Delight Equity wanted systematic buy and sell alerts on Pakistani equities: the same signals a trader would read off a chart, delivered automatically, every trading day, without sitting in front of a terminal.
Two constraints shaped everything. The exchange publishes no retail order or write API, so automated execution was never available, and building toward it would have been a compliance liability rather than a feature. And the suite had to run on an ordinary Windows machine with no developer setup and no ongoing maintenance burden.
Technical depth
Parabolic SAR is path-dependent and cannot be reliably rebuilt from raw prices; our own reconstruction once disagreed with the reference chart on a live name. The engines now read the reference value directly, and moving averages are split-adjusted and verified against the chart to within roughly 0.1 to 1 point.
Anything launched from a tooling session is killed by session cleanup minutes later. Resident operation was solved with WMI process creation, logon-revival shortcuts and direct launchers, so the engines outlive whatever started them.
Each engine binds a dedicated localhost port as a single-instance lock; a second copy fails to bind and exits. The lock is per-machine, and the operational rule that follows from that is documented in the handover.
The backtester supports a proper out-of-sample split: warm on the first 70 percent of history, judge only on the last 30. It produced an inconvenient result, that across a strong bull market the default trend-following parameters beat buy-and-hold on only a minority of names. We reported that plainly and wrote the caveat into the delivered documentation rather than tuning until the numbers looked better.
Result
Delight Equity trade from these alerts every session, and report a win rate of up to 80% on trades taken from the bots.
This is the client’s own reported result, not a measurement we took. We have not independently audited their trade log, and we would rather say that than present someone else’s number as our own.
Note also that win rate and total return are different questions. Our own out-of-sample backtest, described below, found the default parameters beat buy-and-hold on only a minority of names across a strong bull market. A trend-follower can be right often and still trail an index it sits out of half the time. Both results are real, and we publish both.
The suite issues alerts, not advice, and no trading system guarantees a result. All trading carries risk of loss.
Capability used: Software Development
The systems we are proudest of started as the piece nobody else wanted to quote on. Tell us what it is and we will come back with a scope, a fixed price, and an honest answer about whether we are the right people for it.
