coming from someone who has had a hand in actually implementing these types of concepts in real production , I can tell you there is a big gap between ideas and production(i.e. concept and reality). There are hardware limitations that are limiting for many reasons outside of simple physics of the devices and architecture that go into making them, and many standards that are already in place that are the consensus of the big players in this industry. Even if the conceptual idea has merit, it must still conform to hardware/software protocols that are in place. Very big gap between ideas and implementation. Tell him to get a patent if it's that great, and if the idea really is that profound, I'm sure the offers will come knocking. He can try building a prototype, but if it's at the IC level, I hope he has some serious cash and resources to burn.
I thought QUANT funds were doing miserably this year. One of the funds at Renaissance (Long Island ) is underperforming the S&P 500 by 18%. I've also read arguments suggesting that a significant portion of the market's recent rise has to do with all these quant funds desperately liquidating their losing short positions.
Let´s wait and see how they are going to liquidate their desperate long postions in let´s say 6 weeks time when S&P is approaching new lows...
Markets evolve over time, once these guys start their algorithms, that will establish new patterns that other programmers may detect and try to pick off, and so on in the never ending war. They may even produce patterns that the little guy can detect, who knows. Are these programmers really better traders than most and will know exactly how to program a fast response to news? They will predict market response and not get whipsawed? Seems pretty tough to me. Is this more 'efficient'?... Efficient Market theory is a very flawed premise for many reasons but first because it assumes markets can actually reach equilibrium. There is no such thing as market equilibrium- there is never a time when all bids and offers are satisfied and all participants agree on a stable price. Look at price charts, they constantly move around. For a market to be in equilibrium then everyone's percieved value is the same, that's never going to happen. Karl Marx also thought that he could scientifically quantify value. I think high frequency trading is likely to add more volatility as they try and outdo each other and that's about it, the people who get picked off are the ones who can't adapt.
Markets evolve over time, once these guys start their algorithms, that will establish new patterns that other programmers may detect and try to pick off, and so on in the never ending war. They may even produce patterns that the little guy can detect, who knows. Are these programmers really better traders than most and will know exactly how to program a fast response to news? They will predict market response and not get whipsawed? Seems pretty tough to me. Is this more 'efficient'?... Efficient Market theory is a very flawed premise for many reasons but first because it assumes markets can actually reach equilibrium. There is no such thing as market equilibrium- there is never a time when all bids and offers are satisfied and all participants agree on a stable price. Look at price charts, they constantly move around. For a market to be in equilibrium then everyone's percieved value is the same, that's never going to happen. Karl Marx also thought that he could scientifically quantify value. I think high frequency trading is likely to add more volatility as they try and outdo each other and that's about it, the people who get picked off are the ones who can't adapt.
If high frequency trading is the new big thing then why are all the volumes down ? It appears that they process the information and pump orders into the market at high-frequency but don't actually complete anywhere near the amount of trades that they submit into the market. It should be called LOW frequency TRADING.
1. HFT is synonymous to trading the order flow. 2. Order flow can be predicted and it is done by a few. One way we do it is by reverse engineering the algo. order offered by the bigger institutions. 3. Race between AIs. If it's an algo., there is always a tendency. Let's say you find out that Goldman Sachs (... this is only an example...) uses a specific fitness measure and Genetic Algo. to self-adjust their VWAP models' parameter. You can use AIS, SHIAI, NN or other agents to monitor predict a specific change in order flow derived from the information above. From there, all you have to do is front-run the algos. enough so that it doesn't trigger their re-adjustment, and sometime help their model's liquidity, considering this be HFT. 4. Inefficiency comes from 2+ institutions running a similar model. These institutions compete with each other to out-perform the competition. Though, there is only 1 market and the base "formula" for a strategy like VWAP or others are the same. .... Basically, HFT and how orders are routed and handled by the institutional brokers are 2 sides of a coin. Because they IBs use some kind of algo. or logic to handle these orders, they're vulnerable to patterns. Dark Pool was supposedly the ultimate answer to the issue but they're being "edged off" by funds and traders predicting how the orders are routed and handled. Anyways... my point is... Successful HFT operations don't handle and data mine market data like your everyday statistical backtesting. There are plenty of "edges" around to be exposed. The question is whether you have the access to utilize them and the background knowledge to focus your attention to.