University of Penn HFT Paper and Bad Assumtions

A friend shared this paper with me, pointing out that HFT is really not a big deal, as the amount that they make employing their strategies is nowhere near as big as some folks’ estimates of $21 billion, or $8 billion. Their study can be downloaded here:

I will just copy from the first page their assumption on which this paper’s conclusion is based, and add a some commentary:

Despite the growing controversy over HFT, there appear to be no objective, large-scale empirical studies of the potential profitability and impact of HFT. The purpose of this paper is to provide one such study. (I guess we can ignore Tabb, Rosenblatt, ITG, Jeffries, etc.)

In our study we make a crucial distinction between passive HFT, in which a HFT strategy places limit orders that are not immediately marketable, and thus act as providers of liquidity to the market, and aggressive HFT, in which only market orders are used, and thus the HFT must pay the attendant execution costs of crossing the bid-ask spread. In this study we focus only on aggressive HFT, and we shall argue that this is the variety of HFT that should be the primary focus of any concern, since the presence of passive HFT can only provide price and liquidity improvement to any trading counterparties.

Where to begin? I guess fail? Before they even get into their methodology they have excluded any non-taking-the-offer or non-hitting-the-bid orders and executions. So in their example, any order that has not taken the offer must not be nefarious. So if an HFT using a datafeed from an exchange determines there is a large buyer or seller in the marketplace places a limit order, hidden order, middle of the spread dark pool order, 1/1000th of a penny jump ahead dark limit order, slide and hide order, etc… they are excluded from this study? Really? And if  HFT prop limit orders are residing in a dark pool, waiting for the algo they just ran ahead of  to pass through and take them, that does not count in their study? Really? And if a stock has a 3 cent spread, and an HFT runs ahead of a displayed bid-side order by 1 cent, with a limit order, it does not count in their study, as it is still “passive HFT” because it used a limit order and not a market order? Really?

Just wow.

Have fun reading the rest of it.