AI Screener Validator

AI Screener Validator in ScreenerHub is the AI review layer for your saved screens. Use it to check whether a screen is too narrow, too loose, internally inconsistent, or missing an important constraint before you run it repeatedly or pair it with a portfolio in Monitoring.

What AI Screener Validator is for

The validator is useful after you already have a first workable screen. It is not there to invent a strategy from nothing. Its job is to review the logic you wrote and help you spot issues that are easy to miss when you have been staring at the same criteria for too long.

Three questions the validator should help you answer:

Does this screen express one coherent idea, or did I mix several styles together?
Are my criteria likely to produce too few or too many results to be useful?
Which blind spots should I fix before I monitor this screen or rely on its results?

In practice, that means catching things like a value screen that forgot a quality guardrail, a dividend screen that ignores payout sustainability, or a momentum screen that layers so many thresholds together that almost nothing can pass.

How to use it

Start with a clear first draft

The validator works best when the screen already contains the core of your idea. Build the first version in the Screener: choose the universe, add the main factors, and save it as a screen. The review runs on saved screens, so save before you ask for feedback.

Give the screen a short description of your investment intent. The validator reads the name, the description, and the criteria, and a clear description leads to more relevant analysis.

This first draft does not need to be perfect. It does need to be specific enough that the AI can evaluate the trade-offs. A draft with P/E Ratio TTM < 15, ROE > 15%, and Debt / Equity < 1 is reviewable. A draft with one vague filter and no real structure is not.

Run AI Review from the screen page

Open the screen in My Screens and select AI Review. The validator reads the criteria as they are configured at that moment, so review the same definition you actually intend to run.

That matters if you are iterating quickly. A review is only useful when it matches the exact state of the screen you are about to rely on.

Read each section of the review for what it is

The review comes back in four sections, and they do not carry the same weight.

Strategy Analysis restates the investment thesis the AI sees in your criteria. If it does not match what you meant, the logic is probably not expressing your idea yet.
Strengths lists what the current combination already does well.
Potential Improvements points to gaps: logic that is hard to satisfy together, a screen that may be too broad or too narrow, or one that leans too heavily on a single class of metric.
Suggested Criteria proposes concrete rules that would complement the strategy, each with a short reason. Select Add to put one into your screen.

For example, a screen that combines deep-value multiples with very high growth thresholds may be logically possible but uncommon enough to deserve a second look. A dividend screen with Dividend Yield > 5% and no check on payout ratio or free-cash-flow coverage may deserve a warning even if it still returns names.

Revise the criteria and rerun the review

The validator is most valuable as an iteration loop, not as a final score. Adjust the weakest part of the logic first, then run the review again.

Typical revisions inside ScreenerHub look like this:

replace stacked valuation filters with one core valuation rule and one quality rule
add a balance-sheet check such as Current Ratio or Debt / Equity when yield or value rules are aggressive
remove duplicate filters that all express nearly the same idea
split one overloaded screen into two separate strategies instead of forcing one definition to do everything

If you are unsure whether a change made the screen better, compare the revised result set in the Screener, then use Monitoring Runs or your own review notes to see whether the logic now behaves more consistently.

Common issues the validator catches

Contradictory logic

Some criteria can coexist mathematically but still work against each other strategically. A screen that asks for both deep cyclicals at distressed multiples and highly predictable compounders may describe two different businesses rather than one investable universe.

Missing guardrails

A screen often starts with the exciting part of the thesis and forgets the protective part. The validator is useful for spotting missing balance-sheet, profitability, or sustainability filters that would make the output easier to trust.

Overly narrow rule stacks

If you keep adding one more threshold to improve the list, you can end up with a definition that only finds a few names under one market regime. The validator can flag when the logic looks so tight that it risks overfitting.

Criteria that are hard to interpret together

Even good metrics can become noisy when combined without a clear hierarchy. If the screen uses valuation, momentum, dividend, and turnaround criteria all at once, the validator can help you decide whether the strategy needs one primary axis.

When to ignore a suggestion

AI review is a second pass, not an investment committee. You should ignore a suggestion when the trade-off is intentional and you can explain it clearly.

Examples:

you want an unusually concentrated screen because it is meant to surface only exceptional setups
you are deliberately screening for stressed balance sheets in a turnaround strategy
a metric is missing because you plan to judge that factor manually on the company page

The standard is simple: if you can defend the choice in plain language, keep it. If you cannot explain why the rule belongs there, revise it.

What it is not

AI Screener Validator is not a stock-rating engine. It reviews the quality of the screening logic, not whether a specific stock is a buy.

It is not a replacement for Screener Quick Check. Quick Check evaluates one stock against a screen. The validator evaluates the screen itself.

It is not a historical backtest. If you want to pressure-test a saved screen over time, combine it with Monitoring and the process described in How to Backtest a Stock Screen.

Feature overview: Stock Screener
Build and edit the underlying logic: Screener
Check operators and thresholds: Operators
Validate one stock against a screen: Screener Quick Check
Track how a saved screen behaves over time: Monitoring