Case studies
Every project can be shown from specification to test report.
Manual M15 execution caused late entries and inconsistent sizing, and real performance could not be reviewed.
Publishable results: Execution became fully systematic and the parameters are reviewable and extensible. Performance figures are published only together with their measurement conditions.
Prop account rules were not reliably respected during manual, high-pressure execution.
Publishable results: Risk rules are applied deterministically in code; the system did not breach its caps in the tested scenarios.
Manual breakout execution in the opening minutes of the session was error-prone under wide spreads and violent swings, with results varying widely day to day.
Publishable results: Breakout execution became fully mechanical, and execution metrics such as average slippage and order success rate are measurable and reviewable.
The existing grid EA kept opening positions through strong trends with no cap, pushing drawdown past the tolerable range.
Publishable results: The EA's core logic stayed untouched, and the guard layer did not breach its configured caps in any tested scenario.
Three MT4 experts shared one account, with scattered and inconsistent position handling, magic numbers and basket logic.
Publishable results: All three modules now run on MT5 through the shared order layer, with the behavioural difference against MT4 documented and reviewable.
Running three symbols at once while keeping total exposure in view: each pair behaved on its own, but on volatile days the combined exposure ran away.
Publishable results: Combined exposure never crossed the agreed ceiling during the test, and the desk could run all three symbols without manual coordination.
The NAS100 and XAUUSD signal was being taken by hand with a visible delay — in the first minutes of the session the entry differed by several index points.
Publishable results: Entry slippage in testing fell from a visible average to under half an index point, and the first month on demo missed …
Unwanted entries in the minutes around rate announcements were the single biggest drag on the results.
Publishable results: Across a year of data, entries inside news windows dropped to zero, and the status panel made it obvious whether the system …
The two indices are highly correlated; managing positions separately left both open through every reversal.
Publishable results: During the test the time spent holding open positions shortened, and the account never broke the prop rule on simultaneous exposure.
The team ran several challenge accounts with different caps and adjusted risk by hand — one human slip would end an account.
Publishable results: Four accounts now run from one bot, and across the period no account came closer than 30% of its rule ceiling.
Crypto volatility made a fixed pip-based stop meaningless, and the team's reporting was always out of date.
Publishable results: One build now runs three crypto symbols, the team's journal updates hourly, and the test drawdown came in below their manual version.
Every night the team drew levels by hand and entries differed between members; the missing shared rule caused recurring arguments.
Publishable results: The team's nightly decisions became one rule set, and in the six-year test over 90% of trades followed it.
