5 ms·

Hey HN, I’m one of the makers. Enjoying the discussion about buying to rent! We’re building a tool to let anyone create models like this one — models that inco

Hey HN, I’m one of the makers. Enjoying the discussion about buying to rent! We’re building a tool to let anyone create models like this one — models that incorporate uncertainty in their variables and outputs. This is the kinda stuff people do in Excel right now, with much difficulty. We’re working on a couple more demo models — let us know if you have any suggestions. And if this kind of tool resonates with you, then we’d love to chat — hi@causal.app. Thanks!

jedharris 7y ago
I didn't see how to look at the model "code". How can we see that?
kartickv 7y ago
It looks polished at first glance. The bottom line isn't clear because there's too much information. I'm expecting a single number in 60-point font like "You'll save X amount of money by renting instead of buying". Some information can be hidden behind progressive disclosure. Also, it's US-specific. Let me choose a currency, and format numbers accordingly, like "₹2 crore", not "$20m". In India, we use lacs and crores, not millions. I also like that you're using Monte Carlo simulations. Though it would be good to augment that with real-world data from the user's country. If I'm in India, which Indian city has had the worst-case real-estate investment performance? The worst-case is useful information to consider when making an investment.
georgespencer 7y ago
Hey! I'm a really big fan of this idea! Well done on tackling something really important. I think the examples you've chosen are good but not great examples of where you can add significant value. A house being rented where there are a few variables and a good/better/best outcome is not especially gnarly Excel for the Hacker News audience. I think marketing and sales forecasts and stock option pricing and DCF might be more impactful, if more esoteric, in showing the power of what you're building. BTW, it'd be cool to be able to anchor a variable to a historical base or average. So maybe if you have seven or eight forecasts of CAGR, Casual could accept 'citations' and in its output show a user of the model that this isn't totally speculative. Love what you're doing and have signed up. Well done again.
- refrigerator 7y ago
  Thanks - glad you like the idea! You’re totally right that these aren’t the best examples to show the real value of the approach. We wanted to keep things super simple to start getting feedback, but are working on a more involved example now. DCF valuations and options pricing sound like great ideas – will look into them! What do you mean re: citations? Validating Causal models using forecasts from other means?
  - georgespencer 7y ago
    In forecasting folks will often use external data. For example let's say you have six different estimates on the growth potential of a market from six external reports. It'd be cool to be able to specify that a range of assumptions have citations. At the moment folks do this using accompanying appendices to financial models, which is pretty gross.
    - refrigerator 7y ago
      Ahh gotcha — yes that makes a lot of sense! Would love to chat to you more about this stuff - will email you :)
  - somberi 7y ago
    Taimur - Congrats and I signed up as well. To me, one of the advantages of representing the story in a probabilistic way, similar to how you have shown it, is that I am able to sell an idea to my audience. If you are looking for use cases to build for a demo, in addition to DCF, as someone suggested before, you might want to consider currency movements and also IRR calculations. One feature request, if I may, is the ability to manipulate the end result (time horizon in the demo of rent vs buy) and see how it impacts one of the features (rent, market return, etc). All the best and glad to discuss.
    - refrigerator 7y ago
      Thanks, will email you!
grey-area 7y ago
Pension growth over time (and depletion after retirement), and investments in the stock market with compounding gains would be good ones to make. Also perhaps one predicting the probability of lottery wins over a lifetime?
jwr 7y ago
I immediately signed up, as this is the kind of tool I could really use. There are a number of scenarios (especially financial ones) that involve uncertainty in my life, and I've been attacking them with spreadsheets and later Mathematica. But spreadsheets are too simple, and Mathematica is too cumbersome. Applying probabilistic programming tools is too complex: I would spend way too much time on details. So, this seems like just the right tool for the job. Example scenario: should I pay off my mortgage, which is in a foreign currency?

jedharris 7y ago

I didn't see how to look at the model "code". How can we see that?

kartickv 7y ago

It looks polished at first glance. The bottom line isn't clear because there's too much information. I'm expecting a single number in 60-point font like "You'll save X amount of money by renting instead of buying". Some information can be hidden behind progressive disclosure. Also, it's US-specific. Let me choose a currency, and format numbers accordingly, like "₹2 crore", not "$20m". In India, we use lacs and crores, not millions. I also like that you're using Monte Carlo simulations. Though it would be good to augment that with real-world data from the user's country. If I'm in India, which Indian city has had the worst-case real-estate investment performance? The worst-case is useful information to consider when making an investment.

georgespencer 7y ago

Hey! I'm a really big fan of this idea! Well done on tackling something really important. I think the examples you've chosen are good but not great examples of where you can add significant value. A house being rented where there are a few variables and a good/better/best outcome is not especially gnarly Excel for the Hacker News audience. I think marketing and sales forecasts and stock option pricing and DCF might be more impactful, if more esoteric, in showing the power of what you're building. BTW, it'd be cool to be able to anchor a variable to a historical base or average. So maybe if you have seven or eight forecasts of CAGR, Casual could accept 'citations' and in its output show a user of the model that this isn't totally speculative. Love what you're doing and have signed up. Well done again.

refrigerator 7y ago
Thanks - glad you like the idea! You’re totally right that these aren’t the best examples to show the real value of the approach. We wanted to keep things super simple to start getting feedback, but are working on a more involved example now. DCF valuations and options pricing sound like great ideas – will look into them! What do you mean re: citations? Validating Causal models using forecasts from other means?
- georgespencer 7y ago
  In forecasting folks will often use external data. For example let's say you have six different estimates on the growth potential of a market from six external reports. It'd be cool to be able to specify that a range of assumptions have citations. At the moment folks do this using accompanying appendices to financial models, which is pretty gross.
  - refrigerator 7y ago
    Ahh gotcha — yes that makes a lot of sense! Would love to chat to you more about this stuff - will email you :)
- somberi 7y ago
  Taimur - Congrats and I signed up as well. To me, one of the advantages of representing the story in a probabilistic way, similar to how you have shown it, is that I am able to sell an idea to my audience. If you are looking for use cases to build for a demo, in addition to DCF, as someone suggested before, you might want to consider currency movements and also IRR calculations. One feature request, if I may, is the ability to manipulate the end result (time horizon in the demo of rent vs buy) and see how it impacts one of the features (rent, market return, etc). All the best and glad to discuss.
  - refrigerator 7y ago
    Thanks, will email you!

grey-area 7y ago

Pension growth over time (and depletion after retirement), and investments in the stock market with compounding gains would be good ones to make. Also perhaps one predicting the probability of lottery wins over a lifetime?

jwr 7y ago

I immediately signed up, as this is the kind of tool I could really use. There are a number of scenarios (especially financial ones) that involve uncertainty in my life, and I've been attacking them with spreadsheets and later Mathematica. But spreadsheets are too simple, and Mathematica is too cumbersome. Applying probabilistic programming tools is too complex: I would spend way too much time on details. So, this seems like just the right tool for the job. Example scenario: should I pay off my mortgage, which is in a foreign currency?