This is the part that turns a checklist into an actual finding. Comparing a 3,000-share order against a 100-share order and concluding that internalization failed because the bigger order cost more tells you nothing useful. Size alone is expensive to execute.
The comparison that matters is a 3,000-share order against other 3,000-share orders of the same type, in similar names, under similar liquidity and volatility, split by execution or routing path where that path can actually be identified. These comparisons are most informative when they are matched within the same symbol, order type, liquidity, and volatility bucket, and time-of-day window.
Once you do that, a lot of the variation caused by order difficulty can be controlled for, and the routing path becomes a more plausible contributor to whatever gap remains.
An illustrative case makes this concrete. Let’s say a manager’s 100-share orders in liquid large-caps come in around 1.5 basis points of effective spread, while 2,000-share orders in the same names run closer to 7 basis points. That gap alone proves nothing — larger orders are just harder to fill.
The real question is what comparable 2,000-share orders of the same type look like when routed externally under similar conditions. If those land around 6.8 basis points, internalization isn’t obviously the culprit; difficulty explains most of it. If this 2,000-share order lands around 3.5 basis points instead, the routing model for that size and liquidity bucket deserves a closer look.
