"When buy and sell orders come into an exchange, they are first flashed to those paying to see them for 30 milliseconds — 0.03 seconds — before they are available to everyone else."
One thing I don't understand about "flash trading", and what I've never seen an explanation of in the media: How does this interact with price-time priority?
In a regular old auto-matching system, if the spread is two ticks wide and a buyer sends an order to lift best offer (instead of just improving best bid):
1. If you had a sell order at best offer in the front of the queue the time the buy came in, you're matched, and it doesn't matter if you try to cancel your sell a nanosecond later.
2. If you had no sell order, it doesn't matter if you improve the sell price a nanosecond later --- you're not matched.
So does "flash trading" change either of these? Or is it just a fancy name for "getting market data real fast" --- similar to what you'd get by colocation, but artificially enforced (e.g. they delay everyone else's data by 30ms so the "flashers" can see it first)?
Thanks. So the exchange really are messing with price-time priority in exchange for cash? Wow ...
But then why would any non-flasher ever place a single order on these exchanges? Couldn't a boycott could solve this problem without any legislation --- they could just go to an alternative venue which wasn't selling them out in exchange for more fees?
---
Appendix: Bear with me while I think out loud so anyone can point out mistakes I'm making.
Time n-1, Order #0: Some seller posts an offer of size N > 5000 at $21.00. (It has to be $21.00, otherwise Order #2 would have nothing to buy at that price).
Time n, Order #1: Mutual fund sends buy 5000 "at market" (i.e. fill me at any price from here to the moon).
Time n+1, Order #2: Would-be flasher sends buy N at $21.00.
Time n+2, Order #3: Would-be flasher sends sell 5000 at $21.01 (and prays that no other flasher beats him to it).
Time n+3: Repeat from time n-1.
Under normal price-time priority, #1 matches #0, #2 matches #0 and becomes best bid, #3 doesn't match anything, and your new market is $21.00 / .01. The total traded volume is N. (The would-be flasher is now long N - 5000 and has to trade out at a loss of fees + stamp duty, so the first time around the total traded volume is 2N - 1, but that doesn't happen again: he either stops doing this shit or goes out of business).
With "flash trading", #2 matches #0 (despite #1 having time priority) and #3 matches #1 (despite not even being in the market at the time #1 arrived). The total traded volume is N + 5000. The exchange earns some extra fees, the flasher gets the arbitrage profits, the mutual fund gets screwed, and the sellers (#0) don't care either way because they were willing to trade at $21.00 all along. Weird.
Yeah, we're getting ever closer to the era in which:
- you step out to home depot to buy a $20 hammer
- when you get there "The Flash" (of dc comics fame) figures out you're looking to buy a hammer and buys all the $20 hammers
- he leaves a post-it note saying "if you want a hammer meet me in aisle 7"
- you get there and he offers to sell you a $20 hammer for $21
- either you buy it or you don't; once you've made your decision he returns either (N-1) or (N) hammers back to the store (30 day return policy and all that)
- if you head over to lowe's instead well he is "The Flash"
Not an exact metaphor for "flash trading" (as sketched it's riskless) but it's imho a relevant thought experiment.
At the individual level the winners and losers are pretty obvious. At the systemic level it's unclear any actual efficiency is gained; differentiating between actual "efficiency gains" and "red queen" scenarios is tricky.
Because the truth is that this is business and all the people who aren't in on the flash would be if they could afford to be/had the expertise in-house.
> The programs can gauge market response and, using pre-programmed manoeuvres, called algos (for algorithms), catch the demand before the public (that’s YOU), grab lots of it cheap, enter and cancel hundreds of small orders in order to find out the public’s limit for a stock’s price and dump all it’s holdings to them at a penny below their limit, but above what the computer paid for it, for a small ($10,000, say) profit, in a matter of seconds.
This isn't flash trading, this is just high frequency trading. All the stuff mentioned in this paragraph about "cheating" has nothing at all to do with flash orders. bellringer seems to suggest that high frequency trading is "cheating". This argument that guys with faster computers and better connections are beating at home traders is nonsensical. If I tried making cars at home in my garage, of course Toyota is going to beat me. They have state of the art factories, trained engineers, and years of experience. Toyota isn't cheating. Similarly, Goldman Sachs, along with lots of smaller guys, has huge computing clusters, blazingly fast and connections and they're going to beat you, a guy with a laptop at home. But it's not cheating.
Cheap cars are useful. A super liquid stockmarket is a zero sum game in which everyone is encouraged to participate but a few sharks basically have extra cards.
Its really not obvious why such a responsive market is necessary, it is simply said to be a good thing and a progression, yet it moves at speeds that clearly do not reflect the underlying fundamentals.
Because the stock market doesn't generate money. Money changes hands based on rising and falling values, but the money hasn't just appeared out of thin air, nor is it destroyed.
a "super liquid" stock market as you put it means lower spreads... that benefits you. The people who are losing money because of high frequency trading are algo traders... that is not you (unless you're a hedge fund)
Your reasoning is wrong. Limiting the frequency of trades (by some mechanism) does not imply more money going to brokers or higher spreads. (It may effect clearing times, certainty over being able to buy quantities in short times & exact prices).
Then the second part is wrong also, as the low frequency traders will lose money, getting on average worse deals. The high frequency traders will make sure you get a worse price, because any price spike you may have fortuitously gotten will be removed. Alternatively you can just reason that the high freq traders are costing money (using resources, making profits) and ergo must be extracting money out of the market.
You'll have to make a much better case for the value for such liquidity.
>because any price spike you may have fortuitously gotten will be removed
this is true, but "fortuitously" doesn't always apply. "price spikes" could be high or low, and could affect you positively or negatively. with out high frequency traders, it's equally likely that you get a fortuitous or unfortuitous spike. what high frequency traders do is even out prices so that they are always "right". they ensure that when you trade, you aren't trading on some random spike and are instead trading at the correct market price. despite what you think, real low to medium frequency traders enjoy having correct market prices at all times, which is what high frequency guys ensure.
Sorry, this is also wrong. Why is it equally likely? A high spike will be when they buy and a low when they sell. As a low freq trader you'll get worse prices buying/selling on average as a result.
Your notion of the existence of a precise right/correct price for anything is an illusion, forget about it,
This is why it is best to start with big picture reasoning - in this case that they make money and ergo others lose/make less in a zero sum market with only tangential impact on real economic activity.
"Then we divide all the incoming orders into discrete time chunks called quantums. All the orders in one quantum will be processed independently of the rest of the world and before the next quantum is processed."
Basically, instead of having continuous trading (which could mean huge amounts of market data every second), they just do 30-second auctions all day long. You get an order book update, send in your orders, and they figure out who matched whom 30 seconds later. At which point they send out the next order book update, consisting of all the orders which didn't match in the last auction. Time priority still applies within the 30-second chunks. So if you have a trading algorithm, it only has to wake up once every 30 seconds ... but when it does, it has to be fast.
(However, time priority doesn't apply in the opening auction --- the priority for orders at any given price is randomised and has nothing to do with who got their orders in first.)
Of course the little local brokers loved this mechanism, and the big foreign banks all hated it (and probably lobbied against it behind the scenes). Maybe they lobbied so hard that they don't have this mechanism anymore; I've been out of the industry for a few years now so I'm way behind the times ...
Well no because this is how exchanges have always worked. You can look anywhere and see this, TV, the web, stock prices are delayed by 20 minutes. Market participants get them without this delay and no-one's ever had a problem with this. So the principle of paying a fee in order to get faster access to pricing is well-established. This is just taking it to the next level.
Secondly, anyone who wants to can do this, they just need to pay the fee. There is zero evidence of a cartel.
You don't need evidence of a cartel, monopoly, or cheating to decide that society/economy may be better off if the rules were changed. I have no evidence that things would be better if the rules were changed but some people that know better than me see value in taking a closer look to see if this might be the case.
Simply force purchasers to hold stock for at least 6 months, or some other long period, so that fundamentals will have to be carefully looked at. Buying a stock signals confidence in a companies stock price for the duration you expect to hold the stock. If you make people hold it longer, the signal becomes more meaningful. Finally, market manipulation becomes much less rewarding, since you can only flip your money twice a year (profit from manipulation is % gained each time * capital * number of manipulations), and you can't foment at all really.
Perhaps it would also be helpful to put a 5 (or more) day delay on any sell orders. I would personally be in favor of a 2 year delay or more. Currently, if you get really bad news (like a companies only drug was not approved, making the company worthless) you can sell the stock immediately via a market sell order, and get whatever the highest offer is. This is no different than collecting pre-signed contracts to buy your house, then signing the best one when you see smoke coming out of the windows. The person offering to buy the stock has obviously not had time to adjust their order based on the news, and so you are selling them something they weren't bidding on. If you give buyers time to adjust, selling on a panic would be completely retarded, since you will either lose all value (selling into a market with buyers who have adjusted their offers way down), or the panic will subside and you will get a fine price, but then could have just held it as you were planning before the panic anyway. Discouraging panic selling makes prices more stable (does anyone honestly think that the best estimate of the value of a large company could possibly vary 50% in a few days??) Also, automated trading triggers would be eliminated, which would be great. If a firm can setup computers to sell a stock if it drops 5% in 1 minute, then they are taking much less risk than someone who has to hear news, find a computer, put in login information, etc. However, the firm's buy order signals the same confidence as the careful stock researcher, while in reality they can buy stock with far less confidence since they are taking less risk.
Both of these measures would reduce market volatility, if not effectively eliminate it over short periods. This would mean that corporations could borrow money on much better terms, since their stock prices would be more stable, meaning that stock is better collateral. It would also mean shareholders would insist on things like dividends and long term profitability and planning, since they are going to be holding the bags if things aren't good in the long term.
I'm against flash trading but 6 months (or even 6 days) is throwing the baby out withthe bathwater. Look, I'm broke - lend me $10/ thanks. OK, I paid my rent and got paid, here's $11, thanks a lot. That's what liquidity is, and it's important - with your suggestion everything would become glacially slow and tiny credit hiccups would become hugely amplified. You should be able to buy when you see an opportunity and sell when you need cash, even if your need of cash is to finance another buy.
The problem with flash trading is that its practitioners are in effect paying the exchanges for the right to join in on a trade after it's been proposed and they can see whether it has profit potential or not. Sure, they are making tiny margins of a fraction of 1%, so they say it's not greedy. Good point. But you can only afford to do that if you've got fat money to participate and to make up for those tiny margins they're putting up big money on every deal that does show profit potential.
I find it hard to believe it's been legal up to now. Every time I study it, I'm reminded of some old movie where a gangster has paid out money for a private telegraph line to get the results of the horse race while the betting is still open at a remote location.
I think the long-term quantities you propose like 6 months and 2 years are absurd, but I agree with your basic premise that too much speed trading undermines the market's price discovery function and evaluates arbitrage over considerations of fundamental value. It would be interesting to model this behavior with an agent-based simulation of a virtual market.
The twist here is that in the movies it was all a con, but here in Real Life it is actually working, that is certain people are getting information faster than others for what appears to be risk-free profit.
The last part(risk-free) is what I have a problem with, if you are providing liquidity, then you should be making profit while taking certain risk. The flash order trading seems like a licence to print profits without the risk.
Well, there's multiple flash traders. If you're trying to buy at $21 and sell at $21.01 you have two trades you need to make and you run the same risk as a regular old non-flashing market maker: that you get one leg of the opportunity and your competitor gets the other, then both of you are stuck with a position that you trade out of at a loss.
True, each flash trader is competing for profit, but it seems that the typical mutual fund holder, who invests on time scales longer than 24 hours, is uniformly screwed.
Considering that the typical mutual fund holder represents the vast majority of the public, and that flash trading is only profitable for a handful of people, I would guess it is time for a change.
Ok, I'm not sure why my post was modded down, it obviously shouldn't have been.
The market _should not_ be liquid. Liquidity is great, but you haven't provided any reason it's better that the market be liquid. Your first example (loaning you money) is credit markets, not stock. And credit markets become looser if stock markets become stricter, since stock is better collateral for loans, so here your example supports my scheme.
Credit hiccups would not go away in my scheme, but day to day sensitivity to credit markets would be reduced if executives were incentivized to look after long term stability of companies instead of quarterly profits. Do you believe Ford or Standard Oil, in their heyday, would have been crippled by being unable to get loans from WallStreet? If a company fails for inability to get a loan, they have been painting themselves into the corner with bad decisions for years.
Finally, why shouldn't the stock market be glacially slow? Stock prices could change just as quickly as they do now, since buy orders could be entered and removed instantly, just as now. The only thing this would prevent is people flipping stock as short term investments, and so stock _would not_ fluctuate as much, since there would be so many less trades overall. Short term investments are always (no exceptions) an attempt to game the system or trade on inside information, and there is certainly no way to profit in the short term except by fomenting, using inside information or trading far quicker than the average investor so you can respond to news more quickly. All of these activities hurt the efficiency of the market, so the market becomes more stable and prices more accurately signal company health than they do currently. When you have a stable market, price ratio's can increase since risk is reduced, and that is good for everyone when it is sustained.
IT's true that my first example is a case of credit rather than stock, but I kind of wrote it as a throwaway example of why you'd want to be liquid.
Rather than a loan, suppose that in the case of being broke I instead opted to sell something - a book or a guitar or a car, say. Now, would it be fair for someone to say 'no way, you can't sell that because you haven't owned it long enough?' I do think your heart is in the right place - managing for the long term and so forth - but your 6-month hold is extremely unfair to the individual investor, and would just result in the creation of unregulated off-market exchanges, like pawnshops for stock certificates. I mean suppose I buy IBM but next month they are hit by some financial scandal, I have to sit with my bad investment for another 5 months and watch the price crater as most of the existing shareholders sell?
Short term investments are always (no exceptions) an attempt to game the system or trade on inside information
I don't agree. I can think o a wide variety of reasons for short-term buying, and don't think regulators should be in the business of judging motives. I'm afraid I think this forceful imposition of long-term ownership criteria would only cause the market to become bloated and sclerotic and the cure would be just as bad as, if not worse than, the disease.
Finally, why shouldn't the stock market be glacially slow?
If Alice wants to buy and Bob wants to sell, and they've agreed on a price, what right do you have to demand they wait?
If I want to buy GOOG with a stop-loss under last week's low, you have no right (legal OR moral) to prevent me from doing that. You may call that an "undesirable short-term trade", but I'm not gaming the system, I'm limiting my losses.
If you get your way and trading becomes severely restricted, just watch how fast capital moves offshore to more user-friendly exchanges. That would not be good for the USA.
Wall St has too much political power for the government to stop this. If they can get the Glass-Steagal act repealed, they can squash an inconvenient policy initiative.
Maybe it's my programmer's mindset... I never really understood why flash trading is such a problem. If I buy a stock, and you buy a stock the next hour, is it a problem? What if I buy half an hour before you? A minute? A second? A us? Where do you draw the line? Why would you draw the line?
To me, it seems like a scrub argument. Note the usage of 'cheating' and 'fair' in this thread and article.
> Critics say flash orders favor sophisticated, fast-moving traders at the expense of slower market participants.
Why shouldn't someone who moves fast have an advantage?
One thing I don't understand about "flash trading", and what I've never seen an explanation of in the media: How does this interact with price-time priority?
In a regular old auto-matching system, if the spread is two ticks wide and a buyer sends an order to lift best offer (instead of just improving best bid):
1. If you had a sell order at best offer in the front of the queue the time the buy came in, you're matched, and it doesn't matter if you try to cancel your sell a nanosecond later.
2. If you had no sell order, it doesn't matter if you improve the sell price a nanosecond later --- you're not matched.
So does "flash trading" change either of these? Or is it just a fancy name for "getting market data real fast" --- similar to what you'd get by colocation, but artificially enforced (e.g. they delay everyone else's data by 30ms so the "flashers" can see it first)?
(edit - formatting and grammar)