Its ranking barely sorted and the single run lagged. So why did GARP usually beat the market? Here is the resolution.

Start with the uncomfortable part, because an honest case has to. If GARP were a precise ranking, its best scored names would have led and its worst would have trailed. They did close to the opposite.

The fine ranking did not pick the winners

Sort every stock by its combined cheapness and growth score, split into ten buckets, and the return should climb from worst to best. It did not. The relationship was slightly backwards: the very top scored names lagged the middle of the pack, and the rank correlation came out positive, the signature of an ordering that barely sorts and, if anything, leans the wrong way. GARP's edge, then, was never in naming the single best stock. It was in the company you kept by running the screen at all.

Only the sharpest end kept pace with the market

Look at the single full run a different way, by how tightly you concentrated. The broad decile and the top thirty lagged badly over the straight line from 2018, but squeezing into just the top ten names roughly matched the index at about 12% a year. That is the reverse of a safety screen, where concentration hurts. Here the handful of strongest cheap growers carried the load, and diluting across a wider basket watered them down, on this one run at least. Held across many start dates, as the overview showed, even the broad basket usually won; concentrated, it kept pace even on the unlucky single line.

The weather it needs is growth

Here is the engine under everything. GARP is a bet on the economy expanding, and the data says so plainly. Split the months by macro backdrop and its returns are night and day: when growth was rising it earned around 4.6% a month, and when growth was falling it earned almost nothing. Cheap, cyclical, small companies are geared to the upswing; in a slowdown they simply mark time. Any decision to run GARP is really a view that the expansion has further to go, and it is worth holding that view consciously rather than by accident.

What it really is, and what the returns cost

Strip away the label and GARP resolves into something plain: a cheap, growing, mostly small cap basket. Its closest match in the real world is simply a broad small cap index, which it tracks with only a whisker of extra return, so the honest description is a rules based small cap tilt you could run yourself, not a source of hidden alpha. And the returns, when they came, came dear. For every unit of the wild swings you endured, GARP paid you very little: on a risk adjusted basis the broad cuts barely cleared zero while the index earned a normal, comfortable amount, and only the concentrated top ten came close to fair pay for the ride. GARP rewarded patience, but it charged a great deal of volatility for the privilege.

So how do you actually use it?

Three rules fall out of the evidence. Hold it for years, not months, so the distribution of outcomes rather than one unlucky start decides your result. Expect it to work in expansions and to mark time in slowdowns, and size it accordingly. And lean on the basket, not the ranking: the screen's job is to herd you into cheap growth, not to crown the single best stock. Run that way, GARP stops looking like the disappointment its headline suggests and starts looking like what it is, a volatile, cyclical engine that rewards patience and a rising economy.

So we took it to Krest, and ran it through the whole test.