electronic market making

################################ 
####     market making      #### 
################################ 
 
note: market maker is aka designated specialist (NYSE term), aka "dealer" (as opposed to investor/customer) 
 
### 
###  volcker rule (as a result of dodd franck) 
### 
 
effective 2015.07.21 (FRB,SEC,FDIC,CFTC are involved) 
 
prohibits banking entities from 
- engaging in proprietary trading 
- investing in "covered funds" (i.e. hedge funds, private equity funds, who can still prop) 
 
### 
###  proprietary trading 
### 
 
prop trading = engaging as "principal" in purchase/sale of financial instruments (except spot FX/commodities and loans) 
 
exempt types of trade = agency transactions, securities lending/borrowing, liquidity mgmt planning, repo. 
 
### 
###  permitted "principal" trading 
### 
 
principal = on your own(firm's) money 
 
(1) market making and related hedging 
(2) underwriting 
(3) risk mitigating hedging 
 
underwriting = undertaking the reponsibility of selling stock/bonds to investors on behalf of some company wanting to raise capital. 
 
### 
###  market making 
### 
 
MM is required to 
(1) routinely stand ready : must have two-sided quotes near/at the market price. 
(2) RENTD (reasonably expected near term demand) : subjected to position limit. 
(3) compensation: cannot incentivize prop trading 
 
the following factors are considered consistent with MM activity (not mandatory) 
-- two-sided quoting 
- for liquid equity   : very regular, continous quoting 
- for normal equity   : regular quoting 
- for illiquid equity : trading intermittently or at the request of customers ("trading by appointment") 
- for swaps           : at the request of a counterparty 
 
- pattern of trading (roughly) equal amounts on both sides (throughout mkt cycles) 
- trading (on exchange) should happen via passive limit orders 
- register as designated MM or supplimental liquidity provider on relevant exchanges 
 
passive VS aggressive : aggressive chases the market price, passive waits for the price to come to it. 
 
note: MM cannot hold one same position for too long, but this is not a problem with a rapid position rolling, high turnovers, MM can collect spreads along the way. 
 
note: MM can still have a target portfolio and trade to realize it. 
 
### 
###  MM related hedging 
### 
 
hedging-purpose trades can easily be interpreted as pure for-profit prop trading unless you have good explanation to justify why that trade was for risk hedging purpose, how it mitigates the risk exposure from MM, still consistent with the required position limit, etc. 
 
it is permitted to hedge with products other than those the desk market makes. 
 
### 
###  RENTD 
### 
 
there must be limits to; 
- the amount, types, risks of its market making inventory (product, position size, etc), exposure to relavant risk factors. 
- the period of time a financial instrument may be held. 
 
desk must 
- get pre-approval before exceeding those limits. 
 
limits take into account the following factors: 
- recent market trends (trading volumes, index rebalance events, etc and more fundamental stuff like economic indicators, industry forecast, etc) and customer trends (including customer portfolio status, etc) 
 
### 
###  volcker reported metrics ()assessment criteria   #  there are 7 of them, must be reported to SEC monthly 
### 
 
risk mgmt measurements 
- risk and position limits usage 
- risk factor sensitivities 
- var andd stressed var 
 
source of revenue measurements 
- comprehensive PnL attributions (existing day, new positions, residual, volatility) 
 
customer facing activity measurement 
- inventory turnover, aging 
- customer facing trade ratio (trade count based, value based) 
 
### 
###  other/misc 
### 
 
volcker rule requires rigorous complicance programs, CEO attestation, mgmt framework, policies & procedures, testing/monitoring/training, record-keeping, etc. 
 
### 
###  market making VS auction method 
### 
 
major exchanges use auction method aka order driven, matched bargain, where basically the exchange simply facilitates the direct matching of orders. 
 
https://en.wikipedia.org/wiki/Market_maker 
https://www.sec.gov/answers/trdexbd.htm 
 
Nasdaq and LSE use MM method. pros is better liquidity (from investors perspectives) but cons is your orders dont affect the quotes. (due to this cons, spread used to be huge, until the below OHR got enforced.) 
 
OHR: order handling rule (by SEC) 
 
(1) limit order display rule 
 
the limit order display rule states that market makers and specialists should display publicly the limit orders they receive from customers when they (the orders) are better than the market maker or specialist's quote. the rule ensures that the general public will be able to compete directly with market makers in the quote-setting process. it benefits investors since this publication of trading interest at prices that are better than the market makers' and specialists' quotes provide improved pricing opportunities for investors. 
 
(2) quote rule 
 
the quote rule requires specialists and market makers to provide quotation information. the quote information the specialist or MM publishes must be the best prices at which he is willing to trade (the lowest price the dealer will accept from a customer to sell the securities and the highest price the dealer will pay a customer to purchase the securities). 
 
a specialist or MM may still trade at better prices in certain private trading systems like ECNs, without publishing an improved quote. this is true only when the ECN itself publishes the improved prices and makes those prices available to the investing public. the quote rule entures that the public has access to the best prices at which specialists and MMs are willing to trade, even if those prices are in private trading systems. 
 
==> this sounds like a common sense, but was not enforced. now it enforces the transparency to the market. 
==> also, when client order and MM's quote are at the same price, client order must be prioritized. 
 
==> MM gets special perks, like collecting spreads as profit, permission to nake short sale. MM is prone to the accusation of front-running. 
 
### 
###  good papers to read 
### 
 
"the new stock market: sense and nonsense" 
 
"eletronic market making : initial investigation" 
http://www.cs.cmu.edu/~softagents/papers/CR_nevmyvaka_sycara_seppi.pdf 
 
==> discusses how to optimize the spread. what factors to consider. (obviously, as tick-size gets smaller, not as much to exploit.) 
 
(1) initial spread 
look at NBBO and take some interval within that range. 
 
(2) how to update the spread 
- predictive model (based on order book imbalances, short term pattern, etc) 
- non predictive model (based on solely on NBBO spread) 
 
"we can re-formulate this task: adjust the bid-ask spread in such a way that the orders generated by other market participants will transact with the dealer's bid quote and ask quote with the same frequency." 
 
(a) timing 
(b) quote positioning 
- timing should be as fast as possible. when your MM engine analyzes the order book at t0 and decides/updates the spread at t1, already the book may have changed enough that both your ask & bid are placed on the same side of the market, then you end up paying the spread instead of profiting from it. the bottleneck for timing is not computation speed by your c++ engine, but network//QR/infra side. one good modification to the quote positioning problem is instead of always taking spread inside NBBO, take somehting like 3rd-best-bid ~ 3rd-best-ask (or fixed amount of cents away from NBBO), then you are more safely protected from mkt fluctuation, straddling the inside mkt, preserving the property of standing on both sides, plus wider spreads, but less volume. this is where your MM engines has to balance (if using non predictive model). 
 
(c) inventory mgmt 
ideally, MM buys and sells at equal frequency. i.e. MM inventory should fluctuate around zero. but when a stock price is consistently going up for some period of time, then MM's ask gets his more often than his bid, then his short position accumulates, and vice versa. in such a case, MM can adjust its ask deeper inthe book so the hit on ask gets less. 
 
==> formula: QuoteDistance = minDistance + slpha * max (0, inventory - initial limit) / inventory * minDistance 
    i.e. when the position is within the initial limit, the quote is always set minDistance away from the mkt, but if the inventory gets outside the limit, the quote moves further away, encouraging the inventory movement in the opposite direction. 
 
==> non predictive model is not perfect if mkt is moving drastically. we need predictive models too. 
 
 
### 
###  vocab 
### 
 
turnover = the number of shares traded for a period as a percentage of the total shares in a portfolio or of an exchange (as in shares outstanding). 
 
it is a measure of stock liquidity calculated by dividing the total number of shares traded over a period by the avg # of outstanding shares (for that period). the higher the share turnover, the more liquid the share of the company. 
 
for example, if the total amount of shares traded over the year was 10 billion and the avg amount of outstanding shares for the same period was 100 million, then the share turnover for the year is 100. 
 

  1. 2018-10-23 23:55:33 |
  2. Category : financial
  3. Page View:

Google Ads