The 4-3-2-1 Method

We have designed a scoring system for each risk asset under our coverage, including SPY, QQQ, BTC, AAPL, AMZN, GOOGL, META, MSFT, NVDA, TSLA, PLTR.

The total score is 10. We assign:

  • 4 points to our ML model outputs

  • 3 points to the price technicals

  • 2 points to valuations

  • 1 point to seasonality

1. ML Model Scores

Everyday, we run a set of macro-driven, machine-learning algorithms ahead of the market open to deduce the risk-on versus risk-off regime for the specific risk asset. The machine produces a binary result of 1 (risk on) versus 0 (risk off).

For the purpose of our 4-3-2-1 method, we award:

  • 4 points to the “risk-on” result, and

  • 0 points to the “risk-off” result.

2. Price Technicals

Everyday, we examine the price chart using both the Elliott Waves theory and traditional technical analysis methods.

We award:

  • 1 point to the medium-term technical uptrend (months)

  • 1 point to the short-term technical uptrend (days)

  • 1 point if the asset is en route to our target but yet to reach key resistance

This section will be better explained in practice with example charts.

3. Valuation

We follow each individual asset’s valuation in the form of P/S (or EV/S) and P/E ratios, both in a historical context and relative to their growth rates.

We award:

  • 1 point if the growth-adjusted valuation is more attractive to peers

  • 1 point if the absolute valuation is below the asset’s historical 5-year mean.

4. Seasonality (SPY, BTC)

Using the past 10-year seasonality (where available) data, we assess the likelihood of a seasonal correction to SPY or BTC. The SPY seasonality is used for SPY, QQQ, and Mag-7 stocks due to their high positive correlations.

We award 1 point if the forward 10-calendar-day returns are historically positive for SPY or BTC.