Your portfolio,
engineered
like a fund's.

How much your holdings move together. Green ≈ they rise and fall as one (less diversification); red ≈ they offset each other (more).

Which holdings actually generate the risk — often one or two names dominate even when the weights look balanced.

How deep each strategy sank below its own previous peak — the pain you would have had to sit through.

Deeper cuts of risk: Downside dev = volatility of bad weeks only · Omega/Gain-loss = money made per money lost (higher is better) · Ulcer = how deep and long drawdowns felt · VaR 95% = a bad week\u2019s size · Time in DD = share of weeks spent below a previous peak.

Cumulative return of each strategy through notable market drawdowns present in your data window.

The mirror image of the stress table: how much each strategy gained through the market's best runs in your data window. Together they show the trade-off — the strategies that fall least usually also rise least.

Scans bonds, gold, commodities, utilities and defensive stocks; ranks by how much each would cut your portfolio's volatility if added at a 10% weight. Low or negative correlation is what does the work.

Every dot is a possible portfolio; the upper-left edge is the best return available for each level of risk. Diamonds mark where the strategies land.

Each bar counts how many weeks landed at that return. A tall narrow hill around 0–1% = a calm ride; long tails to the left = occasional nasty weeks.

Approximate decomposition: each holding's weight × its total return over the window.

The main leaderboard grades every strategy on the same history it was optimized on — its own homework. Here each strategy is built using only the first ~70% of your data, then graded on the unseen remainder. A strategy whose Sharpe holds up out-of-sample is one you could actually have trusted. One caveat: the Market Cap, Income, Value and Quality engines use today’s fundamentals even in the training window, so treat their results as indicative.

Tip: change Rebalance in step 03 (Buy & hold → Weekly) and re-run to see how frequency and transaction cost affect net performance.

An ARIMA model is fitted to the strategy's weekly log returns using the auto-ARIMA stepwise search (Hyndman–Khandakar), which walks the (p,q) order space up to ARIMA(3,0,3) and keeps the fit with the lowest AIC — then projects 52 weeks ahead analytically, with no random simulation. The line is the model's median path; the shaded band is the analytic 5–95% forecast interval, properly compounded from log returns. Wider bands mean the model sees more uncertainty. Illustrative — not a guarantee.

For every strategy: the ARIMA(p,d,q) order chosen by the auto-ARIMA stepwise AIC search (orders up to (3,0,3), each with a constant; d is fixed at 0 because weekly returns are already stationary), its AIC, and the model's 1-year projection. Lower AIC is better only among candidates fitted to the same strategy's returns; AIC values are not comparable across strategies, because each is fitted to a different return series. ARIMA(0,0,0) means the returns look like white noise around a drift — the honest answer for most equity portfolios. Click a row to plot that strategy above.

The Planner answers two practical questions about the portfolio you just analyzed: 1) how much cash income would it pay you in dividends, and 2) if you keep investing every month, what could it grow to over the years? All amounts use your capital and currency from step 03. Estimates, not guarantees.

Each strategy's dividend yield is the weighted average of its holdings' current yields. Income = your capital × yield. A 3% yield on $10,000 pays about $300/year.

An analytic lognormal projection — every cash flow is compounded at the median growth rate, with the shaded band showing the exact 10th–90th percentile range implied by your volatility input (no random simulation). With volatility at zero all three lines collapse to exact steady compounding. Illustrative — not a guarantee of future results.

Enter what you actually own — ticker, number of shares, and (optionally) the average price you paid. The desk values it at the latest close, shows your real profit or loss, measures how far you've drifted from your chosen strategy, and writes out the exact trades to get back on plan. Holdings are saved in your browser, so they'll be here next visit.

Top recent market-wide stories (indices, Fed, earnings, economy), newest first. Sentiment scored 0–100 from the headline.

Sentiment scored 0 (negative) → 100 (positive) from headline text.