Our Wyckoff-gated buy list is up 2.71% per position. Here is the number we don't lead with.

August 13, 2026

Sixty positions of real money over two months. The closed trades show 91.8% wins and +3.63%. Including the eleven still open, the honest figure is +2.71% and 85%. Here is why we publish the second number, and what Wyckoff actually contributes.

We run a daily buy list on a funded account. It is oversold-in-high-volatility mean reversion. A candidate must first clear a structural gate: a current bullish read from at least one of three classical frameworks — Wyckoff, Smart Money Concepts or Dow Theory — plus a hard liquidity filter.

Here is the entire live record, and both ways of reading it.

The flattering number

Ninety-two percent winners. That is the number a marketer publishes.

The honest number

There are also eleven positions still open, and they are not doing as well:

The open book holds the two worst positions in the entire record: -8.99% and -7.10%. Neither has closed, so neither appears in the 91.8%.

Why that gap exists, structurally

Our buy list exits on a 50-day moving average touch — a profit target — or a 30-day time stop. The catastrophe stop is off by default, deliberately: the edge being harvested is the deep dip, and a tight stop cuts exactly the trades the strategy exists to hold.

That design has a consequence you can predict without seeing any data:

Winners hit the moving average and close, entering the statistic. Losers sit

until the time stop, so at any moment the open book is enriched with the trades

that are not working.

This is not a flaw we discovered in someone else's product. It is arithmetic that applies to ours, and it is why we publish the 60-position figure rather than the 49-position one. A win rate over closed trades is only interpretable when losses are forced to close on the same terms as wins.

Is it beating the market?

Profitable and better than doing nothing are different claims.

Average holding period is 10 days. Over the same span SPY returned +3.1% over one month and +4.8% over three, which is roughly +0.8% per 10-day window.

So per position, +2.71% against about +0.8%. That is a real gap and we are not going to pretend otherwise.

What it is not, yet, is proven edge, for three specific reasons:

• Sixty positions over two months is a small sample. Our own advisory audit

refuses to draw conclusions below 50 rows and three distinct days; sixty positions is past the first bar and nowhere near comfortable.

• These are higher-beta names than SPY. Deeply oversold mid-caps move more in

both directions, so excess return over an index is not the same as alpha. The correct benchmark is risk-adjusted and we do not have enough history to compute one that means anything.

• The unscreened version of this strategy fails out of sample. Backtested

raw, it returns a profit factor of 0.90 — a loser. It only turns positive through a fail-closed liquidity screen (ADV over $100M) and a regime gate. We wrote it up rather than burying it, because a strategy that needs two filters to stop losing is a strategy on probation.

So what does Wyckoff actually contribute?

This is where we have to be careful, because "Wyckoff" is doing less work than the name suggests.

Wyckoff's framework describes markets as cycling through accumulation, markup, distribution and markdown, with a specific vocabulary for the events inside a trading range — the selling climax, the secondary test, the spring. It is a structural read: it tells you where in a cycle a chart plausibly sits.

In our stack it is one of three confirmation engines, not a signal generator. A candidate must show a current bullish read from at least one of Smart Money Concepts, Dow Theory or Wyckoff — the job is to reject names whose oversold reading is a falling knife rather than a base.

Two details matter. We use Wyckoff's phase == accumulation, deliberately not its own bullish recommendation, because accumulation is the "it has based" signal we actually want. And markdown is excluded outright: that is the falling knife the gate exists to catch.

Our overlap analysis found it flags accumulation on roughly 24% of names the SMC and Dow gates miss, and overlaps SMC only 33% of the time. So it is genuinely additive in coverage.

Additive coverage is not the same as additive edge, so we built a specific test for that distinction with the threshold set before running it: the same oversold population, split by Wyckoff phase, on 10-day forward returns. Baseline for all oversold names is +4.66%. With the SMC/Dow gate, +5.87%. To earn inclusion, a Wyckoff accumulation read had to reach roughly +5.87% — matching the filter we already had. Landing near baseline would have meant it adds names without adding edge.

It returned +7.24%. It cleared the bar it was set, and it was added to the gate on that basis rather than on its reputation.

Pre-registering the number is the whole point. A threshold chosen after seeing the result is not a test — and this one could have failed.

What happened when we mechanised the rules

Separately from the gate, we extracted 130 mechanical Wyckoff rules from trading books and put them through our backtest mill.

Two of them ever accumulated 20 trades. Those two, across 213 trades: 57.7% win rate, profit factor 0.98 — a hair below break-even — and zero clearing a deflated Sharpe ratio of 0.95. Today exactly one of the 130 is active.

That is not a verdict on Wyckoff. It is a verdict on mechanising Wyckoff. The framework asks you to read the relationship between price, volume and time across a whole range and judge whether supply has been absorbed. Most of what makes that judgement work does not survive translation into "if RSI < 30 and volume > 1.5x average."

The honest summary: as a structural filter, it earned its place on a pre-registered test it could have failed — +7.24% against a +5.87% bar. As a set of standalone mechanical entry rules, it did not survive our testing. Those two findings are not in tension. They say the judgement Wyckoff asks you to make is real, and that the judgement is what resists being written as an if-statement.

Why publish the weaker number

Because you can check it. The 60-position figure includes every position we hold, priced today, winners and losers alike. The 49-position figure is the same strategy with the inconvenient trades still hidden inside it.

If we only showed you 91.8%, the first time an open position closed at -9% you would have learned something about us that we could have just told you.