An SQX export can easily run into hundreds of candidate strategies. Before any portfolio-construction question can be answered sensibly, that library needs to be organised well enough to actually compare candidates against each other. This is unglamorous work, and it makes almost everything that follows faster and more reliable.
Consistent naming and metadata
The single highest-leverage habit is keeping naming and metadata consistent across an entire export — symbol, timeframe, direction, and strategy family all identifiable at a glance, ideally from the filename or an accompanying field rather than something you have to open the strategy to check. This sounds trivial until you’re trying to compare forty similarly-named files at 11pm and can’t tell which ones share a symbol.
SQX exports carry structured metadata already, and preserving it — rather than renaming files by hand in a way that loses that structure — pays off the moment you start filtering or grouping the library.
Know what you actually have: symbol, timeframe, direction
Before comparing strategies on their statistics, it helps to know the shape of the library itself. A quick pass to tag or group by symbol, timeframe, and direction (long, short, or both) turns a flat list of files into something you can actually reason about — and often reveals gaps or imbalances immediately, long before any correlation analysis.
Preserve genuine trade history
A strategy is only as useful as the trade history that comes with it. Keeping the full backtest trade log — not just the summary statistics — matters, because portfolio-level analysis needs to reconstruct a combined trade stream across multiple strategies, and that requires each strategy’s actual trade-by-trade record, not a single aggregate number.
Remove obviously weak candidates early
Not every strategy an SQX search produces is worth carrying forward. A first, coarse filter — removing candidates with clearly insufficient sample size, an unacceptable Return-to-Drawdown, or other obvious quality issues — reduces the library to a more manageable set before the harder comparative work begins. See Profit Factor, Expectancy, Stability and Trade Count for how to apply this kind of filter sensibly rather than on a single number.
Maintain enough variety — don’t over-filter toward sameness
There is an opposite failure mode worth guarding against: filtering so aggressively on one metric that the surviving library becomes narrow and self-similar. If every retained strategy shares the same entry logic because that logic happened to score best on the metric you filtered by, you’ve accidentally removed the behavioural variety a portfolio depends on. It’s worth deliberately checking that the surviving library still spans different timeframes, directions, and entry styles — not just that every survivor individually looks strong.
Avoid a library full of near-identical systems
Large-scale strategy generation can produce many variations on a small number of underlying ideas — the same entry logic with slightly different parameter values, for instance. A library full of these near-duplicates can look large and impressive while actually representing very little genuine diversity. Grouping strategies by their underlying logic family, not just their individual statistics, helps surface this before it becomes a portfolio-construction problem later.
Use structured comparison, not a spreadsheet by eye
Once a library reaches even a few dozen strategies, comparing candidates by scrolling through a spreadsheet stops being reliable. Consistent metadata and preserved trade history are what make a structured, tool-assisted comparison possible — sorting, filtering, and grouping by the dimensions that actually matter (quality, market, timeframe, direction, behaviour) rather than relying on memory or manual review.
Where this fits in the SIPS workflow
Importing SQX Strategies covers exactly which files SIPS needs from an SQX export and how the import process checks that a Databank and its trade files line up correctly. Dataset Management covers how an active dataset, import history and replacement are handled once strategies are in your workspace. From there, Strategies is where the validated library becomes something you can browse, filter and compare directly.
The practical takeaway
Time spent organising a strategy library before building portfolios is rarely wasted. Consistent metadata, preserved trade history, an early quality filter, and a deliberate check for surviving variety together turn a large, unwieldy export into a library you can actually make good decisions from.