The Go-Live Checklist

A strategy can have a genuinely excellent backtest and still lose money on its very first live day for reasons that have nothing to do with whether the underlying idea is sound. A position limit set in the wrong units, a signal that reads yesterday's close instead of today's, a symbol mapping that points at the wrong instrument, a risk system that isn't actually subscribed to the new strategy's fills — none of these show up in a backtest, because a backtest doesn't touch the live trading infrastructure at all. The go-live checklist exists to catch exactly this category of failure: not "is the idea good," which was decided earlier, but "will this specific piece of code, running against this specific infrastructure, do what everyone believes it will do."

A typical checklist covers several layers. Signal and data checks confirm the live data feed matches what the backtest assumed — same fields, same timestamps, same corporate-action handling. Execution checks confirm order routing goes to the intended venue, order sizes are in the intended units, and a manual test order behaves as expected. Risk checks confirm the strategy is correctly registered with every relevant limit — gross exposure, net exposure, single-name concentration, sector caps — so that a limit breach actually triggers an alert rather than silently sailing past an uncaptured desk. Operational checks confirm someone is actually monitoring the strategy during market hours, that a kill switch exists and has been tested, and that whoever is on call knows what escalation looks like.

A useful worked example: a mean-reversion equity strategy passes its backtest and paper-trading review, and is one day from launch. During the go-live check, the team places a single manual test order into the live venue at minimum size and confirms it fills, appears correctly in the position system, and is picked up by the firm's real-time risk monitor within the expected latency. It is — except the position shows up under the wrong strategy tag in the P&L attribution system, a mapping error that would have made the strategy's live performance invisible to daily monitoring for weeks. The checklist catches a bug that had nothing to do with the trading logic and everything to do with plumbing, exactly the kind of failure a backtest cannot see.

What this means in practice

The checklist is deliberately unglamorous — it is a list of switches, mappings, and test orders, not a re-litigation of the strategy's edge — and that is precisely its value. It is signed off by someone other than the strategy's own developer wherever possible, because the person who wrote the code is the least likely person to notice their own blind spot. Skipping or rushing it under pressure to launch on schedule is one of the most common, and most avoidable, sources of early-life strategy losses that have nothing to do with the strategy itself.