The Deep Value Screen Needs Three Filters, Not One
Zorro's Deep Value preset stacks trailing P/E, price-to-book and FCF yield together - and why any one of them alone lets in the wrong stocks.
A trailing P/E under 12 sounds like the whole definition of "cheap." It isn't. A stock can clear that bar and still be a melting ice cube - a business whose earnings are about to fall off a cliff, or whose book value is propped up by assets nobody would actually pay for in a sale. Zorro's screener handles this the way most useful value screens do: by refusing to let a single ratio decide anything on its own. Its "Deep value" preset stacks four separate conditions, and the reason is that each one is blind to a failure mode the others catch.
The four conditions
A stock has to clear all of the following at once to show up in a Deep value result:
- Trailing P/E between 0 and 12 - priced cheaply against the earnings it has already reported.
- Price-to-book between 0.1 and 1.5 - trading at or below a reasonable multiple of stated net asset value.
- Free-cash-flow yield of at least 5%, or a pass for financial companies, where "free cash flow" isn't a clean measure to begin with (a bank's cash movement is loan and deposit activity, not owner earnings in the normal sense).
- Forward P/E divided by trailing P/E no higher than 1.25 - the market's own forward estimate isn't already backing away from this year's earnings.
Commodity plays and bond-like vehicles - gold and silver miners, ETFs, trusts, royalty companies - are excluded before any of the four conditions are even checked, because they score artificially "cheap" on book value for reasons that have nothing to do with being undervalued operating businesses.
Why trailing P/E alone is the wrong filter
Trailing P/E is built entirely from the past: the twelve months of earnings a company has already booked, nothing about what it expects next. A commodity producer or a shipping company at the peak of its cycle can print a trailing P/E of 8 while sitting on margins that everyone - including its own analysts - expects to compress. A screen that stops there waves the stock straight through on a number that is about to stop being true.
That's what the fourth condition is for. The ratio of forward P/E to trailing P/E is a durability check: below 1 means the market expects earnings to grow, and a value above roughly 1.25 means the opposite - a double-digit earnings decline already priced into next year's estimate. This same ratio gates Zorro's "Small-cap value" preset too, for the identical reason: a low trailing multiple only means something if the market isn't already anticipating that the number driving it won't repeat.
Why price-to-book alone is the wrong filter
Book value has its own blind spot, which is why Deep value doesn't stop at "trailing P/E cheap." A company can trade at 0.4x book and still be a bad trade, because book value is an accounting figure, not a liquidation estimate - inventory that's half-obsolete, receivables that won't collect, goodwill left over from an acquisition that didn't work out. None of that shows up as a lower book value until a write-down finally forces it, by which point the stock has usually already re-rated on the news. Capping price-to-book at 1.5 (not 1.0) while also requiring cheap trailing earnings is the compromise: a company doesn't have to be trading below its stated net worth, just not paying a meaningful premium for it, and the earnings and cash-flow conditions have to agree independently before it counts as "cheap" rather than merely "cheap-looking."
Why free cash flow is the tie-breaker
The FCF yield floor exists because both of the first two conditions can be satisfied by a company that's cheap because its accounting earnings are cheap, not because the business generates cash. Depreciation schedules, one-time charges and working-capital swings can all push reported earnings in directions that have little to do with the cash actually coming into the business. A free-cash-flow yield of at least 5% means money is actually arriving, measured by what's left after real capital spending - a much harder number to flatter with accounting choices than net income.
What four filters buy you that one doesn't
Each condition on its own passes a wide net of stocks, and most of what any single filter catches is noise - a low P/E caused by a one-off earnings spike, a low P/B caused by stale assets, a decent FCF yield on a business about to lose its core customer. Requiring all four to clear at the same time is what turns "looks cheap on one measure" into "cheap on earnings, book value and real cash flow, with no sign the market expects this year's numbers to be the high-water mark." That's a materially smaller, and more defensible, list than screening on any one ratio alone - and it's a different list from Zorro's Quality compounder preset entirely, which skips price altogether and instead tests return on capital, margin and leverage. A stock can fail Deep value for trading too far above book and still pass Quality compounder on the strength of its balance sheet, because the two presets aren't asking the same question.
What it still can't tell you
None of this is a forecast. Every one of the four numbers is backward-looking - last reported earnings, last reported book value, last reported cash flow - recalculated each time new financials land. A stock can clear all four conditions this quarter and fail next quarter the moment one of its inputs changes, and clearing the screen says nothing about where the price goes from here. It says only that, as of the last filing, a company was cheap on three independent measures of value at once, with no obvious sign from the market's own forward estimate that that was about to stop being true. Reading the recent-IPO tracker makes the contrast clear: almost nothing that priced in the last year shows up on a Deep value list, because a freshly listed company usually hasn't accumulated the multi-quarter earnings and book-value history the screen's first three conditions depend on - it isn't that young companies can't be cheap, it's that this particular screen can't yet tell.