just a head up: battling out queue position is basically the most HFTish way to trade. There are hundreds of papers out there that describe how to model the FIFO queue. Just don't do it. The queue position you see on your screen is at least 1.6ms old which is the lag of the fastest monitor you can find...and I didn't even factor in your internet connection. Machines that are looking to optimize PiQ work in nanoseconds
I have heard that HFT's are typically aggressive in the order queue and quick to cancel orders therefore providing a greater chance for others to get to the head of the queue i know my statistics are correct, however i do not see how this would decrease my chance of getting filled or is it that it is only my chances of an adverse selection fill occurring and not the two way needed to earn the spread? In terms of internet connection i submit orders and rarely cancel i know that this means that that i might get adverse selection however in this paper it shows that only 25% of the time do you get a fill from an aggressive HFT which makes an average of 36 cents a contract whereas you have a 37% of getting filled by an opportunistic trader which loses roughly 50 cents per contract under this logic it makes sense not to cancel a vast majority of orders i also understand that aggressive HFTs will make a majority of their money picking of stale limit orders after announcements and therefore offer a chance to cancel because these announcements are generally scheduled, additionally i trade outside NYSE market hours where there are less other MM and fundamental traders less fundamental traders that typically exploit most MM into providing liquidity at a loss (excluding A-HFT) means less adverse selection and less Other MM means less competition aggressive HFTs struggle because order book quantities are typically 80-150 contracts at each level meaning that there is less liquidity to trade on and as a result provides better and more frequent opportunities to trade in a lower level order book additionally I trade ZN which can go for seconds without a trade and when they do occur they are usually not HFTs as the market is governed by a large amount of block transactions occuring in the CME BROKER-TEC cash market. Baron_Brogaard_Kirilenko.pdf (nber.org) This shows that after a day in trading ES PROFIT OF DIFFERENT COUNTERPARTIES HFT aggressive made an average of $21,952,215 HFT Mixed: $12,771,060 HFT PASSIVE: $502,338 OTHER MM: -$2,856,583 FUNDAMENTAL: $1,129,413 OPPORTUNISTIC: -$30,895,400 SMALL: -$2,603,088
I guess you mix things up a bit here. HFT are either battling for queue position to run a 0+ strategy aka if they have a good position, they stay, wait for a fill and flip it for a tick or they don't so they cancel. Orders are layered across all prices which requires a sufficient amount of margin in case the book gets swept. Most MMs are spreaders, however, and they run a multi asset portfolio. It could be basis trading or ETF arbs or an options dispersion portfolio, there are a gammut of choices. All you want to do is trade a range and opt for better fills via good PiQ but this has nothing to do with exploiting the book.
HFT's to my knowledge tend to be highly aggressive and their margins are very low with HFT's like Virtu having a 6% market share in equities but only making 2 million on average per day means that they have very low margins of course Retail traders are not on average going to make money from HFTs in forms of a HFT mark-up of roughly 0.0007% of the trade size or 1/20 of a cent per stock due to either wider than optimal spreads or adverse selection depending on order type market or limit retail traders can still make money an compete with them on some scales but they just need to emulate fundamental traders when they trade by finding statistical anomalies over time periods either seconds to months and ensuring they can overcome this mark-up. Either way in the near optimal markets HFTs compete to provide a little under 50% of traders will actually lose money to them statistically.
wasn't referring to HFT but other traders finding statistical anomalies is larger time frames the difference between microseconds and minutes on relative basis is the same as minutes to a fortnight the time period is not actually much of a difference given the only factor change is interest in some fair value equations the difference is not as much as people think especially as you go further out volatility only goes up by the square root of the number times period volatility so volatility is not much of a difference either. Lots of fundamental investors say that from one day to the next a company cannot be a full 1% different however they also might say that business prospects and values change far more than the 16% SQRT(256)*1% which you would deduce from the equation in short we think longer time frames have a far greater impact and difference than they actually do.