Hi HN,<p>I’m Justin, and I’d like to get some feedback on Indie Adviser. It’s a web app I initially built to support my own independent state-registered investment adviser (RIA) business. I’m now planning to market it to other financial advisers. I’ve set up a dummy login button for HN visitors, so there is no need to create a real account.<p>It’s designed to complement an adviser’s existing brokerage portal, adding portfolio optimization and cashflow projection tools, as well as some client recordkeeping that I think is specifically useful to RIAs. These are modules typically included in a big-box wealth management firm’s proprietary toolkit. The idea is to help financial advisers go independent using a modular toolkit that can work alongside any brokerage portal. I personally use Interactive Brokers, and that’s the only brokerage connection currently implemented.<p>The modern portfolio theory optimizer generates mean-risk efficient portfolios (<a href="https://en.wikipedia.org/wiki/Modern_portfolio_theory" rel="nofollow">https://en.wikipedia.org/wiki/Modern_portfolio_theory</a>). Modern portfolio theory has been criticized for 1) relying on variance as a risk metric and 2) forcing an assumption that asset returns are normally distributed.
The portfolio optimizer solves the first problem by allowing for alternative risk measures such as semi-variance and loss thresholds. I prefer semi-variance as a risk measure because it doesn’t count upside movement as risk. Semi-variance has been traditionally omitted from portfolio optimizers due to the simplicity of calculating plain-old portfolio variance. Portfolio variance doesn’t require a full calculation of the weighted portfolio returns distribution, as long as the covariance matrix of the individual assets is known. But with modern computing, I don’t see any reason to not “upgrade” risk measures. Note: the frontend only has the typical variance risk measure implemented.<p>To solve the second problem, I use a vine copula model for asset interrelatedness rather than the typical joint normal return distribution assumption. The joint return distribution generated with the copula model is the input to the portfolio optimizer algorithm.<p>I would very much appreciate any feedback. I’m particularly interested in feedback from any other financial advisers, especially those who do their own portfolio models. I’m also interested in feedback on the web app itself—I’m not a professional so I’m mostly learning as I go. And finally, if you work in institutional portfolio management and are interested in bespoke MPT optimizations, contact me!<p>Thanks very much!
Justin Luther
justinluther@lutherwealth.com
<a href="https://www.lutherwealth.com" rel="nofollow">https://www.lutherwealth.com</a>