About
My name is Sébastien Assohou. Some people know me as Badoun.
I came to this work through coaching. Bodies, biological mechanics, what holds and what breaks under real load. At some point it became obvious that the discipline I applied to bodies, which is to look honestly at what actually works rather than what should work, transfers cleanly to other systems. To the systems that claim to predict financial markets, for example.
That is how Nyalai started.
What Nyalai does
Nyalai examines the evidence behind systematic investment strategies, against five published criteria. Its first operational framework is an adversarial calibration and validation framework for quantitative finance. In plain terms: when someone claims "this model has an edge, we should allocate capital to it," Nyalai tests that claim against five adversarial validation gates and documents what was measured. Verdict-form 002 was published as a refusal, and its verdict was withdrawn in October 2026 by correction notice CN-003, with no conclusion on the strategy; Historical evaluation 001 recorded the historical output of CalibrationJudge, NOT-CALIBRATED.
Each examination is designed to end in a dated, reasoned and traceable conclusion: it states which criteria were met, why any were not, which could not be computed, which were not tested because an earlier gate failed, and the limits of the analysis.
Nyalai does not sell investment advice. It does not recommend allocations. It does not predict anything on its own. It returns a binary verdict, CALIBRATED or NOT-CALIBRATED, with each finding traced to the evidence and the criterion behind it.
Why this project exists
Because the industry has a hole in the middle of it.
Quantitative finance runs on an uncomfortable asymmetry. The person who builds a strategy is also the person who validates it. When they validate, they are human, therefore biased, therefore optimistic about their own work. Research on multiple testing in finance shows what can happen when many hypotheses or configurations are tested. Harvey, Liu and Zhu (2016) conclude that many published factors are likely false discoveries. Bailey and López de Prado (2014) argue that a backtest cannot be assessed without accounting for the number of trials behind it, an input they note is virtually never reported in published backtests. Real edges exist. They are just drowning in the noise of false ones.
Someone has to have no interest in a strategy passing. An infrastructure whose only product is the refusal when the refusal is deserved.
What I believe
I do not believe in self-proclaimed edges. I believe in methodologies that expose their own failure modes.
I do not believe in pure financial theory. I believe in empirical checks, permutation tests, bootstraps, and cross-regime robustness.
I do not believe in methodological neutrality. I believe a validator must carry an explicit bias, the bias of default refusal, and that bias must be documented publicly rather than hidden.
The gate specifications and thresholds are published under Creative Commons Zero. The Refusal Doctrine lays out the reasoning. Anyone can reimplement it. Nyalai does not sell a secret. It sells disciplined application, conclusions traced to the evidence and criteria behind them, and infrastructure that regulators and allocators can inspect.
There is a second reason for publishing the methodology in full. Our discipline requires that a small number of independent validators eventually operate in parallel, each with its own methodology, each contestable, each auditable. We publish ours as a matter of institutional discipline, not as an invitation. We aspire to remain the reference, not the monopoly.
The name
The Nyala is an African antelope that lives on the boundary between forest and savanna. Two worlds at once. It never fully commits to either. It watches.
The male carries spiraled horns. The female is smaller and unarmed. Together they form an asymmetric pair where one gives the frame and the other gives the judgment. That is also how a well-designed probabilistic system should work. Someone builds. Someone judges. They are not the same person.
Nyalai. Nyala plus AI.
What Nyalai does not do
No allocation advice. No external capital management. No co-development of strategies with the entities it validates. That last rule has a name in our internal doctrine. We call it sovereignty rule seven.
Nyalai will never say publicly that a strategy is going to work. It can say that the process that produced it passed the five gates, which is a much weaker and much more verifiable claim.
Contact
Nyalai is in a pilot phase.
The published examples are records or constructed illustrations, not results in force. Verdict-form 002 was an external submission from a personal acquaintance of the founder, as its page discloses, and not a paid client engagement. It was published in September 2026, and its verdict was withdrawn in October 2026 by correction notice CN-003. It illustrates how Nyalai corrects its own work; it is not a verdict in force, nor evidence that the framework works. Historical evaluation 001, run in July 2026, records the output of an earlier calibration judge on the founder's own system and is not a verdict. Specimen R-01 and Specimen N-01 are constructed illustrations of a form, not verdicts.
If you allocate to systematic strategies and want to discuss what evidence a strategy should be able to show before you rely on it, write to contact@nyalai.com. Researchers and operators of systematic strategies are welcome too.
The "it" pronouns used above refer to a methodological discipline carried by Sébastien Assohou, founding operator.