Research at Ostera begins with observation rather than with a model. Most ideas start as something noticed in the market: a pattern in how liquidity behaves around a certain kind of event, a recurring response of price to a particular kind of flow, or a relationship that seems to hold more often than chance would suggest.
01From observation to hypothesis
An observation becomes useful only once it is stated precisely enough to be wrong. The first step is to turn it into a hypothesis: what exactly happens, under which conditions, over what horizon, and what would be seen if it were not true.
02Is the behaviour repeatable?
Many market patterns are real but rare, or real but explained by something else. Before any testing, the question is whether there is a plausible mechanism behind the behaviour: who is trading, why they would behave this way, and whether that reason is likely to persist. A pattern with no mechanism is treated with more suspicion, not less.
03Testing against historical data
Hypotheses are then tested against historical data, with particular attention to the ways such tests mislead: trying many variations until one appears to work, using information that would not have been available at the time, ignoring costs and the effect of the firm’s own trading, and drawing conclusions from too few occurrences. A result that holds only under one set of parameters, or in one period, is not considered validated.
A hypothesis earns trust by surviving attempts to break it.
04From research to discretionary execution
Validated research does not become an automated rule. It becomes part of the context a trader brings to a decision: what to look for, what usually happens next, and the conditions under which the observation stops holding. Execution remains discretionary, because the conditions in which an edge exists are rarely identical twice, and recognising when they have changed is part of the judgement.
05Review
Research is revisited as conditions evolve. Behaviour that once held can fade as markets and participants change, and the framework is designed to notice when that happens rather than to defend past conclusions.
Insights are research notes published for general information and education. They describe market mechanisms, not the firm’s positions or results. They are not investment advice or research recommendations, and nothing in them is an offer or solicitation.