By full spread, do you mean when bid is 20.92, ask is 20.94, you're paying the full 0.02 for immediate execution? I thought it would only be 0.01 from last price.
How do HFT strategies combat slippage
- Thread starter traderzhang
- Start date
By full spread, do you mean when bid is 20.92, ask is 20.94, you're paying the full 0.02 for immediate execution? I thought it would only be 0.01 from last price.
%%
Some got helped by location;
top 100 funds may have plenty good performers, but since they are listed by AUM + not performance=just a guess
Implementing the strategies in terms of trade limits
Yeah, you wish
. When you send a market order, you're paying for the entire spread, whether that's 0.01 or 0.25 (like right after FOMC meetings). If you dont' wanna pay the spread, use limit orders. It's as simple as that.Actually, in stat arb making roughly bid/ask or even less is not uncommon for the first research pass. Assuming the signal is real, either because it's based on some sort of a real life effect or because you verified it somehow (e.g. it's forecasting methodology and you have an R^2 of some sort), you can take it to the next step. A robust portfolio formation methodology with some sort of hysteresis can improve your PnL per trade value to something reasonable. For simple strategy where each signal is generated per asset, it can be as simple as a signal band. PnL quality (Sharpe/Sortino/total pnl) will get worse, but PnL/trade will increase.
It's best to think about your execution quality in two stages. There is slippage, which the difference between the price used for your signal formation vs the benchmark price. Then there are transaction costs, which is the difference between the benchmark and the actual execution price. The latter will include the actual costs (spread and the vig) plus your impact if you are trading large enough.
Back when we traded in 1/16, 1/8, 1/4, 1/2 up to 1 pt etc.. I would watch when a specialist widened the spread on NYSE and AMEX then I would slip in with a buy and sell limit order (both) within the spread and scalp out a cut. Word quite well until…..they went from fractions to decimals and bam that strategy evaporated cause the spread tightened too tight.Yeah, you wish
IMO the only way a retail trader can maneuver around the HFTs is by what I call manual HFT trading. That is min scalp …..1 point.
I'm currently reading "Dark Pools" by Patterson where he documents this shift. His story references the "SOES Bandits" as the first adopters of computer based trading having an edge over stale NASDAQ market maker quotes. In a way, the computer democratized trading from "its who you know" to "its what you know."
Still, there are always insiders and outsiders looking in.
Microseconds is so last decade, it's nanoseconds now. You can get stable 20-25 mike median latency form a commercial system these days and the cost would be fairly reasonable (about 4k for the gateway and another 1-2k for your own colocated server). Even FPGA is a commodity these days, you can buy/rent trading boxes with pre-programmed FPGA boards that will do things like parse market data.HFT firms have their servers running in the exchange and pay a lot of money for that benefit. They also develop trading FPGA boards to shave off microseconds from the reaction time. It's quite a different ballgame
In any case, back to OPs question - in a medium-frequency setting, a low PnL/trade is usually not driven by latency but rather by low alpha in the signal.
That's a load of bullshit. What you describe existed for a brief period on NASDAQ Level II. It was never an edge in trading NYSE and AMEX stocks precisely because you had to send your order to the specialist.
If what you describe was fact, firms like Bright Trading that traded only specialist markets at the time would have been all over that strategy. I can still remember Bob getting irritated over the phone when I asked why his firm didn't offer trading in NASDAQ stocks. He said there was no edge because there was no specialist. He had just tried to sell me on their "trading with the specialist" opening orders strategy. I decided to go with Momentum. Fortunately it was not the Atlanta office. Not many will get the meaning of that reference. I had a laugh about a year or two later when I saw that BT began offering NASDAQ stocks.
Not to rag on the Bright brothers too much, but Don Bright used to come on here and say you couldn't make money trading index futures because futures led stocks and there was nothing that led futures. He challenged anyone to show up in their Las Vegas office to trade in front of him and prove him wrong.
Right before the pandemic, a buddy of mine who moonlights as a backgammon hustler got into an argument at a gathering about trading. The guy he was arguing with, who said he was a former prop trader, must have traded with BT because he said the same thing about index futures--they lead stocks but nothing leads futures so it's impossible to get an edge. He wanted to come over to my house and watch me trade so I could prove him wrong.
Rest in peace, Don.
Yeah, I was there in the thick of it. Unfortunately, I wasn't quick enough to make money using that strategy.
Back then, the specialists had no better technology than us dumb money. But the balance is completely tipped in favor of Wall Street now. You can't scalp like you can back in the 90s. It just won't work. What you need is an edge, a niche of your own.