Two Plus Two Older Archives  

Go Back   Two Plus Two Older Archives > Other Topics > The Stock Market

Thread Tools Display Modes
Old 01-09-2002, 09:43 PM
Posts: n/a
Default an complex but interesting problem (LONG!!!)

Buyers aren't like oil deposits. You can't send out a team of geologists to find out whether they're there or not. The best you can hope to do is sit on your trading floor and wait for a sign.

Moreover, the inventory risk for a trader is essentially the inverse of the temporal overlap when offsetting orders become visible. If a market order to sell pops out of a computer - coming from Heaven knows where - and a market order to buy pops out of the same computer 10 seconds later, your risk is 10 seconds - if you know the buy order is coming.

Or it could be forever!

If you don't talk to your clients on the phone and know who they are and what they want, or you don't have a paper buy-limit order in hand - or even worse, if you are seeing the same view of your buyer-client as everybody else sees without everybody else knowing it (because computers enable the order-shopper to be everywhere at once one moment and then cancel the next), and you all end up trying to sell to the same person at the same time - you might be SOL.

So suppose, to bring some friction and some visibility back into the market - so that you can reflect true ambient demand and show your clients a narrow bid-ask, that reflects a seller in the afternoon to a buyer in the morning for instance - you decide to pay them to submit to friction, and not cross-shop their orders. In other words, you say what if I agreed to pay you not to wait and see where I am at our market center, and not to try to lean on us and free-ride on our information and try to pick us off, what if instead we paid you to give us market-makers some idea of your buying and selling plans in advance, so that we could lean on you and price you in and even pick you off? (Then, we market-makers could settle it amongst ourselves how to have just one inventory per client and not trade redundantly.)

In other words, you say what if there were a payment mechanism flowing to the people being looked at and at the expense of the ones doing the looking? What if there were an incentive to get people to migrate from the pickoff side back to the get-leaned-on side, so as to bring some new information into the market? Because in a marketplace where it is a battle to have the least friction - and in the end nobody has to submit to any friction - you end up with all pick-off artists and nobody left to lean on - meaning no real idea where anybody actually wants to buy and sell.

Because everybody who does actually want to buy or sell is nevertheless either lying silently in wait one moment, or picking off someone in twice the volume they actually need the next! And when that last person to look at is gone, all that will be left is a self-perpetuating nonsense-cloud spinning around itself like a bunch of flies! It drifts off, evaporates, and gets replaced Darwinistically.

No matter how badly you need to sell, it will almost always make sense to hold back your pickoff an infinitesimal second longer. But in the end, you and your counterparties all die and get replaced - by people participating in a system which either capitalizes on pre-existing friction or induces friction incentives!

In other words, for a zero-friction player who perceives he can bang the immediate bid at any moment, it's like an option which it would be irrational to exercise before expiration. So even if prices are at 30 and such a seller would be willing to sell as low as 20 and "needs" to sell immediately, he'll effectively sell to himself and go long as a middleman, and his apparent willingness to buy and sell will be a function of where he thinks he can make a profit relative to the current spread. His order reflects to the market where he thinks he can get done against everybody, rather than where everybody can get done against him.

In other words, every participant's order becomes a house of mirrors, reflecting the market's apparent needs back to itself!

So, rather than actually locating counter-parties, the focus shifts to hiring the most central dealer, with the most reliable proprietary decks of orders to lean on and

bang against, and the lowest friction, to dynamically replicate as the market rises and falls. The illusion is that lower RELATIVE friction actually creates more counter-parties. BUT IN REALITY, LOWER RELATIVE FRICTION MERELY SHORT-CIRCUITS THE DISSEMINATION OF ACTUAL NEEDS.

So the sleight-of-hand is the relocation of decision-making and inventory-management capacity up hill to compensate for lower actual liquidity, which also creates the illusion of higher immediate liquidity as selling interest is kept off the market and out of the visible region where it could be priced in (like S&P futures

sellers in 1987). And again, why does shifting order and risk management to lower-friction operators conceal true supply-and-demand conditions? Because the time in advance an order will be "exposed to the crowd," and the distance from current prices, will be proprortional to the time it takes to place and cancel an order.

So this brings us back to the problem of how do we turn it outside-in again, how do we reverse the flow of information from the nether reaches - from deep out in investor-land - back into the market center? How do we, in essence, construct a payment mechanism, flowing from one investor, into the market center, and evenutally back out to another investor (or even to a supplier of stock such as a venture capitalist) in the form of firm bids, non-volatile portfolios, and fairly-priced securities that reflect actual demand, rather than mutual investor speculation?

You might say to investors, okay, we'll pay you to warn us in advance of your market orders. And we'll rate you on how reliably you follow through on your warnings. Meaning, not only will we pay you based on your "rating," but we'll believe you based on your rating, so if you never tell the truth we won't really get mad at you or even notice you're talking (by pinging us with indications) any longer!

So, we give the investor a number, say .08%, and say for every order you warn us of 1 minute to 30 minutes in advance, we'll put .08% times the dollar value of your order in an account for you. Or, if you say "5 minutes," the amount we pay you will rise to peak at .08% in five minutes, drop off to 0 in ten minutes, and we'll actually debit you .08% - times your historical followthrough percentage - if you still haven't followed up with an order in 15 minutes! Of course, the sooner you withdraw your paid indication of interest, the less we'll debit you for misleading us (again, assuming we even believe you at all any longer).

Or, as a market-maker or market center, we can put together prints based on certain times of day - like "on close" - or to take place based on specific news events - such as if the Fed eases 1/2%. Then we say, okay, here is the price shaping up for this print. If you give us your indications of interest to participate in this print at such-and-such price or better, and depending how far in advance, we'll pay you .47% of the cash value of your order for telling us!

And meanwhile, we'll be advertising your time-semispecific interest on your behalf - and at all points along the time axis at once - so as to pay other people to indicate back to us and, transversely to you, that they want to take the other side! The maximum amount you get paid rises depending on how far in advance you warn of activation of an actual live, matchable order. But the flip side is that the longer you take to retract your indication, the greater the penalty (assuming you usually follow through of course).

Notice, the difference using advance non-price-specific indications, rather than immediate/no-advance-warning market, or price-bound limit orders, is it allows a market center to collect supply-and-demand information from sellers who are holding back because they think the price will rise, and from buyers who think it could go down. Selling an indication, they can advertise, and still stand to benefit from the expected move, rather than getting leaned on or picked off.

The offsets they expect can accelerate to meet them exactly, rather than both hitting a vacuum (that not even market-makers will touch) in an unecessary, zero-visibility, three-second whipsaw.

Longer-term indications, such as for "today" - or for "this month" - where such indications shift into narrower and narrower timeframe columns, (such as from today to "impending" - and earn a higher payscale as activation time nears) could be accepted and compensated. The payment would come from the fact that a market-maker, having purchased an indication, would be able to advertise a firm bid leaning on it, and win prints at a price he can count on as being favorable to all parties involved. The discovery of all-around-win/positive-sum swap opportunities would be accelerated, at the same time as the risk of inventories, capacity decisions, and general economic planning would be removed.

So, anyway, for the grand prize,

1) What are some of the pitfalls of pricing, compensating, and penalizing such indications of need - "gestures" - from end users? Consider utility to recipients (acting as information intermediaries), cancellation costs, and possibly fuzzy indications which have a fluctuating payscale based on confidence. Will payment give the intended incentive to provide useful information, or just result in a lot of people gaming the payment structure wihtout doing anything worthwhile for others?

2) What are some of the various types of such indications, or "time-contingent" orders that you might accept at a market center? What is the time-dependent value of information, what are you trying to encourage? Would you use narrow windows and log curves, or fuzzy payment, or what?

3) What are some of the problems you might run into with the collective purchase of unique bits of information by multiple market-makers trying to take advantage of them at once? How could market-makers, observing the same bits of data, coordinate between one another, at the same time as competing AND carpooling costs?

4) How does the payment flow-through work? What are the different actions and decisions created at each point in the supply chain, and how do the other points benefit? Does it necessarily have to be cash payment, or could advantages and disadvantages in the auction - such as asymmetric friction, awkward minimimum front-running increments, or price-breaks - be allocated on the basis of behavior, so as to incentivize individual contributions to the collective good of market stability? In other words, once you have nailed down the actions and benefits, are there any opportunities for straight-through or payment-in-kind exchanges, without the necessity of an intermediate cash currency?

5) How do you contain the data so that outsiders can't free-ride on the data purchased at considerable costs, by leaning on it with their own bids away from the market center? Do market-makers have to charge back to investors and end-users just to get a momentary glimse, a window into the market-center spread? Do they have to either pay or follow through with a transaction - are indications only disseminated in turn for indications traveling the other way? Is discretion only valuable when combined with information in a proprietary setting? What is the different utility of information at different points in the chain, investor, exchange floor, etc.

6) Notice how sellers of indications don't take on any immediate risk like sellers of options. They are, in fact, selling information, if for no other reason than you cannot lean on them because their price will slide with the market. Can you elucidate the difference, between options, futures, and "gestures," based on what happens at different points along the time axis, the states they transition between - indication, activation, execution, cancellation... - and the value of these events and the time between them to counter-parties?

7) Notice how, if the market moves away from a "gesture" with a price limit, it would be irrational for its source-gesturer to ever cancel. Better for him to hope he gets away with it, so we'll never know if he was misleading us or not! One way we can overcome this problem is by only paying for Gestures which get followed through with "activated" orders which actually get filled - rather than those that merely activate, sit a moment to qualify as having been exposed, and then cancel. In other words, it is the actually finding an offset, rather than the mere auto-transition to active status, that wins the credit. Plus, that way, we only end up paying and penalizing gestures that turned out to be useful - by having any relevance to emergent market events - anyway. Fed-1/2% "event" orders just evaporate unpaid when the Fed eases only 1/4. Can you think of any other safe way to compensate gestures with price limits away from the market, and even as the market moves towards and away from them over their pre-active/pre-canceled indication-life?

[img]/images/glasses.gif[/img] Can members sell their rating? Can they rent it? Can they pay to bring it up to speed? Think about the difference between what is known as a principle versus an agent transaction. Is it possible to be an agent in the underlying good traded, but to be an information principle? How could brokers compensate their clients for submitting to friction which they can then sell as pooled, unique information to a market center - by bartering what types of services? Think about the separate information and goods supply chains. Further, think about the difference between price-implicit information, and explicit informaton in the form of gestures having been separated and sold separately from actual orders.

9) How specific does the whole payment mechanism have to be, how precise, how fair? Or is there enough unrealized value, enough of a positive-sum game to play, that even the most awkwardly priced and executed transactions could offer a mutual, if unbalanced, benefit? What is the cost and benefit today of holding information back, who bears and enjoys it, and in what quantities?

10) In a hundred years or so, could such a time-axis information/entropy supply chain actually weave its tiniest tendrils to the futhets reaches of the universe, and cause the whole thing to burn out - pop! - like a filament between the ends of time?

Anyway, I forgot most of it - and I guess that's enough for today anyway(!) - but these are the trillion-dollar questions!!


Reply With Quote

Thread Tools
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off

Forum Jump

All times are GMT -4. The time now is 07:30 AM.

Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2021, vBulletin Solutions Inc.