Robo-Advisor Development

A digital suitability questionnaire covering investment goals, time horizon, risk tolerance, income, net worth, liquid assets, and investment experience. The questionnaire structure references frameworks used by FinaMetrica and Riskalyze -- covering both the investor's capacity for risk (their ability to absorb losses financially) and their tolerance for risk (their psychological comfort with portfolio volatility) -- because the two measures are not the same and a suitability determination requires both.

Each answer set feeds a scoring model that maps the investor to a risk profile tier -- conservative, moderate, moderately aggressive, or aggressive. The scoring model is configurable: you define the point values per answer and the tier boundaries. The system stores a complete suitability record for every assessment with the question responses, scores, and resulting profile determination -- the documentation required under SEC Regulation BI (Regulation Best Interest) to demonstrate that investment recommendations were in the client's best interest at the time of assessment.

Under Regulation BI, which replaced the previous suitability standard for broker-dealers, the obligation is to act in the investor's best interest rather than to simply recommend a suitable product. The suitability record the system generates supports this requirement with a documented basis for the profile determination. For RIA-registered firms under the Investment Advisers Act of 1940, the fiduciary standard requires an even higher documentation threshold -- the system supports both. Profile refresh triggers at defined intervals (typically annually) or when the investor reports a significant life event -- retirement, inheritance, major expense -- so the portfolio allocation stays aligned with current circumstances rather than the answers given at account opening.

A model portfolio library organised by risk tier. Each model defines target asset class allocations -- domestic equities, international equities, fixed income, alternatives, and cash -- and maps each allocation bucket to specific ETFs or index funds. The construction methodology applies Markowitz mean-variance optimisation (Modern Portfolio Theory) to identify the asset allocation on the efficient frontier for each risk tier, maximising expected return for a given level of portfolio variance. For forward-looking return assumptions, the Black-Litterman model provides a structured approach to incorporating market views into the expected return estimates rather than relying solely on historical covariance data.

The construction engine selects securities, calculates target weights for each position in the portfolio, and applies fractional share support so accounts below the minimum investment threshold for any individual security can still achieve full diversification. Without fractional share support, a small account invested in a portfolio with five ETFs might be unable to achieve the target allocation precisely -- the engine handles this automatically by calculating the fractional quantities needed.

ESG screened model portfolio variants are available as an overlay, filtering out holdings that fail environmental, social, or governance criteria based on MSCI or Sustainalytics ESG ratings. The ESG overlay applies at the security selection stage -- holdings below the configured ESG rating threshold are excluded and replaced with alternatives that maintain the target asset class exposure. New accounts are fully invested at the target allocation within one trading session subject to market hours and execution confirmation from the broker-dealer.

Two rebalancing modes work in parallel. Threshold-based rebalancing fires when any asset class drifts beyond a configurable tolerance band -- typically plus or minus five percentage points from the target allocation weight -- from the target weight. The threshold is configurable per risk tier because a conservative portfolio's fixed income allocation has different drift sensitivity than an aggressive portfolio's equity allocation. Calendar-based rebalancing runs on a set schedule regardless of drift, most often quarterly or semi-annually, ensuring that even accounts without significant market movement receive a periodic alignment check.

Cash flow rebalancing directs new deposits toward underweight positions first before placing additional buy orders across all positions -- a technique that reduces the number of rebalancing trades needed and lowers transaction costs for accounts with regular contributions. The rebalancing engine calculates the minimum set of trades required to restore target weights, minimising round-trip transaction costs.

Tax-aware rebalancing adds a pre-trade check before any sell order is generated in a taxable account: the engine calculates the estimated tax cost of realising the gain in each overweight position and compares it against the cost of allowing the drift to persist. If the estimated tax liability exceeds a configurable cost-benefit threshold, the rebalancing trade for that position is deferred to a future cycle when conditions are more favourable. The daily NAV calculation for each account provides the current portfolio value, unrealised gain/loss per position, and performance attribution across asset classes -- inputs that feed both the rebalancing engine and the investor-facing reporting dashboard.

The harvesting engine scans taxable accounts daily, identifies positions with unrealised losses that meet the harvest threshold, and generates sell orders to realise those losses for use as capital gain offsets elsewhere in the portfolio or against ordinary income up to the annual limit. The daily scan compares each position's current market value against its tax lot cost basis -- tracking short-term and long-term lots separately because the tax treatment differs under IRC Section 1061 and the standard capital gains rate schedule.

Before executing any harvest, the engine checks the wash-sale rule under IRC Section 1091: a substantially identical security cannot be repurchased within 30 days before or after the sale that generated the loss, or the loss is disallowed. The engine selects a replacement security from a pre-approved substitution list -- a correlated but not substantially identical ETF that maintains the intended asset class exposure without triggering a wash-sale violation. For example, a harvested position in a US large-cap index ETF from one issuer might be replaced with a comparable US large-cap ETF from a different issuer tracking a different index.

All harvesting activity is logged per account with the harvested position, the realised loss amount, the replacement security, and the lot identification used. An annual tax-loss harvesting report is generated for each investor and their tax advisor covering the year's harvesting activity, realised losses by holding period classification, and wash-sale adjustments applied. Harvesting is automatically disabled when the expected tax benefit falls below a configurable minimum threshold, preventing small harvest executions that generate transaction costs exceeding the tax value.

Order routing integrates directly with broker-dealer and custodian APIs -- Alpaca (REST API with fractional share support and paper trading environment for pre-production testing), DriveWealth (fractional shares for retail investor accounts with FINRA-regulated clearing), Interactive Brokers (IBKR Client Portal API or FIX protocol for institutional-volume order flows), and Apex Clearing (for RIA custodial accounts via the Apex REST API). The FIX protocol connectivity layer is built when the custodian requires it -- FIX 4.2 or FIX 4.4 is the standard for institutional order execution and provides the low-latency order routing path required for large account portfolios with frequent rebalancing cycles.

The execution layer supports market and limit orders, handles fractional shares for accounts below full-share minimums, and processes dividend reinvestment automatically into the target allocation weighting. Every order status update -- new, filled, partially filled, rejected, cancelled -- flows back to the portfolio ledger in real time for reconciliation. The reconciliation engine catches fills, partial fills, rejections, and corporate actions (stock splits, mergers, special dividends) and updates the ledger accordingly without manual intervention.

Order routing logic handles market hours (orders queued outside NYSE trading hours are submitted at next open), trading halts (halted securities flagged and excluded from rebalancing execution until the halt lifts), and reject retry logic (orders rejected by the broker for technical reasons are retried with exponential backoff and flagged for review if the retry limit is reached). FINRA suitability requirements under Regulation BI apply at the order generation stage -- the execution layer checks that each trade is consistent with the investor's suitability profile before routing the order.

The investor-facing dashboard shows portfolio value updated using daily NAV calculation, performance against a benchmark (S&P 500, a blended benchmark matching the target allocation, or a custom benchmark defined per portfolio model) over multiple time periods (1 month, 3 months, 1 year, inception to date), asset allocation as a chart against the target allocation, and a contribution and withdrawal timeline. The daily NAV calculation reflects the prior day's closing prices for all holdings, with intraday estimates available where the custodian provides real-time position data.

A goal projection module calculates projected portfolio value at the investor's target retirement date or goal date using the current contribution rate, current portfolio balance, and the expected return for their risk tier. The projection uses Monte Carlo simulation or deterministic modelling depending on the platform's approach -- showing the investor a range of outcomes (median, 75th percentile, 25th percentile) rather than a single number that implies false precision about future market returns.

Tax documents are generated automatically: 1099-DIV, 1099-INT, and year-end consolidated tax statements are produced from the ledger data and made available in the investor dashboard without manual compilation. Downloadable performance reports cover any date range the investor selects, with time-weighted return (TWR) calculated to eliminate the distortion that contributions and withdrawals would otherwise introduce into the performance figure. The dashboard is built for both web and mobile, with white-label theming for financial advisor or bank branding -- the investor sees your brand, not the platform's.