Training Grounds - Lesson VI

How to read our track record

We publish every call we've made — wins and losses, dates and outcomes, nothing removed. This lesson teaches you how to audit a record like that properly: ours first, then anyone else's who claims to have one.

First question: is the list complete?

Before any other number matters, ask whether you're seeing every call or a selection. The most common way performance is misrepresented isn't lying about returns — it's survivorship bias: quietly dropping the losers and letting the survivors imply a hit rate. The cure is structural, not rhetorical. Calls must be published at the time they're made, timestamped, and never deleted. That's how our record works: every pick appears when it's published and stays forever, whatever happens next. A hit rate that can't be audited against the complete list of calls is marketing, not evidence — that sentence is in our glossary because it's the single most important test in this lesson.

Hit rate is the shallowest number on the page

A hit rate tells you how often a desk was right. It says nothing about how much being right earned or being wrong cost, and the difference can invert the story entirely. A house that's right six times in ten — with small, quickly-acknowledged losses and substantial wins — can serve its readers far better than one that's right eight times in ten and then gives it all back on a single blowup it never closed. When you read our table, read magnitude alongside frequency: what did the wins deliver, what did the losses cost, and how quickly were the losses recognised?

Sample size and time keep everyone honest

A record of eight calls proves nothing, however good it looks — at that sample size, luck and skill are indistinguishable. And short windows flatter whoever matched the recent market mood. Look for enough calls, across enough time, spanning more than one kind of market. A record that has only lived through a rising market is an unfinished exam. This cuts both ways: it's the standard you should hold us to as our record grows, and the reason we'd rather show you an honest young record than an impressive curated one.

The audit only we make possible

Here's where the pieces of this course meet. Every pick carries a frozen conviction score (Lesson I). The record shows outcomes. Cross-reference them and you can ask the question that actually validates a methodology: were the high-conviction calls better than the low-conviction ones? If scores mean anything, outcomes should improve as conviction rises — not on every call, but across the sample. If they don't, the score is decoration. We publish both halves precisely so you can run that check on us. Very few research houses hand you the tools for that audit; treat it as a standing invitation.

Scars are the credential

Every honest record has losing calls, losing streaks, and stretches that look ordinary. Their presence is evidence of honesty; their absence is the red flag. When you find a loss in our table, look at what surrounds it — was the bear case published up front (Lesson II)? Was the loss acknowledged at the time rather than reclassified later? A desk's behaviour around its losses tells you more than its behaviour around its wins ever will.

What no record can tell you

Our table reports the outcomes of published calls, measured from publication. It cannot report your outcomes. Your entry timing, your position sizes, your costs and tax, and which calls you acted on all differ from the published baseline — sometimes materially. And the oldest warning in finance applies with full force: past performance is not a reliable indicator of future performance. A record tells you how a process behaved. It cannot promise the process's future, and anyone who reads it as a promise is misreading it.

The checklist, portable

Take this to any provider's record, including ours: Is every call shown, with publication dates? Are losses displayed as prominently as wins? Is the sample big enough and old enough to mean something? Does conviction line up with outcomes? Is the measurement basis stated? Anyone who bristles at those questions has answered them.

URSA

My favourite page on the site. Every wound, catalogued and on display.

TAURUS

That's the deal. The wins mean nothing unless the losses stand next to them — same sand, same light.