What's the grey bar at the bottom?

bowlhead99

noClue said:

Thrugelmir said:

When trading in lines worth £ millions only need to shave a fraction of a basis point to make a profit. 

What do you mean by this?

A basis point is one hundredth of a percent

If a high frequency trader places a CFD or derivative trade to buy £5,000,000 of a stock at £10.001 per share and then sells it five seconds later for £10.002 per share, he would make £4,999.50 of profit even though the share price only went up by a little less than a basis point.

He will obviously incur some sort of cost from the broker or CFD issuer but if such costs are less than £5k for that five seconds of activity, there is plenty of money to be made for both the high frequency trader and the broker. Obviously if the price falls from £10.001 to £10.000 over the five seconds instead, the trader is going to be out of pocket because he has almost £5k of losses, plus costs.  But the trade may last only a few seconds, or even less than a second, so plenty of time for the trading robot's algorithm to get back into profit later the same minute.

The computers executing the transaction will be situated close to the exchange with faster connections to both a data feed and for placing the orders than you could possibly hope to have for your own investments. Most of the market volume is automated rather than individuals thinking about what to buy and sell.

John464

bowlhead99 said:

If a high frequency trader places a CFD or derivative trade to buy £5,000,000 of a stock at £10.001 per share and then sells it five seconds later for £10.002 per share, he would make £4,999.50 of profit 

Wouldn't he have to pay Stamp Duty?

noClue

bowlhead99 said:

noClue said:

Thrugelmir said:

When trading in lines worth £ millions only need to shave a fraction of a basis point to make a profit. 

What do you mean by this?

But the trade may last only a few seconds, or even less than a second, so plenty of time for the trading robot's algorithm to get back into profit later the same minute.

What do you mean by this? Do you mean that since robot can execute trader so quick that they can afford to wait for the later chance if there is one occuring?

The other question I have is that the observation we see in March. Let's say majority of them are automated, but the decisions are still made/algorithmed by human being? So do they predicted wrong, i.e. predicted rising price but end up losing, or they intended sell at a loss? I can't believe automated trading just trades for the sake of trading?

bowlhead99

John464 said:

bowlhead99 said:

If a high frequency trader places a CFD or derivative trade to buy £5,000,000 of a stock at £10.001 per share and then sells it five seconds later for £10.002 per share, he would make £4,999.50 of profit 

Wouldn't he have to pay Stamp Duty?

Not on a CFD or swap issued by a broker or other counterparty rather than buying the actual underlying share. And the broker who is actually buying the underlying instrument to hedge his CFD  exposure with the customer will have an exemption - see https://www.gov.uk/hmrc-internal-manuals/stamp-taxes-shares-manual/stsm118040

noClue said:

bowlhead99 said:
But the trade may last only a few seconds, or even less than a second, so plenty of time for the trading robot's algorithm to get back into profit later the same minute. 

What do you mean by this? Do you mean that since robot can execute trader so quick that they can afford to wait for the later chance if there is one occuring?

The trading robots investing millions on a high frequency trading strategy an make profits in split seconds over small share price variances that would not be profitable for you., as you have no practical way to exploit a share price movement of a fraction of a percentage over the course of a fraction of a minute, because you can't trade that fast and don't have enough money to do it efficiently even if you could trade that fast.

The part of my comment that you quoted was just me saying , of course it doesn't always make profits, but if it makes losses over that few seconds of trade it might easily make them back in the next few seconds. 

The other question I have is that the observation we see in March. Let's say majority of them are automated, but the decisions are still made/algorithmed by human being? So do they predicted wrong, i.e. predicted rising price but end up losing, or they intended sell at a loss? I can't believe automated trading just trades for the sake of trading?

You can see there was a large volume of trades in March - share prices of major stocks and the index as a whole were moving several percent in a day. More volatile markets lead to more transaction activity because people will embrace the idea of doing something about their existing positions or taking advantage of perceived opportunities much more readily, as it will impact their wealth or potential profits more than when markets are flat and doing nothing. 

A human writes the software and builds the algorithms although much of it is automated. The human is not literally sitting there saying OK let's buy x amount for thirteen seconds and then I think afterwards we should probably sell y amount of this other stock until the price moves to z amount. The machines will react quicker than it would take the human to think it and say it and press the buttons to do the trade. Transaction instructions are fed back and forth quickly following an automated or semiautomated process that is reactive to feedback and the actual data feed from the market.

Some traders (human or robot) will be selling and some will be buying. They are not trading 'for the sake of trading', but trading 'with the hope or expectation of making money from it'. Of course, they may be caught out if the they predicted a rising or falling price and actually got a falling or rising price. Large losses and large gains are to be expected, but each market participant believes that overall when they add up all their net gains and losses they will have done better by acting on information and transacting based on what they know or what they see, rather than sitting there doing nothing all year.

MaxiRobriguez

noClue said:

bowlhead99 said:

noClue said:

Thrugelmir said:

When trading in lines worth £ millions only need to shave a fraction of a basis point to make a profit. 

What do you mean by this?

But the trade may last only a few seconds, or even less than a second, so plenty of time for the trading robot's algorithm to get back into profit later the same minute.

The other question I have is that the observation we see in March. Let's say majority of them are automated, but the decisions are still made/algorithmed by human being? So do they predicted wrong, i.e. predicted rising price but end up losing, or they intended sell at a loss? I can't believe automated trading just trades for the sake of trading?

No predictions on the underlying shares profitability are made by the algorithsm (which was created by the human). They are designed to react to price movement over a short period time of time, and are based on known investor behaviour in reaction to certain scenarios and if possible known HFT behaviour too.

For example the dead cat bounce is a well known phenomena which can happen intraday as well as over a longer period, so you may have a certain HFT which watches for large drops, waits for some sort of flattening, purchases in bulk when occurs, and then unloads them all again a few minutes later if the price has edged up slightly. This sort of behaviour may be playing out on DS Smith's share price today, with algorithmic purchase and selling between 12:30 and 12:40 and 13:15 and 13:40. 

Such coding would imply no thoughts either way to the future of the underlying share. The only predictions being made is on participant behaviour in relation to certain events.

noClue

Thanks both @bowlhead99 @MaxiRobriguez. These all make sense now. Another question. I take it that these algorithm companies/individuals are private, i.e. not institutional but people like you and me? I would understand that the return is not guaranteed but since they trade so frequently the maths behind it is it will eventually go into positive? Also, how these companies gather money? You said huge amount (not exactly your wording)...

I remember a while ago I sometimes went to listen to risk management department's seminar one of their projects was betting on horse racing, and the maths model shows that if they bet frequent enough (with their own researched input) they will always return positive in the end. The certain client actually came back with money supporting more research in this area. What you guys described has its fundamental principle very similar to this, at least sound like this.

bowlhead99

noClue said:

Thanks both @bowlhead99 @MaxiRobriguez. These all make sense now. Another question. I take it that these algorithm companies/individuals are private, i.e. not institutional but people like you and me? I would understand that the return is not guaranteed but since they trade so frequently the maths behind it is it will eventually go into positive? Also, how these companies gather money? You said huge amount (not exactly your wording)...

The high frequency traders are the people moving high volumes of institutional money for short amounts of time to capitalise on an opportunity or grind out a profit over time .  There may be a few billionaires with the infrastructure to do it, but no they are not people like you and me with 'normal' amounts of resources.

They are people employing quant analysts and developers to build expensive software and infrastructure and processes and having a lot of money to 'stake' to make the small fractional returns on the large stake to be enough to pay for the costs of doing it and the risks of getting it wrong.

You will see some scam artists all over the internet willing to sell you access to a trading model or tip sheets created by a special clever algorithm. They are not credible because if they actually worked they would be billionaires and not selling software or seminars or tips on subscription. Any kind of day trading results in losses for the majority of people who try it (not that it stops people trying it, just like other forms of gambling, but they generally learn an expensive lesson). Whereas the algorithms created by the huge investment banks and hedge funds are the ones that are more likely to actually work. 

Just google "high frequency trading" and related terms to find out more. But don't expect to be doing it for yourself.

MaxiRobriguez

bowlhead99 said:

noClue said:

Thanks both @bowlhead99 @MaxiRobriguez. These all make sense now. Another question. I take it that these algorithm companies/individuals are private, i.e. not institutional but people like you and me? I would understand that the return is not guaranteed but since they trade so frequently the maths behind it is it will eventually go into positive? Also, how these companies gather money? You said huge amount (not exactly your wording)...

They are people employing quant analysts and developers to build expensive software and infrastructure and processes and having a lot of money to 'stake' to make the small fractional returns on the large stake to be enough to pay for the costs of doing it and the risks of getting it wrong.

The scale of this shouldn't be understated.

The I.T infrastructure to monitor thousands of stocks price movement every second, compare the price action against what will be hundreds if not thousands of algorithms", enact the trade, monitor specifically the subsequent price action on the trade made, compare against "post-purchase" algorithms, all the while also monitoring for other HFT opportunities, and ensuring the software has rules in place to ensure that all the combined trading, HFT and 'normal', isn't breaking any law, institutional risk or composition of ownership rules.

I don't work in the finance industry like Bowlhead so he's your man for the real nitty gritty, but I am a qualified and experienced programmer that has designed real-time applications (albeit not something I do anymore) - this absolutely isn't a retail investor game. Only the largest of the large have the ability to pay for the smart people designing this stuff, the ongoing infrastructure costs of hosting the software, the deep pockets to ride losses etc.

Stick to buy-and-hold long positions. 

bowlhead99

MaxiRobriguez said:

I don't work in the finance industry like Bowlhead so he's your man for the real nitty gritty, but I am a qualified and experienced programmer that has designed real-time applications (albeit not something I do anymore) - this absolutely isn't a retail investor game. Only the largest of the large have the ability to pay for the smart people designing this stuff, the ongoing infrastructure costs of hosting the software, the deep pockets to ride losses etc.

I don't really work in the bit of the finance industry that gets involved with brokerage or public markets trading, just have a general appreciation for how markets work. But you're right that essentially all the mindbogglingly fast, complex or opaque aspects of institutional stock market participation (high frequency trading, dark pools etc) mean that the trillions of dollars moving the markets are not coming from the pockets of retail investors who think that they have heard a good tip.

Even the most superficial understanding from reading the financial press or googling some of the terminology should make people think 'OK, there is more going on here than me getting a price quote and planning to sell later with some stop loss or limit order based on my clever reading of the charts'. The smart people are the ones that recognise the superficial understanding of market dynamics that they got from a book or blog or forum is not nearly enough to be competitive and you would be better simply assuming you don't understand it, rather than think that you read something on a forum or clickbait tipsheet which will now let you invest at an advantage to others.

When 'people like you and me' engage in trading we are at best going to be intermediate poker players, winning a little off the people who don't have a clue what they are doing while losing to the deep pocketed institutions who control the game, and paying out transactions costs (house rake). Mostly our real returns will come from long term gains (i.e. companies and stocks generally go up in the long term or give us a share of their profits through dividends), rather than intraday movements (values go all over the place)... even if it can be fun to take a contrarian position if the market seems to have overreacted and you don't mind being wrong for a long time before it comes right.

noClue

MaxiRobriguez said:

bowlhead99 said:

noClue said:

Thanks both @bowlhead99 @MaxiRobriguez. These all make sense now. Another question. I take it that these algorithm companies/individuals are private, i.e. not institutional but people like you and me? I would understand that the return is not guaranteed but since they trade so frequently the maths behind it is it will eventually go into positive? Also, how these companies gather money? You said huge amount (not exactly your wording)...

They are people employing quant analysts and developers to build expensive software and infrastructure and processes and having a lot of money to 'stake' to make the small fractional returns on the large stake to be enough to pay for the costs of doing it and the risks of getting it wrong.

Stick to buy-and-hold long positions. 

Yes, this will always be my strategy.

The whole conversation did an unexpected but very welcomed turn from my initial post ehehe. Since myself is from a maths/algorithm background, it's easy to grasp the general idea you two shared. It's just a bit unbelievable that big institutions spend that amount of money engaging in these zero sum games. If these resources get spent in more constructive activities, they might gain larger return in a long term. "Poverty limited their imagination."