My 2022 Posts Beat My 2026 Posts. Badly.

Simple trading newsletters get read because they commit: one idea per issue, one claim specific enough to check, written straight at the reader. I know because I counted — I went back through my own archive and compared the unpolished 2022 issues against the produced 2026 format. The scrappy version was opened at better than twice the rate of the polished one, and the gap was too large to be noise. This is an archive audit of my own writing, not evidence about anyone's results.

Four years ago I started writing a daily market letter in the least professional way possible. No content calendar, no format. Just a guy who traded all day, sat down after the close, and wrote up the one thing he actually believed about tomorrow. One name. One level. One claim you could check against the tape the next afternoon and tell me I was wrong. Last week I finally did what I tell every trader to do with their own journal: I stopped remembering and started counting.

What did I actually measure?

Buried in the newsletter's settings were seven old sections from 2022 — hidden years ago when I "cleaned up" the archive. That cleanup buried a few hundred posts nobody has seen since. I pulled the stats on them and ran them against what I publish now.

I didn't trust a gap that size on sight, so I tested it the way I'd test a strategy: a two-proportion test on the pooled open rates. It came back far outside anything chance produces — the difference is real, not a rounding artifact of a good month.

The honest limits, stated out loud, because they exist. The platform reports a "% opened" figure and a separate view count rather than raw send-and-open counts, so the test approximates its trials from views — a proxy, not exact opens data. The current-format sample is a small recent cohort, and I stopped short of the last handful of old posts rather than grind them page by page. A proxy and a small modern cohort could shave the result; they can't erase a gap this wide. The old material genuinely outperformed on engagement, and I'm reporting it the same way I would have reported the opposite finding.

None of this is a claim about trading. Open rates measure whether writing earned attention. They say nothing about whether the market calls inside it were right, and nothing about what any reader got out of them — that question is graded separately, in the ledger, where the misses stay on the page.

Why did the old posts win?

I reread every one of them. The pattern is not subtle, and it comes down to four things.

1. They committed. Every winning post was about one name, one setup, one claim. Today's format is a roundup — five names in the title, a little something about each. The archive is unambiguous about which the reader prefers: the roundup reads like a menu, and nobody remembers a menu.
2. They said something falsifiable. The old posts named a specific level and a specific expectation — this holds, or this thesis is dead. Statements you could grade. The single most-discussed post in the entire archive, with nothing else close, was exactly that: one clear, specific, checkable claim about a defined cycle event on a major index. Engagement didn't track with how much I covered. It tracked with how much I risked being wrong in public.
3. They talked to you. The old voice was second-person and had stakes in it. "Nobody is coming to save you" was a real line from that era. Somewhere along the way I traded that for narrator voice — "markets digested the data today." Accurate. Forgettable.
4. They were numbered. There was a serialized run in the archive — issues #2 through #13 — and readers showed up for the next number the way you show up for the next episode of a show. The series stopped at #13 in 2022. It shouldn't have.

One more data point that reframes the whole thing: the two most-viewed posts in the entire archive are both from 2022, when the list was a fraction of its later size. Raw reach isn't what changed between the eras. Engagement quality is.

How did I score the archive without fooling myself?

Two measures, deliberately, instead of one over-claimed number.

First, a within-section composite percentile for ranking individual posts. Each post got three percentile ranks computed against its own section only: views, open rate, and a weighted engagement score (likes, plus comments at 1.5×, plus new subscriptions at 3×). The open-rate rank gets down-weighted to 0.3× when a post has fewer than 15 views, because a "100% opened" on a couple of views is noise, not signal. The composite is the mean of the three. Ranking within sections matters: the 2022 lists were far smaller than today's, so comparing raw views across eras would flatter the present and tell you nothing.

Second, the one aggregate significance test above — the only one in the whole analysis, run once, on the era-level question. Everything else is ranking, and I kept the two jobs separate on purpose.

The composite produced a top-15% shortlist. That shortlist is where the real finding was hiding.

What was the winning ingredient — and why can't I keep it in a newsletter?

Here's the part a growth-hacker would leave out.

Every single post on that shortlist uses cycle-timing language. A defined cycle. A named point within it — bulls enter here, at this stage of a structure with a stated length. Numbered episodes of a method-driven series. The specificity that made those posts falsifiable, and therefore engaging, came from a particular way of timing markets — and that vocabulary was always a strange fit for a daily generalist letter. It deserved to be taught in full depth, not scattered through inbox fragments.

That's what this site is. The cycle-timing work that powered my newsletter's best-ever material now lives here at Cyclical Markets, where it can be laid out properly — the 80-day cycle and the rest of the framework, start to finish, at the methodology hub. The honest reading of my own data is that the best thing my newsletter ever ran was this site's material, living in the wrong building.

So no, the old winners don't get republished. The engine moved shops. But the chassis transfers anywhere: commitment, falsifiability, direct voice, and serialization are format, not method — and format is what the readers were actually voting on.

What makes a trading letter worth opening?

Distilled from the shortlist, five transferable patterns — the actual answer to why simple newsletters get read:

That last row changed how I publish. A falsifiable claim is only half the contract — the other half is coming back and grading it, hit or miss, where everyone can see. Here is what that looks like in practice, from the public calls ledger, outcomes stated:

2026-07-13, on a recorded session: I said that if CrowdStrike broke above the resistance zone a few percent overhead, you could expect continuation, or a breakout through the band about 12% above that close. The next session it traded through the upper band intraday and closed above it, finishing roughly 12% higher. Hit.

2026-07-09, same ledger: I flagged Rivian after it opened above the levels we track and ran early. It fell the next day and never recovered — about 3.7% lower by July 17. Miss, stated in print.

The most-discussed post taught me the audience doesn't want coverage. It wants skin. The full scoring rules I use for that ledger are written up in how an educator should score their own calls, and the graded multi-year result is the scoreboard article.

Questions traders ask

Does a higher open rate mean the 2022 trading calls were right?

No, and the distinction matters. Open rate is audience data — it measures whether the writing earned attention, not whether the market claims inside it resolved correctly. A post can be opened by everyone and still be wrong about the market; a correct call can sit unread. Grading the calls themselves is a separate discipline with its own rules — logged before the outcome, scored against a stated reference, misses kept in the record.

Why not just republish the hidden posts?

Three reasons. The individual trade write-ups are dated — a specific setup from 2022 has no live value, only evidence value about what format worked. The method vocabulary that powered the best of them now has a proper home here, where it's taught in full rather than in fragments. And the real asset in the archive was never the posts — it's the five patterns they proved, which transfer to everything written next.

Should a market letter cover five names or one?

My own archive says one, emphatically. The multi-name roundup format was opened at less than half the rate of the committed single-idea format, and the most-discussed post in four years of archive was a single checkable claim. A roundup feels safer to write because it can't be entirely wrong — which is exactly why it doesn't get remembered. Commitment is the product.

Is polish bad for a trading newsletter?

Polish isn't the villain — hedging is. The 2026 posts didn't lose because they looked produced; they lost because production came bundled with menu-style coverage and a neutral narrator voice that risked nothing. Substance that commits, dressed plainly, beat polish that hedges. If a post ships nothing checkable, no amount of formatting rescues it.

The archive verdict, in one line: readers open letters that risk being wrong in public. If you want to check whether the person writing to you deserves that trust in the first place, how to verify a trading guru's credentials is the five-minute checklist — and the cycle framework the winning posts were built on starts with the 80-day cycle.

Keep reading