How signals are scored
Every signal run ranks your trading universe (the list of symbols your strategies may trade) with a four-factor tilt. Each symbol gets a raw value per factor, each factor’s scores are put on a common scale across all your symbols (so a momentum score and a value score are comparable), then weighted by your active strategy version and blended into one rank.
The four factors
| Factor | Measures | Computed from |
|---|---|---|
| Momentum | Recent strength | Price change over roughly the past 3 months |
| Low volatility | Calmness | How much the price actually swings, as a yearly rate, flipped, so calmer stocks score higher |
| Value | Cheapness | Price compared with the past year’s earnings, flipped so cheaper stocks score higher; only profitable companies qualify |
| Quality | Business strength | Return on equity |
A factor that can’t be computed for a symbol (missing company financial data, say) is treated as neutral rather than penalized. A symbol with no usable price history at all is excluded from the run: the tilt’s core is price-based, so a symbol we couldn’t price isn’t ranked on company financials alone.
Buy, hold, sell
The composite ranking splits into bands: the top of the universe reads buy, the bottom sell, and the middle hold. No order is generated for holds, and existing positions are left alone. A universe where every score ties (tiny lists, missing data) collapses to all-hold rather than forcing trades. In a short universe (just a handful of names) every name scoring above the universe’s own average reads buy, so your cash spreads across all your good names instead of piling into the single best one, while only clearly-below-average names sell.
In a larger universe there’s one extra guard on selling: a name near the bottom of the ranking is only marked sellwhen its own momentum is actually falling, not just because it placed last. Ranking is relative — if every symbol you hold is still climbing, the lowest-ranked one is still climbing too, and force-selling it just to rotate into a fractionally-higher name would churn a perfectly good holding. Requiring real, negative momentum before a sell keeps a strong book intact.
Tuning the factors
Each factor has an on/off switch and a weight, set under Advanced: tune current strategy on the Strategies page. Tuning adjusts the strategy you’ve selected in place: once your settings differ from the preset you started from, its card reads (tuned), and Reset to default restores that preset’s original factors and weights. The weight decides how hard a factor pulls on the final rank.
Turning a factor off removes it from the decision entirely: it stops affecting the ranking, as if it weren’t there. The factors still on share the whole blend between them: with all four on and weighted equally, each is a quarter of the decision; switch one off and the other three become a third each. Because the weights are relative, the ones left on scale up to fill the gap rather than the blend shrinking. (You can’t set a weight for a factor that’s off: its weight box turns off with it.)
At least one factor has to stay on, and an enabled factor needs some weight: a strategy with every factor off (or with every enabled factor at zero) has nothing to rank on, so it won’t save.