Australian (ASX) Stock Market Forum

Alan Kohler's Take on High Frequency Trading

Why the Market Profile legacy analysis of price and time has changed
How electronic market order flow contrasts to open-outcry order flow
Analyzing volume and price
How the influence of supply (volume) effecting price is the new paradigm
... right on cue, just as VSA gained traction. What's the next fad the algos can exploit?
 
... working systemic instability into the market microstructure.

I would have thought that any momentum following trading introduces instabilities. The price goes up, everyone gets on board, and the price goes up some more until it is far from it's rational/fundamental value. This happens with trading of all asset classes (houses, food, financial derivatives etc), and probably reflects something fundamental in human nature. It doesn't seem to be qualitatively different if I make a machine do the trading, and make decisions at a slightly faster pace than my poor little brain can manage.
 
I would have thought that any momentum following trading introduces instabilities. The price goes up, everyone gets on board, and the price goes up some more until it is far from it's rational/fundamental value. This happens with trading of all asset classes (houses, food, financial derivatives etc), and probably reflects something fundamental in human nature. It doesn't seem to be qualitatively different if I make a machine do the trading, and make decisions at a slightly faster pace than my poor little brain can manage.

Punta, when you simplify it like that the issues are lost. It's not simply a matter of 'making a machine to do the trading', what if the machine fails? What if the machine is only tested for a few hours before being dumped into the live market, like the NYMEX crude bot that f'd stuff up? What if the firm doing the HFT accounts for more than 50% of all the market? What if the machine fails, and then the failsafes fail? This has already happened in repeated instances across many exchanges. Turns out, the 'laws of computer science' apply to HFT code as well, who would have thought! :rolleyes:

Research indicates repeatedly that many HFT strategies are not even viable in lieu of transaction costs despite their high level of predictive capability. HFT by and large gets exchange rebates to perform their churn, without these rebates there would be no viability. A concept so retarded that the London Stock Exchange has already abolished it after being one of the first to utilise it. Of course a whole bunch of HFTs jumped ships to BATS Europe (which holds <10% of share compared to LSEs 65%) straight away, since they wouldn't be able to churn viably without.
 
Punta, when you simplify it like that the issues are lost.

Yeah fair enough, there's obviously problems with how individual algos are implemented, but I can't see anything fundamentally wrong with HFT.

There is currently legislation in place to prevent asset bubbles, or large price fluctuations, on time scales that the human brain is familiar with (i.e. months/years). I guess with the advent of HFT there could be legislation in place to stop HFTs causing big price fluctuations at "their timescale", of a few mins.
 
Yeah fair enough, there's obviously problems with how individual algos are implemented, but I can't see anything fundamentally wrong with HFT.

It' all about market structure and its integrity, rather than the individual algos.

Right now the only barrier to entry for 'driving on the HFT highway' is money. Got money? You can drive on our highway, even if you don't know how.

But it turns out the highway is pretty important to the real world, so letting on a jalopy with a rocket strapped to just to score the toll seems more than a little stupid.
 
How does any of this affect Joe and Martha Sixpack with their occasional share purchase to put in the bottom drawer?

Much ado over nothing methinks. :2twocents
 
...What if the firm doing the HFT accounts for more than 50% of all the market?

Wouldn't this mean that it would have to be trading with itself?? That doesn't seem to make too much sense.

You could say that, if an algo is not trading with itself, then there must be another party willing to take the other side of the trade, and so the algo isn't pushing prices around unfairly.
 
Wouldn't this mean that it would have to be trading with itself?? That doesn't seem to make too much sense.

You could say that, if an algo is not trading with itself, then there must be another party willing to take the other side of the trade, and so the algo isn't pushing prices around unfairly.

Lots of algo activity is not matched trades but rather huge quote stuffing. Also there is the well known 'pass the parcel' technique, algo A has a block of 100 shares and sells it to algo B for one tick more than it paid, then algo B sells the block back to algo A for one tick higher again, etc etc. No reason the algos have to be owned by different firms.
 
Also there is the well known 'pass the parcel' technique, algo A has a block of 100 shares and sells it to algo B for one tick more than it paid, then algo B sells the block back to algo A for one tick higher again, etc etc.

That sounds like flat out market manipulation. It makes sense though - if there is a fundamental instability due to the fact that people/algos follow momentum, it might be in your interest to generate a bit of momentum, so that you can ride the wave you've created further down the track. I'm not saying it's right, but if I could make waves and surf them, I'd be pretty happy...
 
The thing is that Mr & Misses average think these things are wildly profitable at the cost of their own "investment" loses. They aren't, mostly the BOTs are hardly profitable.

I've seen the evolution of HFT bots on ASX since 2007-8 up to now and 99.9 percent of the time they have greatly reduced volatility. To the point that many a local has had to move on because they cannot get intraday moves big enough to cover cost. Yet idiot journalist state that they wildly swing the market at will to profit from everyone else. Or they somehow front run other execution bots? Ridiculous!

From time to time they have a bad hair day and muck things up but I've seen many human traders do the same. No problems in regulating them to "do no harm" but firstly it would be nice if Kohler and his Muppets knew what the hell they spoke of.
 
Lots of algo activity is not matched trades but rather huge quote stuffing. Also there is the well known 'pass the parcel' technique, algo A has a block of 100 shares and sells it to algo B for one tick more than it paid, then algo B sells the block back to algo A for one tick higher again, etc etc. No reason the algos have to be owned by different firms.

Then Punter comes along and sees a higher price so puts in a bid for 10000s at market and Bot sells into it with an average price way less than bought.
OR

If they get really lucky there is an announcement related or un related which has punters lined up at open for buy OR sell.
Bot picks the eyes out of it quicker than traders can react!

Bot buys when everyone is selling and sells when everyone is buying.
You just cant see clear cut evidence.


I've seen the evolution of HFT bots on ASX since 2007-8 up to now and 99.9 percent of the time they have greatly reduced volatility.

Absolutely.
My feeling is that long term trends will be rare.
Once bots get on them they will control volatility.
 
Absolutely.
My feeling is that long term trends will be rare.
Once bots get on them they will control volatility.

I agree, they provide liquidity both ways so the effect is minimal.
However they do provide a drain on markets that was not there once, so there is less money for the humans without HFTs machines close to the action. So less overall opportunity in the market, longer term that is meaningful.

On the other side of the argument, it does not seem to make much sense that volatility tends to pick up during holiday periods, where, presumerably, the machines are not lying on the beaches or hitting the slopes. So if the machines are moderaters of volatility then shouldn't the holidays be a little more boring?
Or does everyone go home and day trade on their holidays?
Maybe everyone has a HFT app on their Iphone - younger generation just can't keep up with em.
 
Then Punter comes along and sees a higher price so puts in a bid for 10000s at market and Bot sells into it with an average price way less than bought.

Except, in the 1 human second it took to bid, liquidity has already dropped and the price has moved.

http://www.economist.com/node/21547988
A bigger problem, says Paul Squires, the head of trading for AXA Investment Managers, is that increased liquidity can be illusory. “You can press the button to buy Vodafone, say, and have it executed in a second but in that period 75% of the liquidity has disappeared and the price has moved.”


You just cant see clear cut evidence.

Errr, says you? Actually analysing market microstructure is pretty much the bread and butter for outfits like NANEX. Worth checking out before making claims like that.

My feeling is that long term trends will be rare.
Once bots get on them they will control volatility.

So how has AAPL, or NG futs, or Soybeans, or Swiss Francs, or JPY, or anything that is up to its eyeballs in HFT for the last 2 years been trending so long term? I am wondering how you explain all the volatility and 100% correlation of unrelated markets, of May 6 2010 then? Oh wait, turns out HFT are just as 'fearful and greedy' as everyone else and at the first whiff of trouble they will just turn off and let the market fall/rise to the human price level.
 
No they have not. The vast majority of market participants cannot afford to buy and collocate servers at the ASX.

Well no ****, if you allocate an investor with $1000 the same 'participation' rate as Goldman Sachs Sydney office then of course it seems skewed.

But your statement is not true, the vast majority of market participants ranked by actual participation in the market can in fact afford to buy and colocate servers at the ASX (or similarly connected datacenters).
 
For every dollar the HFTers make, someone has to lose it.

1/ Quite so in derivative markets, Bu Joe Sixpack isn't likely to buy his nest egg there. Not quite so clear in equity markets. Suppose Joe bought CBA for $5ish at, or shortly after the IPO... who lost?

2/ When Joe Sixpack goes into the market to buy some shares, HFTs may have pushed it down or pushed it up. JS is just as likely to benefit by a few ticks.

3/ Ergo, HFTs do not compete in any way with buy and holders, they compete with short term traders. As a private short term trader you always have been a guppy amongst sharks. I couldn't give a fat rat's @ss about the HFTs... bring it on, I'll trade against them (and do).

I want to regulate everything

Corrected for accuracy. LOL

I am incredulous that you claim to be an Austrian. :rolleyes:
 
Top