Saifeddine Boumaza runs backtested, AI-driven models against your treasury data to recommend where excess cash can move without breaking your liquidity requirements.
Most French SMEs keep working capital in low-yield accounts because the alternatives look complicated or risky. That caution has a cost: the gap between what idle cash earns and what a disciplined, tested allocation strategy could return compounds every quarter it is left unaddressed. Traditional savings products are not built to respond to changing market conditions in real time. Saifeddine Boumaza exists to close that gap without asking a finance team to become traders.
The system does not guess. It analyzes historical price and liquidity patterns, runs them through predictive models, and validates every recommendation against past market cycles before it reaches your dashboard.
The engine forecasts short-term liquidity needs from your transaction history, then proposes allocations sized to leave working capital untouched.
Every strategy is run against historical data covering periods of rate increases, contractions, and stable growth before it is activated on live capital.
Recommendations are constrained by pre-set liquidity floors and volatility ceilings, so the model cannot propose moves that compromise short-term solvency.
As new transaction and market data arrive, the model updates its recommendations rather than relying on a static quarterly review.
Backtested results are displayed as a rolling comparison between a static cash position and the model's recommended allocation over the same period. Both lines use identical starting capital and identical market data, so the only variable is the strategy applied.
Strategies are tested on historical market data using walk-forward validation: the model trains on one window, is tested on the next unseen period, and repeats. This avoids fitting the model to data it has already seen.
Every backtest enforces a minimum liquidity reserve, a maximum single-position exposure, and a stop condition tied to volatility thresholds. These constraints apply identically in live recommendations.
Link bank accounts, accounting software, or ERP exports through a read-only connection. No fund transfer happens at this stage.
The engine studies inflows, outflows, and seasonal variation to establish how much of your balance is genuinely idle.
You receive a ranked set of allocation options, each with its historical performance record and risk parameters shown side by side.
Nothing moves without explicit approval. Once approved, the system tracks the position and flags any deviation from expected behavior.
Data is processed and stored on infrastructure located within the European Union, in line with GDPR requirements applicable to businesses operating in France.
Recommended allocations are selected specifically for their liquidity profile. Each proposal states its access terms before you approve it, so there are no surprises later.
No. Saifeddine Boumaza generates recommendations and connects to execution through your existing banking relationships. Fund custody remains with your bank at all times.
Bank connections are read-only wherever supported, data is encrypted in transit and at rest, and access is restricted to the accounts you explicitly authorize.
The model recalibrates continuously and will flag positions that fall outside pre-set risk parameters rather than waiting for a scheduled review.
The workflow is built to require a single approval step per recommendation. No trading desk or dedicated treasury analyst is required to operate it.