It’s in the Little Things

The big calls get the attention: which stocks to own, where the moat really is. But every decision is only as good as the data under it. Get the foundation wrong and even a great model produces confident nonsense.

So we (me and my loyal companion Claude) keep doing the unglamorous work, checking our numbers against the official sources one stubborn detail at a time. Each one is tiny, and each one left alone quietly poisons a ranking and nudges a decision the wrong way. A few we caught recently.

The Case of Negative Equity

Some of the best compounders, TransDigm among them, have bought back so much of their own stock that book equity has fallen below zero. Return on equity then divides profit by a negative number and turns to gibberish, so the off-the-shelf score marks it not available and quietly docks a serial compounder for being too good at returning cash. We read it the right way: a profit earned on less than zero capital is the far end of capital efficiency, so we credit it. And because debt-to-equity is just as meaningless here, we swap in interest coverage, which measures whether the company can actually service its debt.

The Case of the Penalised Banks

A bank has no EV/EBITDA, no price-to-free-cash-flow, no current ratio: those concepts simply do not exist on a bank balance sheet, where deposits are raw material rather than debt. A generic model reads each blank as missing data and quietly docks it, so every bank started a few points behind companies that merely had the right shaped accounts. We now score banks only on the measures that mean something for them, earnings, book value, dividends, profitability, and let the rest step aside instead of counting against them. JPMorgan’s Quality rank climbed nine points on ratios it always deserved.

The Case of Free Cash Flow

Then there are the definitions. Take free cash flow. The off-the-shelf number reads negative for names like Toyota and Sony, not because they are burning cash, but because the feed folds in their lending arms. We rebuild it from the cash-flow statement itself, operating cash flow minus capex, using each carmaker’s own core capex so the lease-fleet noise drops out. Toyota flips from misleadingly negative to a figure that reflects the business.

The Case of the Overcounted Director Trades

One insider feed counted a single director’s sale once for every layer of holding company stacked above it, so an $82M disposal read as nearly −$500M. We deduped it to the one real trade, and an alarming exit shrank back to its true size.

The Case of the Pence and the Pounds

A London income stock showed a dividend yield of 0.1% when the true figure was 6%, a pence-versus-pounds unit slip that made a generous payer look almost stingy. Fixing the units brought the yield, and the income case, straight back.

The Case of the Currency Slip

A euro-reporting company was being read as if it filed in US dollars, overstating its revenue by roughly 14% and flattering every margin and growth figure downstream. We now read each company in the currency it actually reports.

The Case of the Australian Cash Flow

Many Australian filers report operating cash flow the direct way, listing actual cash in and out instead of starting from profit. It is arguably clearer, but it trips up any feed expecting the usual layout, so healthy operators like WiseTech show a blank where their cash flow should be. We read the direct-method statement on its own terms, and the cash they generate shows up instead of a dash.

Why We Bother

None of it is glamorous. It is cross-checks that run on a schedule, comparing our numbers against company filings and official feeds, so errors surface in days instead of sitting in a screen for months. Because accurate data is the base everything else stands on: a sharper signal, a fairer ranking, a better-judged trade. They all begin in the same place, numbers you can actually trust.

It’s in the Little Things.