ms·

Pretty cool demo for causal.app. Feedback on the framework itself: * I found the interface to widen/narrow the variance of ranges painful to use. Why not jus

Pretty cool demo for causal.app. Feedback on the framework itself: * I found the interface to widen/narrow the variance of ranges painful to use. Why not just let me plug in an exact number? * I assume the output for the variance is 95% interval? This should be stated. * There's major math problems with respect to how you are handling variance on compounding rates and reporting it. Most importantly, the distribution should almost certainly not be normal (lognormal makes more sense). If you don't use lognormal distributions, your confidence interval after N years makes no sense as the product of two normal distributions is not normal (nor even symmetric) - I'm actually not even sure how to interpret your confidence interval after 30 years because of this fact! * The line becomes impossible to see if it has no variance! Feedback on the calculation: * I'm guessing all these numbers are real rates, as there is no inflation? * Monthly rent is really "rent realized" as you are ignoring vacancy. * I'm not following how you are calculating accumulated rental earnings. It is more than the sum of historical rental earnings, but less than your rental earnings being re-invested in the market * "We assume that these cover at least the cost of your monthly mortgage payment" -> including interest + principal? I'm not sure if your calculator per se relies on this assumption, but this would not hold in many markets (e.g. SF Bay Area) BTW, for those who want to do a simple buy/rent trade-off: I have a google sheet here: https://medium.com/@usaar33/an-up-to-date-buy-or-rent-calculator-22d0bf9bbbb5 https://medium.com/@usaar33/an-up-to-date-buy-or-rent-calcul...

refrigerator 8y ago
Thanks for the super detailed feedback! * Yeah the "scrubbing" interface is a little experimental. We originally had it for the mean values too, but that was a bit too crazy so we got rid of it. May well get rid of it altogether. * It's a 2 standard deviation range, so a little more than 95%, but you're right - we should be clear about this! * The entire model runs via simulation, so we're not analytically calculating products of distributions etc. I don't think we're reporting the distribution of compounded returns after N years anywhere, but there may well be something misleading somewhere — what specifically are you referring to? * Thanks for raising the invisible line issue! --- * Yup * Yup * I believe we are just taking the sum of historical earnings! What suggests otherwise? * Yeah including interest and principal. And fair point! In the UK I think it's a reasonable assumption but probably not so in other places. Thanks again — really appreciate your feedback. Happy to chat more here or via email: hi@causal.app :)
- usaar333 8y ago
  > I don't think we're reporting the distribution of compounded returns after N years anywhere, but there may well be something misleading somewhere — what specifically are you referring to? The output, e.g. after 21 years. You'll write something like property value is X +/- Y. The problem is that the distribution isn't symmetric. In particular, being at X - Y is far likelier than X + Y, at least if I interpret X as a mean. In addition, the median is less than the mean, which is X? > I believe we are just taking the sum of historical earnings! What suggests otherwise? As I scroll over years, the net rental earnings are clearly not linearly proportionate to the year.

refrigerator 8y ago

Thanks for the super detailed feedback! * Yeah the "scrubbing" interface is a little experimental. We originally had it for the mean values too, but that was a bit too crazy so we got rid of it. May well get rid of it altogether. * It's a 2 standard deviation range, so a little more than 95%, but you're right - we should be clear about this! * The entire model runs via simulation, so we're not analytically calculating products of distributions etc. I don't think we're reporting the distribution of compounded returns after N years anywhere, but there may well be something misleading somewhere — what specifically are you referring to? * Thanks for raising the invisible line issue! --- * Yup * Yup * I believe we are just taking the sum of historical earnings! What suggests otherwise? * Yeah including interest and principal. And fair point! In the UK I think it's a reasonable assumption but probably not so in other places. Thanks again — really appreciate your feedback. Happy to chat more here or via email: hi@causal.app :)

usaar333 8y ago
> I don't think we're reporting the distribution of compounded returns after N years anywhere, but there may well be something misleading somewhere — what specifically are you referring to? The output, e.g. after 21 years. You'll write something like property value is X +/- Y. The problem is that the distribution isn't symmetric. In particular, being at X - Y is far likelier than X + Y, at least if I interpret X as a mean. In addition, the median is less than the mean, which is X? > I believe we are just taking the sum of historical earnings! What suggests otherwise? As I scroll over years, the net rental earnings are clearly not linearly proportionate to the year.