Our dividend screen halves the chance of a cut. It still didn't beat SCHD

Companies that passed our dividend-history screen cut their dividend about half as often afterwards. A basket of them matched SCHD's return, with deeper falls.

After we retired our trading bot, we turned to dividends. A company's dividend record is public and checkable — judging it does not need anyone to predict the share price. So we built a screen that looks only at a company's dividend record, and then asked it two questions it could fail: does it pick dividends that are actually safer, and is it any better than buying an ETF that already does this job?

The short version: it roughly halves the chance of a dividend cut. And a basket of the companies it picked did not beat SCHD once you account for how far it fell along the way.

The screen

A company passes only if all three are true:

• It has paid a dividend in the last 400 days.
• It has raised its dividend for at least 5 years in a row.
• It has not cut it in the last 5 years. A cut means the yearly dividend fell by more than 10%.

Dividend history is messy — a payment that slips from late December into early January can make one year look like a cut and the next like a big raise. So yearly amounts are measured from each company's typical payment size rather than calendar-year totals, which removes that false signal.

Testing it without hindsight

We have the dividend payment history of 1,092 US dividend-paying companies — 80,405 payments. For each of five dates, 1 January of 2011, 2013, 2015, 2017 and 2019, we cut every company's record off at that date, applied the screen to what was known then, and split the companies into passed and failed. Then we looked at what each dividend actually did afterwards, up to September 2026.

Year is the 1 January screening date. Cut is the share that cut at least once afterwards.

Every time, companies that passed went on to cut about half as often as those that failed, and were more likely to be paying more by the end: 87–93% of them, against 72–80% of those that failed. The 2019 group was tested straight through COVID, when many companies cut, and the gap held.

Is it better than just buying SCHD?

Safer dividends are only worth having if you could not get them more easily. SCHD and VYM are low-cost US dividend ETFs, and SCHD in particular already selects companies on their dividend record and financial strength. So for each screening date we bought every company that passed, in equal amounts, held them without trading, and compared the result with SCHD and VYM over the same period — all with dividends reinvested.

Bought on 1 January of each year and held to September 2026; total return with dividends reinvested. SCHD launched in October 2011, so the 2011 basket can only be compared with VYM.

• Against VYM, the basket won every time: 5 of 5, by 64 percentage points on average.
• Against SCHD, it was ahead on total return in 3 of 4 periods, by 4.7 percentage points on average — over seven to fourteen years, that is close to nothing.
• But it fell further every time: its worst drop was 36–38%, against SCHD's 33.4%. Measured as return per unit of that worst fall, it lost to SCHD in all 4.

And the next section explains why even that near-tie flatters the basket.

What would make this result wrong

• Survivorship. Our list is companies we track today. Companies that were taken over, went bust or stopped paying altogether between the screening date and now are missing. In the cut-rate comparison both groups share that problem. In the SCHD comparison only the basket has it — SCHD's real return already includes everything that failed inside it. So the basket's returns are a best case, and a near-tie on paper is probably a loss in reality.
• This test covers only the history rules. Our live screen also checks payout ratio and cash-flow cover, which this page could not test. A later test could, using the figures companies filed with the SEC as they stood at each date: among companies that passed the history rules, the third paying out the least of their profit cut 10.4% of the time, against 30.7% for the third paying out the most (details). That tests the measures, not the exact limits our screen uses.
• The five dates overlap. A company that passed in 2015 usually passed in 2017 too, so these are not five independent experiments.
• No costs or taxes on either side, and each basket was bought once and never rebalanced.

What we take from it

Screening on dividend history does what it says: it finds dividends that are measurably less likely to be cut. It does not find extra return — an ETF that selects on the same kind of record got there with smaller falls. So we present our dividend tools as a way to see and check a company's dividend record, and to build a list you control, not as a way to beat the market.

Method

Payment history from Yahoo Finance, one request per company, stored and re-used. Prices for the return comparison are Yahoo Finance adjusted closes, which include reinvested dividends, for both the basket and the ETFs. A cut is a fall of more than 10% in the yearly dividend at any point after the screening date, even if it later recovered. A company needs at least 8 payments before the date and 4 after it to be counted. Every number and chart on this page is produced by one script in our codebase, re-run on 27 September 2026.