I got tired of trading formulas breaking, so I built a swing-trading framework around inflection points, scenario grids, and morning trend
I have a doctorate in Biomedical Engineering (specializing in Computational Biophysics), and my day job is building AI-based genetics models. For a while now, I’ve also been deeply interested in swing trading.
One thing that struck me early is how often trading formulas seem to work beautifully for a stretch of time, and then suddenly stop working. Markets shift, participants change, volatility regimes change, macro stress hits, positioning changes, and what looked like a durable edge turns out to have been a seasonal costume.
That bothered me enough that I started looking for what I jokingly called a “super-formula” — not some magical universal indicator, but a framework that could adapt to changing market conditions instead of needing to be reinvented every time the weather changed.
So I went down the rabbit hole.
I looked through 500+ indicators, strategies, trend structures, and market behaviors across different regimes, timeframes, environments, ticker types, and, yes, the kitchen sink.
What I eventually landed on was not one holy grail indicator, but a different way of thinking about the problem.
The big shift for me was realizing two things:
First, not every ticker has a meaningful setup every day.
Second, the same ticker can have multiple plausible next-day paths, and the morning trend helps determine which one is actually activating.
That led me to build the framework around three ideas:
- filter for inflection points
- map multiple probable next-day scenarios
- use the morning trend to resolve which scenario is coming alive
That first point matters a lot.
I do not think this kind of framework should fire on every ticker every day. Most days, most charts are not interesting. They are in the middle of nowhere, structurally ambiguous, or lacking the kind of setup where a scenario grid has real predictive value.
So the first job is filtering for what I think of as inflection points — moments where a ticker is sufficiently poised that the next day has a meaningful probability of resolving into one of a few recognizable paths.
Those inflection points can come from a lot of different sources:
- momentum compression or expansion
- volatility transitions
- higher-timeframe pressure building near key levels
- relative strength or weakness vs. sector and market
- failed moves setting up reversals
- exhaustion
- sector / index tension
- VIX backdrop changing how moves tend to propagate
Once a ticker reaches that state, the system uses 180+ features to characterize both the ticker and the environment around it.
The feature set pulls from things like:
- momentum
- volatility and volatility clustering
- multi-timeframe structure
- sector and industry context
- market / index regime
- VIX and broader risk backdrop
- extension vs. mean reversion state
- relative strength / weakness
- breadth / participation-style information
- ticker-specific historical behavior under similar conditions
But the real idea is not “more features = better prediction.” That road leads straight to overfitting, false precision, and spreadsheet-based spiritual ruin.
The real question is:
When this ticker reaches this kind of inflection point under these surrounding conditions, what kinds of next-day behaviors have historically shown up most often?
That’s where what I call ticker psychology comes in.
Different tickers behave differently under the same backdrop. One rewards continuation. Another loves to fake the breakout and roll over. Another needs sector participation to sustain. Another becomes extremely sensitive to how momentum and volume show up right after the open.
So I stopped thinking in terms of universal patterns and started thinking more in terms of conditional behavior:
- what kinds of setups this ticker tends to respect
- how it behaves near inflection points
- whether it usually resolves cleanly or sloppily
- whether early momentum tends to strengthen the move or flip the direction
- whether volume and opening character help distinguish between competing paths
That led to the current framework.
The model does not try to spit out one rigid next-day prophecy before the open. Instead, it builds a scenario grid around the inflection point.
That grid can contain multiple plausible next-day paths. For example, the same setup may historically tend to break one way if the morning opens with directional momentum and real participation, and another way if the open is softer, choppier, or fails to press through.
So the prediction is not fully formed before the bell.
The prediction comes together when the morning trend begins to reveal which branch of the scenario grid is actually activating.
That morning move is not just a yes/no confirmation filter. It is part of the prediction itself.
On some days, the same ticker at the same inflection point can have meaningful probability to move in either direction the next day. What matters is how the morning actually expresses:
- momentum
- volume
- directional intent
- and whether the opening behavior resembles one historical branch more than another
That’s the part I find most interesting.
I’m less interested in making a static prediction in a vacuum, and more interested in building a conditional framework where the setup is defined historically, but the market still gets a vote.
In plain English, the workflow is basically:
- find tickers at meaningful inflection points
- use a large historical feature set to characterize the setup
- build a scenario grid for the next day
- let the morning trend narrow the field and identify the active path
- then evaluate the trade through that lens
To me, that has felt more realistic than pretending a universal premarket call should survive unchanged once the opening tape starts speaking.
The broader lesson from working on this is that markets seem less like a static equation and more like a conditional system. The edge, if it exists at all, seems to come from understanding:
- when a ticker is actually poised
- what scenarios are historically plausible
- and how early live behavior selects between them
I’m not pretending this is a final theory of markets. If anything, the whole exercise made me more suspicious of elegant formulas and more respectful of context.
My bias at this point is that the market does not reward rigid formulas for very long, but it may reward frameworks that are conditional, state-aware, and humble enough to let the opening tape help determine which of several futures is actually unfolding.
I’d be curious how people here think about this in their own trading:
- Have you found that the biggest edge often comes not from predicting every day, but from recognizing the few days when a ticker is actually poised?
- Have you seen setups where the chart looked ready, but the opening behavior completely changed which path was live?
- Are there tickers you’ve traded long enough that you trust them to “behave like themselves” at certain moments, even under the same broader market backdrop as everything else?
That’s really the rabbit hole I ended up in. Leave a comment below with your thoughts.