# Tag Info

9

Using months of proprietary data that labels participants by their participant ID, it has been found that during periods of significant volatility, the composition of HFT participants in the book remains mostly constant as a fraction of the total BBO composition. What really changes, it was found, was that the fraction of low-frequency traders aggressing on ...

6

Financial modeling is often considered as a mixture of art and science. That is a way how you model your data depends on you. You can choose several approaches, for example: calculate max - min price for a given minute data - a very simple approach, calculate the standard deviation of minute-by-minute stock data, calculate GARCH family models for measuring ...

6

Definitely check out Quantopian and Zipline. Quantopian provides a free research environment, backtester, and live trading rig (algos can be hooked up to Interactive Brokers). The algorithm development environment includes really handy collaboration tools and an open source debugger. They provide tons of data (even Morningstar fundamentals!) free of charge. ...

4

Quick summary: Your model should still be well specified, as long as: 1) You do the analysis on a heavily traded asset, e.g. IBM on NYSE, and 2) You use heteroskedasticity-consistent standard errors in your estimation framework, e.g. White's standard errors. I'm going to start the long answer by re-stating the question to make sure I've got it right. Let ...

4

Very interesting question. I am not an expert on the subject, however, I was able to find a collection of papers on the subject that should get you started. Here is a good and very informative paper that walks you through several tick by tick volatility estimators that seek to reduce the volatility imposed by market micro-structure: Efficient estimation of ...

3

In my experience HFT has to balance the reward of any strategy with risk. In the case of a news-based trading strategy, the risk can be enormous, which means the algo will need a very high expected profit in order to trade the news. After important news events, volatility skyrockets and persists for some time (sometimes even days). If the market were able ...

2

I have created some Fourier Analysis of stocks here: http://www.gregthatcher.com/Stocks/Default.aspx I turn the raw data into a series of sines and cosines, show the Fourier approximation as a graph, and then allow you to "turn off" the various sines and cosines, so that you can see how the various "frequencies" contribute to the graph of the stocks values. ...

2

QuantConnect provides an open source, community driven project called Lean. The project has thousands of engineers using it to create event driven strategies, on any resolution data, any market or asset class. Our system models margin leverage and margin calls, cash limitations, transaction costs. We maintain a full cashbook of your currencies. Its about as ...

2

I'll address your questions in order: 1a) For TSRV constructed using high frequency returns from NYSE market open to market close on a single day, the output should be numbers on the order of magnitude of 1e-4 to 1e-5. In other words, your numbers look about right. I got these number from calculating TSRV for IBM data myself using Kevin Sheppard's MatLab ...

2

1) Spurious autocorrelation of non-synchronous trading data was analyzed in this article: http://www.amazon.com/An-econometric-analysis-nonsynchronous-trading/dp/1245789457 During some time intervals a lot of trades occur and during some nothing happens(so prices are stale). So serial correlation of traded prices may be present but this may be due to stale ...

2

First of all, I do not believe the "optimal smoothing" of an estimator (like the mean or the variance) and the "regression case" are the same. The smoothing of an existing estimator (like mean or variance in the blog post), is an univariate problem, where the regression is a multivariate one. In the regression case, you should be able to change the ...

2

Features could include: Bid-ask spread Bid-ask volume imbalance Signed transaction volume The sign in the Signed transaction volume is positive if the buyer has issued a market order and negative if the seller issued a market order. A great introductory plain English paper on high frequency trading machine learning applications can be found here. A ...

2

It is all a matter of frequency. For instance if you want to get annual realized volatility you multiply your last expression by $\sqrt{(N*251)}$ or the second to last expression by $\sqrt{(251)}$. In other words, your last expression is the 5-min realized volatility whereas the second to last expression is the daily realized volatility.

2

The classic text for machine learning is 'The Elements of Statistical Learning' by Tibshirani et al. I believe the term "data mining" is often used synonymously with "machine learning".

1

The code you posted is wrong since you do not model the time series behavior of the up/down process (ie if you have 10 up move and consequently 10 down move it is not the same as the opposite ie 10 down and after 10 up..). I would recommend you to use standards Arma Garch models apply on returns instead of modeling the process of up/down. These are (at ...

1

I second Tibshirani's book. There is an another edition you can download free on internet : http://www-bcf.usc.edu/~gareth/ISL/

1

There are many ways to calculate the volatility. timeframe does not metter. it can be monthly quarterly or daily data. You can call them as volatility metrics. Volatility Metrics Volatility is the degree of trading price over a specific time window. Historical volatility is the degree of price changes of past market prices.Volatility indicates the risk ...

1

By design, market makers do not exacerbate volatility because their trades are, as a whole, net passive.

1

The 50 cent bid was certainly a LMT order and the exchange will not match a 50 cent bid with a 90 cent offer. And the past tense of "front run" is "frant ran".

1

As a beginner in AlgoTrading QuantConnect and Quantopian are great for practice and improving your skills but for a serious Algo Trader , they are basically useless. An Algo Trader requires flexibility to investigate trading ideas and add or remove libraries or parts of the system that do not work. You need to automatically and constantly reevaluate your ...

1

http://bluemountaincapital.github.io/Deedle/ Disclaimer: I haven't used this.

1

From the Nasdaq page, IMBALANCE-ONLY CLOSE ORDERS Provides liquidity intended to offset on-close orders during the Closing Cross. Must be priced (limit), no market IO orders. IO buy/sell orders only execute at or above/below the 4:00 p.m., ET, bid/ask. They simply mean they were +\$0.01 or at \$23.56 from the price on their sell ...

1

I browsed through the work and this is what I see: the lhs $r_{t+1} + \cdots + r_{t+H}$ is the sum of log-returns after $t$. the rhs is indexed by $t-i, i=0, \ldots, H$ thus this has something to do with the past before (and at) $t$. Thus the regression models the future ($r_{t+1} + \cdots + r_{t+H}$) dependent of the past where only PCA projections of ...

1

HF data have a lot of auto correlation so common models to deal with this problems are ARFIMA, FIGARCH, Fractional Integrated GARCH. Engle recently propose the multiplicative components GARCH for high frequency data, which can include a mean model like and ARMA. In this post they explain how to implement it in R with the rugarch package, it takes some time ...

1

First define a quote: this is the bid and ask (price and volume). when any of them 4 change, it is said the quote changed. We all know what a trade is (nevertheless note if you send a liquidity consuming order of 100 on a queue made of 50+20+30, it generates 3 trades). You can play with statistics (like order-to-trade ratio, not quote-to-trade), on te SEC's ...

1

In trading you need to make a lot of simple computation of a very large flow of data. FPGA are perfect that for. It is typically FPGA that will host marketfeed handler (see NOVASPARKS website, or ACCELLIZE) ; analytics computations ; risk computation (see ULLINK solution for instance). For more, this generic article is not that bad: Introducing ...

1

FPGA's are used to run the latency sensitive HFT strategies. They can also be used solely for parsing whatever protocol is in use (FIX, ITCH, etc..) and routing the decoded objects to a CPU for number crunching. They can of course be used for anything else but these two uses are what is most common now.

1

This is a very good observation that I wrote about in my undergrad studies. I also believed that markets were efficient but not precise. I used the example a few years back regarding a tweet (roughly after the Boston bombings). The tweet was regarding terrorist attacks in which markets fell sharply and then recouping all the gains as news later indicated ...

1

https://mechanicalmarkets.wordpress.com/2015/02/16/protecting-client-interests-anonymity-in-us-equities/ does analysis similar to the question here. It examines the post-trade performance of orders grouped by their MPID (only UBSS and anonymous orders had enough data points to report). It also looks at market impact upon the addition of a new order. ...

1

Exchanges provides the following six timestamps: Gateway In Timestamp-T1. Time at which the order was received by the Gateway from the members TCP connection. Gateway Out Timestamp-T2. This is the time when the order was dispatched by the Gateway to the Matching engine. Matcher In Timestamp-T3. This is the time the order was received by the Matching ...

Only top voted, non community-wiki answers of a minimum length are eligible