How the track record works

Every stock GemRadar flags is recorded on the day it is flagged, with its score and its price. The track record measures what became of them — all of them.

What counts as a call

A call is a gem that scored at least 60 on the day itself. That threshold is the point of the page.

The sweep lists every stock it finds, including ones scoring 3. Those are listings, not judgements. Counting them as calls would make the record look worse when they fall and better when they happen to rise, and either way it would measure something other than what we claim. "2 of 447" reads as 445 misses; "2 of 58 calls" is what actually happened.

Sixty sits just below Strong, so it covers the top of Moderate and up — the stocks we actually committed to.

How an outcome is measured

From the first sweep a stock appeared in, against the most recent close we have for it. Not from yesterday: the claim is "we flagged this", so the measurement has to start when we did. Measuring from yesterday would turn every daily move into a result.

A stock flagged today is excluded entirely — its first price and last price are the same day, so it sits at 0.0% and would only pad the denominator.

Stocks that leave the sweep

A gem does not stay in the sweep forever. It drops out when it stops meeting the criteria — and the most common reason to stop meeting them is that it went up: the market cap climbs out of the $300M–$5B band, or the valuation pillar turns expensive and the score falls below the cut.

That matters more than it sounds. The record used to measure to the last sweep day a stock was still listed, which meant a departed gem froze on that day. The bias ran one way: the stocks that ran hardest left first, so the record systematically understated its own best calls. One measured case had a gem recorded at +16.2% while it had actually done +27.1%.

So each sweep now also fetches closing prices for every stock flagged in the last 90 days that is no longer in the list. The outcome keeps updating after a gem leaves, and the page prints the date it is measured to.

What you see, and what is left out of the list

The list shows calls that moved 5% or more, strongest first. Below it is the count of the ones that did not, in words: "50 more calls are not listed: they moved less than +5% since we flagged them."

That sentence is doing real work. The visible list is winners by construction, and a list of winners is not evidence. The count under it, and the denominator in the header, are what make the page a record rather than a brochure. A stock that fell is inside that count — it moved less than +5% — so it is counted, not hidden.

What is excluded, and why

Price series that jump implausibly are dropped as data errors rather than counted as spectacular results. An early version of the proof card advertised "+30268% since 2 September", which was a currency change at the data provider and not a return. See the price checks.

Free and Pro

The counts are for everyone: "2 of 58 calls" is the argument, and hiding it behind a paywall would make it worthless. The list is not — the full outcome of every gem, with score and price, is the content of the daily sweep with a date attached. Free accounts see the calls that cleared the proof threshold plus the counts; Pro sees every outcome, including the flat ones and the losers.

A short record is a short record

Under 20 sweep days the page says so itself. The arithmetic is honest either way, but a handful of days mostly measures how recently we started. Past moves say nothing about future ones.