patrick.net

The Sharpe ratio is supposedly much better with HFT.

http://en.wikipedia.org/wiki/Sharpe_ratio

SkyNet is in control of the Stock Market.

Maximum *attainable* Sharpe ratio for any given time series is MUCH better at the higher frequencies.

But then a black swan is always lurking. And he is very excited about the higher frequency of action.

LTCM enjoyed a great Sharpe ratio until it did not.

The more frequently you disturb a complex nonlinear system, the more likely something unexpected will happen in a given time frame.

thunderlips11 says

SkyNet is in control of the Stock Market.

Complexity in nonlinear systems is WAY scarier than SkyNet.

Peter P says

The more frequently you disturb a complex nonlinear system, the more likely something unexpected will happen in a given time frame.

Word.

The problem is that our policy "makers" focus too much on stability leading to a loss of resilience in the macro economy.

Peter P says

The more frequently you disturb a complex nonlinear system, the more likely something unexpected will happen in a given time frame.

Very well put!

Damn, I was just about to read the article and then I noticed it was Crapipedia. No point since it would take more effort to separate the misinformation from the real information than doing a Google search.

I said that was the start. I am curious (on another thread) what errors you have found in Wikipedia.

I am a high frequency trader at a hedge fund in NYC. My turnover is on the order of 0.5-1% of the US stock market. FWIW I've read the Wikipedia entry and found it to be generally accurate.

BTW, when looking at numbers like the "70% of turnover is HFT" you need to distinguish between agency algorithmic trades (a.k.a. "real money") and efficiency traders like myself. There is only so much market inefficiency to go around so my type of trading is self-limiting; it can never become too large a portion of total turnover. If it did we wouldn't make money and would back off our size until it started working again.

The majority of algorithmic trading is just executions for institutional investors. In the old days they used sales traders for hand execution. Now they put it into an algo, but it amounts to the same thing.

Thanks for the info, krusty.

One thing I don't understand are claims that HFT lifts up the value of the stock market.

Well I'd say it certainly doesn't raise the "value" of the market. In fact, IMO it has no effect whatsoever. Really the only thing that can move markets over any medium to long period is fundamental investment by buy-and-hold investors. Everything else just washes out: day traders, market makers, algos like mine. We always have a sell for every buy and vice-versa so the net impact is zero.

I guess you could make the argument that higher liquidity makes a market more valuable since it is less risky, but really over the long run the only thing that really matters is the cash flow provided by the investment. This isn't affected at all by how you trade it.

Anyway, why would people want the market level to be anything other than a true reflection of future value? All that does is provide a one-time windfall for current stockholders at the expense of future returns. I'd prefer if it was consistently undervalued so I could sit on a portfolio and collect the cash flows (dividends, capital reinvestment, etc.) Its like houses or hamburgers: you want prices to be low, not high.

Thanks. I suspected such, but I had been reading otherwise.

http://finance.yahoo.com/news/high-frequency-trading-worse-thought-165633554.html

Worth noting. I am still learning about this...

http://finance.yahoo.com/news/one-algorithm-made-4-quote-204830799.html