The definition of algorithmic trading under MiFID II is extremely broad, essentially capturing any functionality that determines any parameter of an order automatically. That begs the question: What order flow qualifies as not algorithmic? Hand on heart, how many traders can claim to exclusively use native exchange order types and not rely on at least one of the friendly little helpers in the form of a stop order, an iceberg, delayed start or order expiry? Probably very few, and possibly none, right?
Regulators might, then, be in for a surprise when they receive their new MiFIR transaction reports. With firms required to report either the person OR the algorithm responsible for the investment and execution decision, it doesn’t test too many powers of prediction to assume they’ll see way more algo IDs than traders’ national IDs populating those reports.
When it comes to equity market data, the picture will become even more skewed, with a requirement for trade reports to include an algorithmic flag. If any trade is to be deemed algorithmic if it includes one (or more) algorithmic order(s), we can be pretty certain that close to 100% of all trades will be so categorized. But don’t be fooled by those statistics; they are not evidence of machines taking over our markets; rather they are a reflection of the fact that smart humans are employing the tools available to them to perform better.