How AI Investment Analysis Works

When you describe your investments to Investment Copilot, the AI evaluates your portfolio across six dimensions. Understanding what it checks — and what it does not — helps you use the analysis effectively and know when to seek additional professional guidance.

Asset allocation: The most important determinant of portfolio performance is the mix of stocks, bonds, cash, and alternatives. Studies consistently show that asset allocation explains roughly 90% of the variation in portfolio returns over time. The AI calculates your exact allocation across asset classes and compares it against target ranges appropriate for your age, timeline, and risk tolerance. It flags significant deviations — if you should be 40% bonds based on your profile but you are only 12%, that is a meaningful gap that affects both your expected returns and your risk exposure.

Diversification within asset classes: Owning 10 tech stocks is not diversification. The AI evaluates sector concentration, geographic diversification, market-cap distribution (large vs. mid vs. small), and style balance (growth vs. value). It identifies hidden concentrations — your S&P 500 ETF is already 28% technology, so adding individual tech stocks on top compounds your tech exposure well beyond market weight.

Holding overlap: Many investors own multiple funds that hold the same underlying stocks. If you own both VTI (total US market) and VOO (S&P 500), roughly 85% of VOO's holdings are duplicated in VTI. You are paying fees on two funds that give you nearly identical exposure. The AI identifies these overlaps and quantifies them — 'these two holdings overlap by 83%, meaning $15,000 of your money is paying double fees for the same stocks.'

Fee analysis: The AI calculates your portfolio's weighted average expense ratio and projects the long-term cost of fees using compound growth models. A 0.5% difference in fees does not sound like much, but on $150,000 over 25 years at 8% average returns, it equals approximately $50,000 in lost value. The AI identifies specific high-fee holdings and suggests lower-cost alternatives with similar exposure.

Tax efficiency: Different account types have different tax treatments. The AI evaluates whether your holdings are in the right account types — tax-inefficient investments (bonds, REITs, high-turnover funds) belong in tax-advantaged accounts (401k, IRA), while tax-efficient holdings (index ETFs, growth stocks held long-term) are better in taxable accounts. This 'asset location' optimization can save hundreds to thousands annually in taxes without changing your overall allocation.

Risk metrics: The AI calculates portfolio-level risk statistics including estimated volatility, maximum drawdown potential, and the Sharpe ratio (return per unit of risk). Historical stress tests show how your specific portfolio would have performed during past market crises, making abstract risk feel tangible and personal.

Risk profiling is the cornerstone of investment analysis. Get it wrong, and every subsequent recommendation is misguided. Here is how the AI evaluates your risk profile and why it asks certain questions.

The two dimensions of risk: Most investors think about risk as a single slider from conservative to aggressive. In reality, risk has two independent dimensions. Risk tolerance is psychological — how much volatility can you handle emotionally without making irrational decisions? This is best measured by your actual behavior during past downturns, not by hypothetical questions. The AI asks about 2022, 2020, and any other market experiences you have had. If you panic-sold during a downturn, that is the most honest data point about your risk tolerance. Risk capacity is financial — how much can your portfolio decline before your goals are jeopardized? A 30-year-old saving for retirement in 30 years has high risk capacity because time allows for recovery. A 58-year-old two years from retirement has low risk capacity because there is no time to recover from a major decline.

Why the AI asks about past behavior: Traditional risk assessments ask hypothetical questions: 'If your portfolio dropped 30%, what would you do?' Most people answer rationally in theory and act emotionally in practice. The AI prefers behavioral data — 'What did you actually do in 2022 when markets fell?' If you panic-sold, your allocation was too aggressive for your real tolerance, regardless of what you think you can handle. If you did not even check your accounts, you might be able to handle more risk. This behavioral approach to risk assessment produces more accurate profiles than hypothetical questionnaires.

Timeline-based capacity: Your investment timeline is the single most important factor in determining risk capacity. Money you need in 2 years should be in cash or short-term bonds — period. Money you do not need for 25 years can withstand significant volatility because historically, the US stock market has never had a negative return over any 20-year period. The AI calculates how much time you have for each goal and sets risk capacity accordingly. If you have multiple goals with different timelines (retirement in 25 years, house down payment in 3 years), the AI may recommend different allocations for different pools of money.

The profile output: The final risk profile is not just a label (conservative, moderate, aggressive). It is a specific target allocation range with reasoning: 'Based on your 26-year timeline (high capacity) and your tendency to check accounts during downturns and consider selling (moderate tolerance), I recommend a Moderate allocation: 40-50% US stocks, 15-25% international stocks, 25-35% bonds, 0-5% cash. This allocation has historically produced average annual returns of 7-8% while limiting maximum drawdowns to 20-25% — which based on your described behavior during 2022, should be within your holding threshold.'

Investment fees are the only variable in investing that is certain, controllable, and guaranteed to affect your returns. The AI places heavy emphasis on fee analysis because the math is unforgiving — here is why.

The compounding effect of fees: A 1% annual fee sounds trivial. On a $150,000 portfolio earning 8% average returns over 25 years, the difference between a 0.05% expense ratio (typical for index funds) and a 1% expense ratio (typical for actively managed funds) is approximately $127,000. Not $127,000 in fees paid — $127,000 in total portfolio value lost to fees and the compounding of those fees. That is potentially years of additional retirement income erased by a difference that is invisible on any single statement.

What the AI checks: For each holding in your portfolio, the AI identifies the expense ratio (the annual fee charged by funds and ETFs), any transaction fees or commissions, and advisory fees if applicable. It calculates the weighted average across your portfolio and projects the cumulative cost over your investment timeline. Then it identifies specific substitutions: 'Switching from [Active Fund] at 0.72% to [Index Fund] at 0.03% saves $690/year on your current position, compounding to approximately $52,000 over 25 years.'

The active management question: Roughly 88% of actively managed US large-cap funds have underperformed the S&P 500 index over the past 15 years, according to the SPIVA Scorecard. This means you are statistically likely to get worse performance AND pay higher fees with most actively managed funds. The AI's default recommendation is low-cost index funds for the core of your portfolio — not because active management can never add value, but because the odds are against it for most investors. If you specifically want to own actively managed funds, the AI will respect that choice but ensure you understand the fee implications.

Advisor fees: If you use a financial advisor who charges an assets-under-management (AUM) fee (typically 0.5-1.5% annually), the AI factors this into the total cost calculation. On a $500,000 portfolio, a 1% AUM fee is $5,000 per year. The AI does not tell you whether an advisor is worth the cost — that depends on the value of behavioral coaching, tax planning, and financial planning they provide. It simply ensures you understand the dollar amount so you can make an informed decision. For context, Retirement Copilot can help evaluate whether the services your advisor provides justify the fees they charge.

This is the most important section on this page. Investment analysis inherently involves uncertainty about the future, and the AI is transparent about what it can and cannot tell you.

The AI cannot predict market direction. Nobody can — not fund managers, not economists, not AI. The AI does not attempt to tell you whether the market will go up or down next month, next year, or ever. Studies consistently show that market timing is value-destructive for nearly all investors. The AI's recommendations are based on long-term strategic allocation, not market predictions.

Projections are probability distributions, not predictions. When the AI shows you a projected portfolio value of $520,000 in 25 years, it is saying 'based on historical return distributions, the median outcome is approximately $520,000, with a 90% confidence interval ranging from $380,000 to $720,000.' The actual outcome could be outside even that range. History does not repeat exactly, and the next 25 years may differ from any historical period. The AI presents a range of scenarios, not a single number, precisely because certainty is not possible.

Past performance and historical stress tests: When the AI shows how your portfolio would have performed during the 2008 crisis, it is using actual historical data — your specific allocation applied to actual market returns during that period. This is educational (it helps you understand how your allocation responds to stress) but NOT predictive (the next crisis could be different in magnitude, duration, and which asset classes are affected). A portfolio that 'survived' 2008 is not guaranteed to survive a different type of crisis.

The AI does not evaluate individual stocks. If you ask 'should I buy NVIDIA?', the AI does not perform fundamental analysis (P/E ratios, earnings projections, competitive positioning). It evaluates individual stock positions in the context of your portfolio — 'NVIDIA represents 15% of your portfolio, which is significant concentration risk in a single company' — but does not offer buy/sell recommendations for specific securities. The AI is an allocation and portfolio construction tool, not a stock picker.

Assumptions built into projections: Every projection depends on assumptions about future returns, inflation, contribution rates, and tax rates. The AI explicitly states these assumptions and allows you to modify them. The default assumption uses historical average returns for each asset class, a 2.5% inflation rate, and current tax rates. If any of these assumptions prove wrong (and some will), the projections change. The AI can model alternative scenarios — 'what if average returns are 2% lower than historical?' — to help you stress-test your plan.

Like the other how-it-works pages in this series, this is not an either-or decision. Here is an honest comparison to help you determine the right approach for your situation and portfolio size.

Cost comparison: Financial advisors typically charge 0.5-1.5% of assets under management annually. On a $150,000 portfolio at 1%, that is $1,500/year. On $500,000, it is $5,000/year. On $1,000,000, it is $10,000/year. These fees compound and are deducted from your returns every year. Investment Copilot is included in the Pro plan at $29/month ($348/year). However, the comparison is not apples-to-apples — a good financial advisor provides services beyond investment management.

What the AI does well: Portfolio composition analysis, fee optimization, rebalancing recommendations, risk profiling, and educational explanations of investment concepts. The AI has no conflicts of interest — it does not earn commissions, does not have proprietary products to sell, and does not benefit from keeping your assets under management. It gives the same recommendation whether you have $10,000 or $10,000,000. It is also available immediately, any time of day, for unlimited questions.

What a financial advisor does better: Comprehensive financial planning (not just investments — tax planning, estate planning, insurance needs, Social Security optimization, business succession planning). Behavioral coaching during market downturns (calling you to say 'do not sell' has real value). Complex tax situations (managing capital gains, Roth conversions, charitable giving strategies). Life transition planning (divorce, inheritance, job loss, retirement). Accountability and relationship — someone who knows your full situation over years and can connect the dots between financial decisions.

The portfolio-size threshold: As a general guideline: below $100,000 in investable assets, the AI provides analysis that is difficult to get from an advisor (most advisors have minimums of $100,000-$500,000). Between $100,000 and $500,000, the AI can handle most portfolio management needs, and an advisor is optional unless you have complex circumstances. Above $500,000, a fee-only fiduciary advisor becomes increasingly valuable because the complexity of tax planning, estate planning, and risk management grows with portfolio size. Above $1,000,000, we strongly recommend a fiduciary advisor in addition to the AI — the stakes are too high for any single source of guidance.

The hybrid approach: Many of our users consult the AI between annual advisor meetings. The AI helps them understand what their advisor is recommending (and whether the fees they are paying are justified), monitors their allocation between meetings, and helps them ask better questions at their next advisor meeting. This is not a replacement — it is an informed-consumer strategy.

For tax-specific investment planning, see also Tax Copilot and Retirement Copilot.

You are sharing portfolio values, holdings, income, and financial goals. Here is exactly how that information is handled.

During analysis: Your portfolio data, financial details, and conversation content are processed in server memory (RAM) to generate your analysis. The AI reads your holdings, runs calculations, generates recommendations, and delivers the results. This processing happens in memory only — no portfolio data is written to disk or stored in any database at any point during the process.

After your session: When your session ends or after 60 minutes of inactivity, all financial data is purged from memory. Your holdings, account balances, and portfolio composition are not retained. There is no record of what investments you own, what accounts you have, or what recommendations you received.

No account linking or brokerage access: The AI works entirely from information you provide in conversation. It does not connect to your brokerage accounts, does not use Plaid or similar aggregation services, and does not have read or write access to any financial account. This is a deliberate design choice: it means you control exactly what information is shared, and there is zero risk of unauthorized account access.

No data training: Your portfolio data and financial details are never used to train or fine-tune AI models. The AI's investment knowledge comes from published financial research, market data, and fund databases — not from user portfolios.

No third-party sharing: Your financial information is not shared with brokerages, investment companies, financial advisors, insurance companies, or any other entity. We do not sell investment leads. We do not share data with product providers. Our revenue comes from subscriptions, not from monetizing your financial profile.

Aggregated analytics: We retain anonymized, aggregate statistics only — average portfolio sizes by age group, most common allocation mistakes, most frequently flagged concentration risks. This data contains no individual portfolio information and is used solely to improve our recommendations. You can opt out of even this aggregated data collection in your account settings.