Technology stock ranking
Saved rankings, the corrected historical simulation, methodology, and forward paper-test records—with their limits and next steps.
This saved table shows the top 100 of 174 eligible names as of September 8, 2026. It exposes the alpha score alongside the original combined score, trend score, fair-value score, and liquidity fields so the ranking can be inspected rather than treated as a black box.
Open the complete ranking table ↗
The saved fixed-family simulation runs from October 1, 2021 through September 11, 2026 using the top-10, rank-weighted setting and $30 million liquidity filter. The simulator includes transaction costs, slippage, liquidity and concentration limits, plus turnover tracking; the fixed website run does not enforce a hard turnover ceiling.
These settings were checked against the corrected website-generation path, simulator, and refreshed rebalance log. They describe the main-period export shown here, not an assumption that every experiment used identical settings.
| Setting | Implementation |
|---|---|
| Scope and capital | Saved main-period simulation: October 1, 2021–September 11, 2026. Initial simulated capital is $100,000; charts normalize starting equity to 1. This is the fixed alpha family, not the dynamic model-selection or quarterly top-50 experiment. |
| Selection and weights | Select the top 10 eligible stocks by Alpha_Prototype_XLKCompetitive. Raw rank weights descend from 10 to 1 and sum to 55 before the concentration and liquidity caps are applied. |
| Rebalance schedule | Score anchors are generally monthly, followed by rebalancing on the next available trading date (a one-trading-day delay). The refreshed log contains 61 rebalances, from October 1, 2021 to September 10, 2026, including an additional September 9 score anchor. Prices are daily observations, not intraday execution quotes. |
| Liquidity filter | At least $30 million average daily dollar volume (ADV). The simulator averages price × volume over 20 trading observations, permits a minimum of 10 observations, and shifts the series one trading observation so the trade-date filter uses prior data. |
| Target position limits | At each rebalance, a position is capped at 15% of portfolio equity and at 2% of the stock’s ADV in dollar terms, whichever is tighter. Subindustry exposure is capped at 35%. These are target-weight limits; weights can drift between rebalances. |
| Transaction costs | 10 basis points (0.10%) of gross traded value, counting both purchases and sales. This is applied to the sum of absolute stock-weight changes, not to the reported one-way turnover figure. |
| Slippage | 5 basis points per 1% of ADV traded, scaled linearly for each stock. For example, a trade equal to 0.5% of ADV incurs 2.5 basis points on that trade’s value. This is a simulated assumption, not a measured execution cost. |
| Turnover and trade-size diagnostics | Turnover is half the sum of absolute stock-weight changes plus the absolute cash-weight change. A trade above 1% of ADV increments a breach counter but is not blocked or clipped. There is no hard turnover ceiling in this fixed run; the saved log reports zero trade-size breaches. |
| Cash and fallback | Unallocated weight remains in cash with zero modeled interest. After subindustry caps, excess weight is not forced into lower-ranked names. The fixed website simulation does not activate an automatic QQQ or XLK fallback. |
| Risk metric convention | The reported Sharpe ratio uses mean daily return divided by population standard deviation, annualized with √252. No risk-free rate is subtracted. |
Source check: corrected model revision d6ece32; generate_website_assets.py, monthly_rebalanced_portfolio_simulator.py, portfolio_construction_validation.py, and portfolio_backtest_metrics.py, cross-checked with the supplied metadata, summary, equity series, and rebalance log. Updated website artifacts were generated September 12, 2026 at 5:29 p.m. EDT, following the original September 11 research summary. This page uses that supplied rerun; no additional simulation was run for the website update.
The reported cumulative return difference was +159.88 percentage points versus QQQ and +107.52 points versus XLK. Average turnover was 46.05%. These are historical simulation results, not realized investment returns.
The reported Sortino ratio is 1.7704. Download the updated metrics CSV ↓
The chart below illustrates when a separate dynamic selection policy would rotate to XLK after its selection standards weaken. It does not show trades made by the fixed top-10 strategy above, which uses no active benchmark fallback. The underlying policy-selection records are separate research evidence, not a newly validated fallback backtest.
The corrected September 12, 2026 rerun keeps Alpha_Prototype_XLKCompetitive unchanged, with the $30 million liquidity filter and rank-weighted allocation. It compares a monthly top-10 portfolio with the quarterly top-50 candidate, including rebalance-day price movement before target changes and trading costs.
Overlap with the main backtest: The main test begins October 1, 2021, so the two evaluation periods overlap from October 1 through October 15, 2021. “Aligned” means the two older-period constructions and their benchmarks share dates; it does not mean this window is wholly separate from the main backtest.
| Metric | Monthly top 10 | Quarterly top 50 |
|---|---|---|
| Simulated total return | +90.86% | +114.55% |
| QQQ return · same dates | +99.54% | +99.54% |
| XLK return · same dates | +99.64% | +99.64% |
| Excess vs QQQ | −8.69 pp | +15.01 pp |
| Excess vs XLK | −8.79 pp | +14.91 pp |
| Sharpe ratio | 1.1943 | 1.3087 |
| Sortino ratio | 1.4863 | 1.5541 |
| Maximum drawdown | −33.29% | −34.32% |
| Average turnover per rebalance | 54.93% | 58.40% |
| Sum of rebalance cost fractions | 2.38% | 0.83% |
| Rebalances | 24 | 8 |
Excess returns are calculated from unrounded exports, so subtracting rounded percentages may differ by 0.01 percentage points. Turnover is averaged per rebalance, not per year. The cost row sums the simulator’s rebalance cost fractions; it is not the percentage-point reduction in compounded return.
Monthly top-10 still trails both benchmarks. Quarterly top-50 beats QQQ and XLK over these same dates, but its maximum drawdown is slightly deeper than monthly top-10. The result supports further testing of diversification and rebalance frequency; because both changed together, it does not isolate which change caused the improvement.
Download corrected metrics CSV ↓ · Download evaluation windows CSV ↓
A paper-test tracker records a hypothetical portfolio against new prices, without connecting to a broker. The supplied snapshot began September 9, 2026 with $100,000 of paper capital. By the September 10 mark, equity was $98,515.71: a −1.48% return after just one unseen trading day.
That snapshot lagged QQQ by 0.42 percentage points and XLK by 0.07 points. One day is far too little evidence to judge reliability. The project protocol targets an initial review around 60 unseen trading days and a longer review at 120 or more.
This is the saved September 10, 2026 paper-test snapshot, not a live paper-trading feed. No live trading performance is claimed.
Only selected presentation outputs and code screenshots are included here. Raw vendor datasets, credentials, and data-provider integration code are not included in this website. Data-licensing requirements, especially for search APIs and TV transcript archives, remain a separate item to resolve before any broader release.