How Exent computes the returns, risk ratios, crash drawdowns, correlations, analyst track records, backtests and studies on its pages, how often they refresh, and where the numbers have limits.
Exent uses third‑party market data (daily prices, adjusted prices, company and fund details) and Wall Street analyst ratings data. From them it builds a weekly snapshot that every data page reads, so each number on a page can be traced to a single dated file.
| Universe | Rule | Pages in the Sep 30, 2026 snapshot |
|---|---|---|
| Stocks | The 1,000 largest US companies listed on the NYSE or Nasdaq by market cap, plus other US stocks Exent users hold worth over $300 million | 1,027 |
| ETFs | The 200 largest US‑listed ETFs, bond funds included, plus other US funds Exent users hold worth over $300 million | 378 |
| Analysts | Analysts with a ranking in Exent’s scoring | 1,000 |
| Firms | Firms with at least three ranked analysts | 73 |
| Strategies | Well‑known, non‑proprietary strategies built from ETFs | 19 |
A security gets a page only with at least 36 complete months of returns. If any month shows a fall of 95% or more, the security is left out: that is almost always a data error, and we cannot verify it.
Returns are total returns: price change plus reinvested dividends and distributions, taken from month‑end adjusted closing prices. The series starts on June 30, 2016. Earlier adjusted prices contain glitches in some funds, so we start after them.
| Measure | How Exent calculates it |
|---|---|
| Volatility in performance tables | Standard deviation of monthly returns, annualized by multiplying by √12 |
| Sharpe ratio | Average monthly return above T‑bills (BIL) divided by the standard deviation of those excess returns, × √12 |
| Sortino ratio | Annualized average excess return over BIL divided by annualized downside deviation (only months below the T‑bill return count as risk) |
| Beta | Sensitivity of monthly returns to SPY’s, over the same window as the correlations |
| Max drawdown in performance tables | Deepest fall from a peak in the month‑end total‑return series |
| Trailing max drawdowns | Deepest peak‑to‑trough fall over the last 1, 3, 5 and 10 years, from daily prices |
Month‑end data understates falls that recover within a month, so a performance table’s drawdown can be smaller than the daily one. Both are labelled.
For each security, Exent measures the deepest peak‑to‑trough fall from daily prices inside five fixed crisis windows. A holding that was not trading during a crisis, or had been listed for less than 180 days when it began, shows no value for it.
| Crisis window | S&P 500 (SPY) fall |
|---|---|
| Global financial crisis, 2007–09 | −56.5% |
| 2018 trade‑war selloff | −20.2% |
| COVID crash, Feb–Mar 2020 | −34.1% |
| 2022 inflation bear market | −25.4% |
| 2025 tariff selloff | −19.0% |
A crash replay shows how much room a holding needed in a real crisis. It does not predict the next one, which will have its own cause. Try it on a whole portfolio with the crash test.
Correlations change over time and usually rise in selloffs, so they are a guide, not a guarantee. See correlation coefficient.
The same principles apply to your holdings inside the app, with a few portfolio‑level measures. AI reads your screenshot or file; the numbers come from quantitative models.
| Measure | Definition |
|---|---|
| Risk‑matched benchmark | Each holding is matched to a market index that fits it, and the indexes are weighted like your portfolio: a passive twin with your kind of risk |
| Correlation clusters | Holdings whose returns correlate 0.75 or more are grouped as moving as one |
| Crash liability | Each holding’s share of the portfolio’s loss in its worst 5% of months |
| Risk share | Weight × volatility × correlation with the portfolio, as a share of the total |
| Value at risk (95%) | The loss your worst 5% of months went beyond; conditional VaR is their average |
| Efficient frontier | The best expected return some mix of the holdings you already own could reach at each level of volatility |
| Expected return | A projected yearly return from past data, weighted by each holding’s share; an estimate, not a promise |
The full list of checks is on the portfolio analysis page.
Exent replays every rating that carried a price target over a 7‑year lookback. The current statistics, computed on June 24, 2026, cover 628,792 ratings, 520,742 of them with a price target. For each call it records:
An analyst’s star rating and rank combine success rate, average return and the number of ratings, so a few lucky calls do not make a top analyst. Firm figures are medians of their ranked analysts, for firms with at least three. Individual calls appear on public pages only after a 30‑day delay; live calls stay in the app. Gems, stocks where several analysts rated 4★ and above agree, are built on the same scores. Results so far: are Wall Street analysts accurate?
Strategy backtests use ETFs as proxies for each strategy, monthly total returns, monthly rebalancing unless a page says otherwise, and no trading costs, taxes or fees beyond the funds’ own expense ratios. They are hypothetical and labelled as such.
| Study | Method |
|---|---|
| How many stocks beat the S&P 500 | Total return of today’s 500 largest US common stocks of operating companies (no funds, preferred shares or notes) vs SPY over 1, 3, 5 and 10 years |
| Crash replay | Median and range of crisis drawdowns among today’s 500 largest US common stocks listed at least 180 days before each crisis, by sector |
| What diversifies | Correlation with SPY over all months and calm months; stress months are those when SPY fell more than 4% (16 so far) |
| Lump sum vs dollar‑cost averaging | All 112 rolling 12‑month windows on SPY |
| Retail portfolios | Anonymized aggregates of portfolios analyzed in Exent |
The stock studies use today’s largest companies, so they carry survivorship bias: companies that shrank or failed are missing, which flatters past results. Each study page says so.
The snapshot is rebuilt weekly through the last complete month, and every page shows its as‑of date. Analyst statistics are recomputed on their own schedule and carry their own date.
Nothing on Exent is investment advice. If you find an error, the editorial policy explains how corrections are handled.
The monthly return of the SPDR Bloomberg 1-3 Month T-Bill ETF (BIL), over the same months as the security being measured.
Total returns. They come from month-end adjusted closing prices, which include reinvested dividends and distributions.
Weekly, through the last complete month. Each page shows the as-of date of its data.
Public pages show individual ratings 30 days after they are published. Live calls and the stocks where top analysts currently agree are in the Exent app.
No. They include the funds’ own expense ratios through their prices, but no commissions, spreads or taxes. They are labelled as hypothetical.
Studies built on today’s largest companies are, because companies that failed or shrank are missing. Those studies state it.
Upload a screenshot of your holdings and get concentration, overlap, crash exposure and a risk-matched benchmark in about a minute.
Educational content, not investment advice. Examples are illustrative; past performance does not guarantee future results.