The YMI Long-Term Strategy: Weekly MACD Divergence Plus a Discount Score

The most expensive mistake in long-term buying is treating a falling price as a bargain. An asset that dropped from $100 to $50 hasn't become cheap if the earnings or demand behind it fell by the same amount. The opposite error costs money too: waiting until the chart looks healthy and paying for a recovery that already happened. The YMI Long-Term framework tries to handle both with two checks that must agree before you buy, one for momentum and one for valuation. It is a rules framework to test, not a published track record, and nothing below should be read as a return forecast.

The composite below is an educational example, not YMI's published indicator formula or actual weights. It still needs a documented data source and testing before you treat it as a buy signal.

Check one: weekly MACD divergence

Most people use MACD crossovers, which fire constantly in trending markets. This framework uses divergence instead. Bullish divergence means price makes a lower low while MACD makes a higher low. Price is still falling, but the force behind the selling is fading. That is a hint, not a reversal signal on its own.

The working rules:

• Standard 12/26/9 settings. They're common, so you aren't fitting parameters to a past chart.
• Weekly chart only. Daily divergences show up too often to be useful for multi-month holds.
• Divergence must persist for two or more consecutive weeks. A single-bar divergence is easy to produce by accident.
• The MACD histogram should be contracting bar by bar. An expanding histogram means selling is still accelerating, divergence or not.

Check two: the discount score

The second check asks whether the asset is cheap against its own history, not just cheaper than last month. You build a composite on a 0 to 100 scale, where higher means cheaper. For an equity index or ETF, reasonable inputs are:

• Current P/E as a percentile of its own 10-year distribution.
• Price as a percentage above or below its 200-week moving average.
• Earnings yield minus the 10-year Treasury yield, as a rough read on relative value against bonds.

For a futures market, the inputs change to price as a percentile of its 3-year range, the shape of the futures curve (backwardation versus contango), and Commitment of Traders positioning, which is published weekly by the CFTC.

The framework's working thresholds are 65 or higher to consider buying and below 40 to finish selling. Those numbers are choices, not validated constants. Your percentile depends on the lookback window, how you define earnings, and how you weight each input. Write those choices down before you look at results, or you'll tune the score to fit the past.

The combined entry and a worked example

A valid entry needs all four at once: weekly bullish divergence for two or more weeks, a contracting histogram, a discount score of 65 or higher, and no clear fundamental deterioration (an earnings recession for an index fund, demand collapse for a commodity). That last check is judgment, and it's the one that keeps you out of assets that are cheap for a reason.

You then buy in thirds over three consecutive weeks. Here is a hypothetical: you plan to commit $9,000 to one ETF, $3,000 per week, at weekly prices of $400, $380 and $410. That buys 7.5, 7.895 and 7.317 shares, for 22.712 shares in total. Your average cost is about $396.26. If the first week was early, the second and third buys improved your basis. If price ran away after week one, you still own a third of the position. Splitting entries doesn't improve expected return, it reduces the cost of picking a bad single week.

Exits and the missing stop

Positions are reviewed weekly, and there's no intraday stop. The exit plan is in tranches:

1. Sell one third when the weekly MACD line crosses above its signal line from below.
2. Sell another third when price reclaims its 52-week moving average.
3. Sell the rest when the discount score falls below 40.

If price makes a new 52-week low after you buy and the score drops below 55, stop and reassess the thesis. That is a review trigger, not an automatic sale. Because there's no hard stop, size the whole position so you can sit through a further drop. On the hypothetical $9,000 above, another 30% decline is a $2,700 unrealized loss, and you need to know now whether you'd hold through it.

What you'd need to test first

Be honest about sample size. Weekly divergence plus a score above 65 on a broad index may fire only a handful of times in a decade, so a clean backtest of a few signals proves very little. A fair test would cover several markets and multiple cycles, hold out data you didn't use to choose thresholds, include costs and taxes, and record every signal that fired, including the ones that kept falling. A divergence that resolves lower is the main failure mode, and it will happen.

Tax treatment depends on what you hold. An ETF, a stock and a futures contract can all be taxed differently, so check with a qualified tax professional before you plan around any of them.

This framework is separate from the day-trading side of YMI. The VIP plan, which includes Intro plus model context, is described on the VIP tier page, and that model context is a different tool from this weekly process. For the surrounding process, see scaling up and compounding profits, market regimes, building a trading plan and consistency.