Warning: The author is neither a tax professional nor a retirement advisor. Any statements contained within this document do not constitute legal, tax, or investment advice.

About

Long-term investors try to keep a mix of asset types in proportions that match their risk preferences. Over time, prices fluctuate and the actual worth of assets will inevitably change. This tool calculates the optimal way to invest a fixed sum of money so that each asset type's value is as close as possible to its desired ratio of the whole portfolio.

Broadly, this tool provides a means of "staying the course" through regular contributions into the right mutual funds.

Demo

Just clone and cargo run to see a demonstration with a real GnuCash database:

$ git clone git@github.com:DavidCain/stay-the-course.git
$ cd stay-the-course
$ cargo run
US stocks: $10032 (🎯 42.68%)
  - FZROX: $5120.56 (485.3619 x $10.55)
  - VTSAX: $4911.94 (66.5305 x $73.83)
International stocks: $7749 (🎯 34.14%)
  - VTIAX: $7749.70 (273.7445 x $28.31)
US bonds: $3393 (🎯 14.63%)
  - VBTLX: $3393.63 (311.0576 x $10.91)
REIT: $3330 (🎯 8.53%)
  - VGSLX: $3330.72 (26.1438 x $127.40)
Portfolio total: $24506

Worth at retirement (Assuming 7% growth):
 - 34: $24506  SWR: $980
 - 50: $69858  SWR: $2794
 - 55: $97975  SWR: $3919
 - 60: $137434  SWR: $5497
 - 65: $192750  SWR: $7710

After-tax income: $49700
Charitable giving: $5000 (10% of after-tax income)
How much to contribute or withdraw?
2000
Contribute the following amounts:
 - International stocks: $903.07
   31.62% -> 32.64% (🎯 34.14%) Δ [7.3% -> 4.4%]
 - US bonds: $313.46
   13.84% -> 13.98% (🎯 14.63%) Δ [5.3% -> 4.4%]
 - US stocks: $783.46
   40.93% -> 40.80% (🎯 42.68%) Δ [4.0% -> 4.4%]
 - REIT: $0.00
   13.59% -> 12.56% (🎯 8.53%) Δ [-59.2% -> -47.1%]

Sample GnuCash accounting records

In example/ are two (identical) sample files in XML and sqlite3 format. Each may be opened with GnuCash 3. The transactions, security prices, and account balances contained within are the basis of rebalancing logic:

How it works

The tool accepts a few key inputs:

1. The path to a GnuCash data file (in either XML or SQLite format)
2. The desired target allocation per asset type
3. The amount of money the investor intends to invest

With that information supplied, the tool will:

1. Use the contained price database to calculate the current worth of each investment fund
2. Classify each fund into an asset type
3. Sum up asset values by asset type, calculate the ratio of each asset class to the total portfolio value
4. Sequentially identify asset classes which have deviated most from their target, invest into those asset classes until they are as close to their target as the next furthest asset class (repeat until the desired contribution amount has been fully allocated to all funds)
5. Output the optimal contributions

Fetching quotes from 3rd party APIs

I'm using the AlphaVantage free API. To use it, make sure that:

1. ALPHAVANTAGE_API_KEY is an available env var. (Get an API key first)
2. update_prices = true is set in [gnucash] within config.toml

When configured, this ensures that the latest stock prices per fund are incorporated into the allocation recommendations.

Background - target asset allocation

Asset allocation is the process of reconciling one's risk preferences and investment goals with a long-term strategy.

Stocks are generally understood to grant higher returns in the long run, at the risk of greater volatility in the short run. Conversely, many consider bonds a source of stable income, with lessened potential for long-term growth. Some asset classes (such as TIPS) may provide some insurance against changing economic conditions, while other funds provide broad exposure to varied economic sectors (adding diversity to a portfolio).

Any long-term investment strategy should be tailored to the risk preferences of the investor. A young professional will likely desire a riskier portfolio than somebody nearing retirement: The young professional is able to tolerate short-term volatility in the hopes of greater returns. The retiree needs a steady stream of income, and is willing to forego greater returns in exchange for stability.

Why use this strategy?

Rebalancing can be expensive. When selling taxable funds, an investor must realize capital gains (or losses) in order to move money from one investment to another. If shares have not been held for a sufficiently long period of time, short-term capital gains may even be realized (at a potentially higher tax rate).

For residents of California, capital gains are taxed as normal income. If the investor plans to retire in another state, they may desire to postpone the realization of capital gains.

Selling shares adds some complexity to annual tax returns, since capital gains must be reported to the IRS. A buy-and-hold investor may instead be able to go years without selling any assets, avoiding the need to report realized gains.

When rebalancing should be performed instead

Regular re-investment may be sufficient to keep asset allocation in line with targets. However, a number of scenarios might cause the ratio of various asset classes to deviate too far from targets:

• Significant market changes (e.g. domestic or international expansion/contraction)
• Change in risk preferences (changing target weights for stocks, bonds, TIPS, etc.)
• Regular contributions being too small a fraction of the overall portfolio.

Only an individual investor can decide when their portfolio's composition has deviated too far from its targets. At that time, it may become necessary to sell overweighted assets and direct the proceeds to an underweighted asset class.

Taking dividends in cash (rather than automatically re-investing into the originating fund) can help alleviate the need for rebalancing. The dividends from overweighted funds may be transferred into underweighted funds. Additionally, receiving dividends this way can help avoid a "wash sale" if the investor plains to perform tax loss harvesting.

Assumptions

• While this algorithm can handle any number of funds and corresponding asset classes, we do assume that the user employs a "lazy portfolio" strategy (a portfolio in which desired allocations are spread across a small number of asset classes and underlying mutual funds). The GnuCash integration only considers assets whose underlying commodities are of type FUND.
• Current values are based on the last known price. The user must keep their price database current within GnuCash in order to get current estimates.

Considerations this tool does not make

1. No differentiation is made between taxable and tax-advantaged accounts. It is up to the user to manage efficient fund placement and direct funds to the appropriate accounts.
2. No consideration is made of minimum investments: Investing in a new mutual fund often requires a minimum investment, but this tool does not account for that. The included example files show an investor building a diverse portfolio first before working towards bringing their various funds in line with target ratios.

Major tasks outstanding

This project is very much a work in progress. Some key outstanding tasks:

• Optimize XML parsing (currently takes a couple seconds on a 20MB file)
• Command line interface
• Return Result instead of just panicking on error conditions

External resources

• Optimal rebalancing (archive): Though it has since been taken down (it lived at http://optimalrebalancing.tk, now spam), this was an excellent web tool by Albert H. Mao. It provided a textual interface to lazily rebalance.
• rebalance-app by Alberto Leal: another Rust implementation based off Optimal rebalancing, but without GnuCash integration and relying on a different underlying type libraries. Probably better suited for people who don't make use of GnuCash. Also includes some features not present in this project!