Any single indicator can be wrong. Any single dimension of analysis can mislead. But when six independent analytical dimensions point in the same direction — that's confluence. And the research shows that confluence works.
The Academic Case for Combination
Rapach, Strauss & Zhou (2010) published a landmark study in the Review of Financial Studies demonstrating that combination forecasts consistently outperform individual models in out-of-sample equity premium prediction. Even when individual models are weak predictors, their combination produces robust signals.
Gu, Kelly & Xiu (2020) pushed this further with machine learning methods, showing that ensemble approaches that combine multiple signals doubled the performance of leading single-signal strategies.
The principle is simple: each analytical dimension captures a different facet of market reality. Technical structure reads price action. Momentum reads acceleration. Volatility reads the environment. Flow reads institutional behavior. Fundamentals read company quality. News & macro reads the context. No single dimension sees the full picture — but together, they do.
How Confluence Works
At the core of Alpha Pulsx is a weighted ensemble model. Each of the six pillars produces an independent score on a standardized scale. These scores are then combined using weights calibrated to reflect each pillar's historical contribution to signal accuracy.
The result is a composite confluence score from 0 to 100 that determines three things:
- Direction — Buy, Sell, or Hold. Only fires when the composite crosses meaningful thresholds.
- Conviction — How strongly the pillars agree. 80% confluence means five or six dimensions are aligned. 50% means it's mixed. Below 40%, we stay neutral.
- Expected Duration — How long the signal is likely to persist, adjusted for volatility regime and catalyst proximity.
Why Independence Matters
The power of confluence depends on the independence of the inputs. If all six pillars were measuring the same thing in different ways, combining them would add no value. But trend structure, volatility regime, options flow, fundamentals, and news/macro are genuinely independent dimensions — they can agree, disagree, or be neutral toward each other.
When they agree, the probability of a successful signal rises dramatically. When they disagree, we know to lower conviction or stay on the sidelines. This is why our system issues Hold signals frequently — because the evidence often doesn't converge, and the responsible thing is to say so.
Signal Continuity
Confluence isn't a one-day snapshot. Alpha Pulsx tracks how signals persist and evolve over time. Each day's signal references the prior signal: "In our last note, we were bullish with 74% confluence." This continuity serves two purposes:
- It shows when dimensional evidence continues to support a thesis — the kind of persistence data that Odean (1998) research suggests matters for decision-making.
- It alerts you when confluence is breaking down — a gradual decline from 80% to 60% to 45% is a clear signal that the thesis is weakening, even if price hasn't moved yet.
Flip-Flop Detection
The system includes built-in flip-flop detection. If a signal changes from Buy to Sell and back to Buy within a short window, something is wrong — either the data is unstable or the market is genuinely undecided. In these cases, the instability is flagged and conviction is reduced rather than generating rapidly changing signals.
Pre-Market Signals, Updated at 10:00 AM
Alpha Pulsx generates signals in two phases each trading day. The first run fires pre-market at 6:00 AM ET, giving you a research-backed view before the opening bell — based on overnight developments, pre-market activity, and the prior session's closing data. The second run fires at 10:00 AM ET, after the opening range has settled, earnings reactions have played out, and the morning gap has resolved.
Why two passes instead of one? Because pre-market data and post-open data tell fundamentally different stories. Barclay & Hendershott (2003) showed that the first 30 minutes of trading account for a disproportionate share of daily price discovery — the opening range sets the tone, and signals generated before it can miss critical information. By refreshing at 10:00 AM, our confluence scores incorporate the opening range breakout or failure, early volume confirmation, and any gap fills — all of which meaningfully shift the probabilities.
But we deliberately stop there. We don't refresh every hour or every 15 minutes. Barber & Odean (2000) showed that the most active traders underperformed by 6.5% annually. Chague, De-Losso & Giovannetti (2020) found that 97% of persistent day traders lost money. More frequent updates would encourage overtrading — exactly the behavior the research warns against.
The two-phase cadence is the sweet spot: early enough to plan your day, refined enough to reflect reality, and infrequent enough to keep you in swing/position mode. By aligning with the 3–12 month momentum effect documented by Jegadeesh & Titman (1993), our signals operate on the timeframe where systematic strategies have the most predictive power — while the 10:00 AM refresh ensures you're not acting on stale pre-market assumptions.
Start Exploring
Now that you understand the six pillars and how they combine, explore the Confluence Dashboard to see them in action across the entire S&P 500 universe. Every signal you see carries the weight of six independent dimensions — and the research to back it up.