HFT is a by product of the demutualization of exchanges, true or false ? If it's true as I suspect, then demutualization which should never have happened IMO was clearly a bad thing for markets.
I don't think I have any good advice for home traders to improve execution, and not particularly for less liquid equities since I have very little experience with them. Any execution improvements that I can think of really require advanced infrastructure. My best advice is to agree with RearedMetal's comments earlier in this thread, to lengthen the time horizons of your strategy. That makes execution less of a factor in your profit and lets you focus on aspects of your trading that will have more of a sustainable impact on your bottom line. Commissions for common futures can be about $0.05 per contract (on top of the raw exchange fees that are listed publicly). I've seen as low as $0.03. Fees for equities can be really small, even free in some cases. Typically you're looking at all-in paying $0.003 per share for removing liquidity and receiving $0.002 for providing it, and the clearing side of that is only $0.0001, the rest are non-negotiable exchange fees that are listed publicly.
That just shows how quote updates propagate from each equities exchange to others. Notice that each time an exchange's quote changes, it sends out a packet to each of the other exchanges. Then the box at the bottom maintains an updated snapshot of the best bid and offer available amongst all the exchanges at any point in time. It's actually quite inaccurate since they've adjusted the speed of each packet to arrive at all destinations simultaneously, whereas the reverse is true in real-life. I think that gives it a cooler psychedelic effect that's more attractive to the senses though
An HFT expert is welcome on this board. We have lots of questions 1) How active are HFT participants in ES market? 2) What are the prominent strategies applied by HFT firms while trading ES? 3) Is all the HFT in ES only against the S&P500 basket or is it also against other custom baskets (like basket of top 200 stocks or so). Please note that I am not referring to cash index-futures arb here. I am guessing many HFT strategies on ES must have something to do with the 500 stock basket albeit not in the sense of index-futures arb.? 4) How much is real liquidity in ES contract? So, at 931 EST, can a manual trader buy 300 contracts? If yes, will it affect the prize? How about at 1300 EST? Also, please look at this thread. And respond either on that thread or this thread if you have relevant answers. Thanks in advance. http://www.elitetrader.com/vb/showthread.php?s=&threadid=255417&highlight=es+market+microstructure
1) Very. Every HFT firm probably trades them in some manner. If not directly, they at least use it to price other markets. 2) All sorts. - Market-making i.e. trying to capture bid/ask - Liquidity-removing: Capturing small momentum moves. - Spreading. Trading ES vs. any combination of other markets. 3) I highly doubt that anyone trades the full basket of 500 stocks vs. ES. Typically you develop a tiered hedge schedule, where you hedge with more of the basket as your position gets larger. It's a compromise between fuller coverage vs. cost/complexity/latency as you try to cover with more stocks. Once you get get a certain point though, you're not getting much value trading AK steel to improve your coverage by a tiny fraction of a percent. Much of the market-making in ES is based on arbing vs. SPY at all of the equities endpoints. 4) I don't really understand your question. I don't know how much liquidity is fundamental and long-timeframe based if that's what your asking. As far as manual trader buying 300 contracts, it's obviously easy to do so by crossing bid/ask. If you're talking about resting a bid and getting it filled, that really depends on how much edge you're giving up. ES is priced very well by SPY and other instruments, so if a home retail trader is gets filled on a resting order, you can be sure that they're giving up edge in the near short-term (i.e. market has already or will shortly tick through that price, or other instruments like SPY are pricing through your order already). Your(IB's) feed is just wrong. Some trades are missing, some are interleaved incorrectly, and some pricing is just wrong (the locked market at 1427 that you have for a while for example). In that timespan there were hundreds of quote upates (~700 to be exact, between 9:30:22 and 9:30:24, and about 100 separate trade prints), and even the snippet that you post is obviously missing quite a few. Perhaps the most glaring inaccuracy I see in that screenshot is that the trades aren't aligned with the quote updates properly (that whole clip of trades, from what I can gather, should be shifted up about 10 lines to get in the ballpark). I have raw data captured from a stream connected directly to CME so I can see precision out to nanoseconds. So, long story short, your data is inaccurate, so don't draw any microstructure conclusions from it.
Thanks HFT for your detailed answer on ES. You actually answered my question 4 very well. Thanks. Is it possible for you to attach your data that you are quoting in the above post. Will like to compare. I knew IB data is wrong - but used to think they just provide a snapshot data and bundle all the missing trades in one trade. Seems there is a lot of missing things.
Is every order type you use in your trading available to every market participant including those who MUST trade through intermediaries? If not, which order types do you use the most in your trading? Would you object to a minimum resting time or a minimum trade state time (say 1 millisecond) for all trading activities? What is your own theory behind stories that report comments like ".... And summarizing it all: since the start of the New Normal 2009, Bank of America has had 962 profitable trading days, with just 97 days with trading losses: a 90.8% win record...."? Is that a fair market, a bogus story or something else? Can you equal that return? (Your reported numbers earlier wouldn't do it.) Kudos to you for taking the opposite side when so many write in defending HFT with little understanding of the issues involved.
Well no one *must* trade through an intermediary. So if you *choose* to do so (by not opting paying for direct market access), you're at the whim of the types of orders they support. For instance, many brokers probably don't support Icebergs or Hidden orders. In fact, many HFT firms don't even spend the time to support these orders. Almost all orders most HFT's use are of Limit order, with time in forces of IOC, DAY, or GTC, which everyone should have access to at any major broker. The big exception is in equities, where HFT's can use ISO orders since they guarantee to comply with Reg NMS regulations when routing with our own custom algorithms. The cost-benefit of such a setup is unfeasible to do at home. As long as it applies to everyone it's not that big a deal. I don't think it will help anyone though. Just make your eyes hurt less when looking at a screen, which is a waste of time anyway if you're trying to pick out anything that flashes that often. I know it's true and think it's fair, on the HFT side at least. BOA and HFT are apples and oranges though. BB's have a lot of edge from information access and customer order flow, among other things. I'm sure there's unfairness there, but I can't comment since I'm not familiar with the space. Libor fixing for example though, is one example of illegal activity. HFT firms are getting a return on investment on the capital that we invest in personnel and technology to build our business. When we make markets, there's also an intrinsic service we're providing, not unlike other brick and mortar middle men, or ticketmaster or a casino. HFT is also different from BB's in that any Joe Schmoe with just a decent bankroll can connect directly to a market and be on potentially equal footing with us. Then it's up to you to develop the algos and software that we use to make markets efficiently. Becoming a BB, however, is a completely different story. As far as returns, consider my numbers more closely, and keep in mind that any HFT's reported numbers encompass an entire plethora of trading strategies, not just one. If each strategy makes X per day, and the standard deviation on the return is 0.2X, and say the firm has 1000 of these independent non-correlated strategies (not true, but to an extent there is high independence between sets of strategies), what is the likelihood of a down day? In my 7 years of trading I can count the number of losing days for the firm on one hand, and most if not all were due to technical issues (positive sign instead of negative in code, or shifted decimal point).
I cannot, sorry. A) I'm too lazy to strip out detailed timestamps and other proprietary data we attach to it. B) We sign contracts not to disseminate market data when we connect to the exchange. It really looks to me like IB has got a bug in their pricefeed due to the time shifting. Each packet from the exchange comes with a sequence number (both LastTrades and QuoteUpdates) so they should easily be able to at least provide things in the right order, but they don't. So if you list the sequence numbers of the quotes/trades that you attached in the snapshot, it would look something like this: 1 2 3 10 11 12 4 5 6 7 8 9 13 14 15 And yes, they're also aggregating like you expect, which wouldn't be that big of a deal in and of itself, but combined with being out of order it is useless.
No worries. I am surprised to see that sequence number is so out of whack. Its great what you are doing actually. Sounds like a Ask me Anything kind of session Some more questions if you can please answer them. You mentioned that you have been working for 7 yrs at the firm. 1) What is your exact role in the HFT firm? 2) What has changed in the last 7 years in the HFT space in terms of markets traded, speed, profitability, average trades/second, average quotes/second, # of strategies deployed, infrastructure? I am not sure how much of this information you can share about the industry or maybe about your firm. Basically, I would just like to gauge how HFT space has evolved over the last 7 years especially in terms of technology, speed, infrastructure etc. 3) How many strategies a trader/researcher in an HFT firm develops in a year, on average? I would like to know this to get a benchmark to compare my performance in developing intraday strategies against superfast strategies developed by an HFT employee. Ofcourse, we are targeting different trade frequencies and my infrastructure is much worse compared to infrastructure available to an HFT employee. 4) What is the half life of a typical HFT strategy before it decays or gets very crowded due to competition. Thank You Very Much.