Every month, InFlextion evaluates each manager and estimates its probability of outperforming its benchmark over the next 12 months.
The model analyzes 40 variables describing relative performance, consistency, peer behavior, stability, and the environment the manager is operating in.
40 variables. Five lenses. One Flextion Score.
The model starts with how a manager is performing relative to its benchmark, not simply whether its returns are positive or negative.
Flextion reads the shape and evolution of performance, not just today's trailing return number.
Performance does not happen in isolation. A manager improving while its peers struggle may carry different information than a manager rising alongside its entire category.
The model puts what is happening to the manager into context through five lenses.
Performance is never evaluated in a vacuum.
The five lenses are not five independent ratings averaged together. The model evaluates how the underlying evidence interacts. Improving momentum may carry more meaning when it is persistent, differentiated from peers, and occurring without a significant increase in downside risk.
How likely is this manager to outperform?
72% estimated probability of outperforming its benchmark over the rolling 12-month horizon.
Most traditional quantitative research ultimately tells an advisor where a manager ranks. Flextion is designed to answer a different question:
How strong is the evidence that this manager will outperform?
The Flextion Score is designed to be interpreted as an actual probability. If managers receive a probability around 70%, those managers should outperform at approximately that rate over time.
The number has meaning beyond its rank.
Manager outlooks are not static. Every month, the entire research universe is rescored as new performance information becomes available. As the evidence around a manager strengthens or weakens, its Flextion Score changes with it.
The score evolves as the evidence evolves.
The goal is to recognize changing patterns in a manager before those changes become obvious through traditional trailing analysis.
Finding a relationship in historical data is easy. Finding one that continues to work on data the model has never seen is much harder.
InFlextion uses two independent forms of out-of-sample testing to determine whether a variable deserves to enter the model.
A variable is developed using one period of history and then tested on periods the model has never seen.
Does the relationship persist across different market environments?
The model is developed using one group of funds and tested on a separate group.
Does the relationship work beyond the managers used to discover it?
Adding more variables can make a model better at explaining historical data while making it less reliable when conditions change. InFlextion is intentionally selective.
The final model uses 40 variables supported by millions of historical fund observations and persistent out-of-sample evidence.
The objective is not to explain everything that happened.
It is to isolate the evidence that has been most useful in estimating what happens next.
Across out-of-sample testing, predicted probabilities have historically tracked realized outcomes to within approximately two percentage points across the probability range.
Historical results do not guarantee future performance.
Every manager is evaluated using the same methodology each month, giving advisors a consistent framework for comparing opportunities across a large research universe.
Understand how strongly the evidence supports future benchmark outperformance.
See which parts of the evidence are strengthening or weakening the signal.
Track how the manager's outlook evolves as new information becomes available.
InFlextion turns patterns within manager performance into a consistent, forward-looking probability of what may happen next.