MethodologyHow the screener ranks a trade, and where each step stops.

This is the whole mechanism in order, from the list of stocks to the verdict on a single contract. Each step says what it measures and where it stops being reliable. There is no language model anywhere in it: every number comes from published formulas, so the same inputs always give the same output, and any of it can be checked by hand.

1. The universe

By default, the largest companies on NYSE and on Nasdaq by market cap, taken from Nasdaq's stock screener, fetched separately per exchange and merged by market cap. One listing per issuer: dual share classes collapse to the larger, so GOOGL is screened and GOOG is not. Preferred stock, warrants, rights and units are removed because they are priced off the common and have no chain worth screening. ADRs stay. You can replace the list with your own. Any symbol not listed on NYSE or Nasdaq is set aside before data is fetched, because no source here can price it.

2. Two passes

A quick pass reads a quote for every name. A full pass then fetches each chain and its price history and runs everything below locally. Fetched data is cached and reused, and its age is always shown. A suggestion states when its quotes were fetched, and anything older than an hour is marked stale.

3. Picking the contract

• Delta target: −0.30 for puts, +0.30 for calls, by default. The contract nearest the target is chosen.
• Tenors: about 30 and 90 days, by default, preferring standard monthly (third Friday) expiries whenever at least two are available.
• Liquidity, on that exact contract: open interest of at least 250 and a bid-ask spread no wider than 12% of the midpoint, by default. Stocks under $5 are skipped, and so is a contract paying less than $0.20 a share at the midpoint, where costs take most of the premium.

4. Volatility

• Implied volatility (IV): at the money, interpolated to the target tenor between the nearest expiries.
• RV30: realized volatility over 30 sessions, with the Yang–Zhang estimator (open, high, low and close).
• Variance risk premium: IV ÷ RV30.
• Term slope: IV at the target tenor ÷ IV at a later expiry, about three times further out and at least 60 days. Above 1.05 the structure counts as inverted; above 1.20 it is flagged as event risk.
• Richness: (estimated fill − fair value at RV30) ÷ fair value at RV30, for the specific contract.

More on why these measures and not others is on the variance risk premium page.

5. The score

Each measure is turned into a percentile against the other names in the same run, then combined:

score = 100 × (0.55 × VRP + 0.25 × yield + 0.20 × liquidity)

where liquidity is half open interest and half (tighter) spread. There is no IV-rank term: a live snapshot carries no implied-volatility history, so that weight went to the variance risk premium. When the term structure is inverted, the score is reduced by

penalty = 40 × min(1, (term slope − 1) ÷ 0.3)

so a slope of 1.06 costs about 8 points and anything from 1.30 up costs the full 40. Earnings are read directly, not inferred: a report before expiry is flagged and handled in the verdict (step 8).

6. Sizing

One contract ties up its breakeven times 100: the strike less the credit for a put, the share price less the credit for a covered call. The number of contracts is that capital divided into your maximum (default $20,000) — or into the share of it one name may take (default 25%), whichever is smaller — rounded down. A name whose single contract is over that share but inside the maximum is still one contract, with a note saying how much of the capital it takes. A name whose single contract exceeds the maximum is still shown, priced at one contract and marked Avoid with the figure.

7. Two scoring modes

The risk-adjusted mode resamples trading-day returns to fill a calendar tenor. Drawing calendar days instead would inflate volatility by roughly a fifth and was measured flipping a fair-value trade from positive to negative expected value. The resampling is seeded per stock, so the same data gives the same numbers every time.

8. Verdicts

Rules are checked from the top, so a trade that is both over your limit and high-scoring is Avoid. Premium mode ranks on premium, but it does not call a trade a Candidate when the expected value over the stock's own returns is negative.

9. The shared Suggestions

Suggestions fill themselves from two places: members' finished screens, re-run by the server on its own data with the member's settings, and a daily scan of the 100 largest names. Either way a trade is shown only if it clears a stricter bar than the interactive screen, because a finding served to someone who did not run it should not need a caveat read to be safe. A contract must have IV at least equal to RV, positive expected value, richness above 5%, expected yield on capital above 5%, a term slope no steeper than 1.20, and no earnings report before expiry. When the market itself is stressed — VIX above its three-month counterpart, or the S&P 500's own options inverted — nothing new is added. Findings expire after 24 hours, and every one is kept and settled at expiry, so the list's own record is on the Performance tab. It uses free data sources only, never one person's brokerage data.

10. What it does not do

• It never places, changes or cancels an order. The brokerage connection is read-only and is checked in two independent ways.
• It models each stock's dividend yield in every price, but not fundamentals or your cost basis. The next ex-dividend date is estimated from the payment schedule, so confirm it before relying on it.
• It knows earnings dates, and infers other events from the term structure, which misses surprises.
• It does not forecast. Every volatility estimate looks back, and a high rank is a hypothesis about premium, not a prediction of return.

Related

• Cash-secured putsScreen short puts by premium richness, yield and liquidity.
• Covered callsThe same screen on the call side, sized on the shares.
• The wheelBoth legs of the wheel, and what assignment costs.
• Variance risk premiumImplied against realized volatility, name by name.