The Screener Is the Tip of the Iceberg: What Sits Underneath levelbox

The levelbox screener is the part most people see. It is a ranked list of stocks to consider for the wheel, each with a quality read, a risk read and a suggested strike depth. It is free and useful on its own. It is also the smallest layer. Under it sit a portfolio optimizer and a crash stress for Pro accounts. Under those sits the console I run my own options book on every weekend.

This post goes down through the layers, from what anyone can see to what runs on one account today.

Above the waterline: the screener

The screener answers one question per stock. Is this a company worth owning at a discount, and how deep should a put on it be sold?

Each name gets a quality score from its financial statements, a risk read from its price action, and a delta cap of 0.10, 0.15 or 0.30. The cap is set by explicit rules about each company. How those rules were chosen is its own post. A name that qualifies for 0.30 on its fundamentals drops to 0.10 in any cycle where earnings land before expiry.

What the screener cannot tell you is anything about your book. It looks at one stock at a time and takes no input about what you hold. The risk that ends a wheel strategy is almost never one stock.

Just below: the book-level tools

Pro accounts get tools that work on the whole portfolio at once.

• The premium optimizer plans a set of trades across the book. It maximises premium subject to a crash budget and an assignment budget, and shows what each extra dollar of income costs in modelled drawdown. How it searches is written up in detail.
• The risk stress re-prices every position in a 25% market drop, with implied volatility expanding as prices fall, and reports what is left of net liquidation value.
• Positions load from an IBKR activity statement, so the stress runs on what you actually hold.
• The MCP server gives an AI assistant access to the same data, so you can ask about your book in plain language. Setup is here.

All four measure the portfolio. That matters because correlated positions can turn ten modest risks into one large one.

The bulk of the iceberg: the founder's console

A large part of levelbox is not on the website. It is a set of tools I run locally on my own brokerage account, and it is where the optimizer started. None of it is available to other users today. I describe it here because it explains why the public layers look the way they do.

The weekly gameplan

A six-stage review I run each weekend. The stages are: gather the facts, read the market context, review each position, work out what premium is reachable this month, build a menu of structures, and write orders. It stops after every stage so I can react before the next one starts. I can end the session at any point.

Every stage states what it cannot see. The market-context stage, for example, is context only and does not move position sizes. Writing the blind spot down each week keeps a partial answer from reading as a complete one.

Structures run

The screener only covers cash-secured puts. The structures run takes held names, lists strangles, spreads, iron condors and covered calls, and ranks them by premium collected per unit of upside given up. Strike depth per name comes from the same 0.10 / 0.15 / 0.30 policy the screener shows, so both layers give the same depth for each name.

One rule came from a real mistake: check what is already owned before proposing a structure. On a book that already holds the shares, a covered call can pay more than a whole condor and add no new downside.

Overlay

Each week's chosen trades go into a dated plan file with an order block and a fills table. Fills are recorded against the target credit. A past plan is never edited into the next one. The old file stays as the record of what was decided and when.

Hold-or-fold

This is for when a position goes against me and I want to do something about it. It collects evidence on one open short option: the legs, the cost to close, the probability of finishing in the money, and base rates from my own past fills.

There is no verdict, and puts and calls are handled differently. A short put in the money is the wheel working toward assignment. So the question on screen is whether I want the shares at the effective price. The hazard it flags is taking delivery of a stock that is still falling.

Stock scan

The weekly review only covers names I already hold. The stock scan is where new ones come from. For each of the 14 themes levelbox tracks, it measures the share of names trading above their 200-day average, which shows whether the theme is heating or cooling. Candidates come from institutional 13F filings, post-earnings momentum and screener criteria. Each one is scored on the same wheelability gate the screener uses. Every candidate lands in one of three research buckets, so a rejected name is recorded with its reason and is not re-proposed a month later.

Coaching

Each weekly review is saved as a versioned deliverable: a verdict, a trajectory toward the income goal, a scorecard, a short list of actions, and a disclaimer. Because it is saved, next week's review can read last week's and measure what moved. That is the argument of why a slow portfolio goal needs a coach. It runs for one account today.

Why build it this way round

Option selling pays often and small, and charges rarely and large. A tool can look right for months before the week that tests it. So I run each tool on my own money, every week, and fix it when a number turns out to be wrong.

Ideas have moved between the layers in both directions. The optimizer began as a question about my own book: how much premium could it earn without the crash number getting worse? The per-name delta policy started in the screener and was later wired into the structures run. Most of the rules in these tools exist because a specific figure turned out to be misleading in a specific week.

What this does not change

None of these layers is a recommendation. They produce scores, modelled losses, probabilities and ranked menus. Every crash or margin figure comes from our own conservative stress model and will not match a broker's margin system. The console is set up for one person's goals, delta preference and account, and its rules encode opinions you may reasonably disagree with.

If you want to see the layer that is open to everyone, the screener is here.

Educational and analytical only, not investment advice. This article was drafted with AI assistance; if you spot a mistake, tell us.