If we are playing for money, the question arises on how you should cope with counteparty risk. Amongst friends this is clearly not an issue, but if we are trying to simulate trading conditions then we should introduce either margin calling or stop loss.
If we use a spreadsheet to track all settled trades, then it is easy to track PL during the course of the trading sessions. So one approach is for the person acting as the settlement agent does margin calls to all participants so that if they have run out of money, they will be forced to reduce their positions.
This approach seems the most optimal, but can be high maintenance if it involved movement of banknotes. Therefore if a judicator is available who can run the Clearhouse, then they can monitor PL and send out instructions to the exchange members (MM, IB and LT) to cut their positions by half during the following session.
HF can be included in this process for simplicity, however a more accurate event is that they are given, say £2000 in capital and once their PL has broken through this amount, the IBs will have to create an accord to unwind the trades a a level that leaves the HF with zero cash. This creation is what happens when an OTC counterparty defaults to multiple dealers. The MM should make a special trading session to facilitate unwinding of exposure.
Given that the LT are trading on their own account, they have limited capacity and should have a lesser amount of capital compared to the HF, say £1000. If a local is within 5 points of going bust, then the judicator must instruct the MM to take over the positions for a fee of 5 points and bankrupt the LT. At that point, the LT should try and make money by shining shoes, bringing coffee, or get captial funding from a HF.
Friday, January 28, 2011
Thursday, December 23, 2010
Distribution of the Final payoff - sum of the two packs
I noticed that we have treated the distributions of the two packs separately and not looked at the distribution of the final level which is the sum of the 14 cards. So, the final distrubution has a mean of 164 and a standard deviation of 38.9. The histogram for a population of 10^6 looks like:-
Wednesday, December 22, 2010
Introduction of a structuring desk
John Arvanitis mentioned the idea of introducing a structuring desk (SD) that can offer packages of options to the IB and the HF but not pit members. The idea here would be to provide option types of various strikes - calls, puts, binaries plus combinations such as call spreads.
These would be over-the-counter trades which means they are private between the two counterparties and bilateral in nature so would not go through the exchange settlement process. Secondary pricing is provided on request by the SD in the form of a close-out price, settlement at the end of the deal would also be bilateral.
This introduces significant bilateral counterparty risk which is now non-transparent - which reflects the issues faced in 2008/9. This will require a form of waterfall that would enforce settlement with the exchange first before bilateral agreements are considered.
The only condition is that if the package is executed delta neutral (for example a 160 call might be priced at 26 points together with a short position in the traded contract at 164 at a ratio of 10:6) then the contract must settle through the exchange.
This is an interesting additional layer of complexity that has yet to be tested. The ability to price call options throughout the life of the game has yet to be developed. however I did manage to price a series of calls and puts at different strikes based on a simple average of the payoffs of a sample of 10^6 games. The results are as follows:-
Strike__ Call___ Put
100 ____65____ 1.0
110 ____55____ 1.5
120 ____46____ 2.5
130 ____38____ 4.0
140 ____30____ 6.5
150 ____23____ 9.5
160 ____18____ 14
170 ____13____ 19
180 ____9.0____ 25
190____ 6.0____ 32
200____ 3.5____ 40
210____ 2.5____ 48
220____ 1.0____ 57
More analysis on binaries and implied deltas to follow.
These would be over-the-counter trades which means they are private between the two counterparties and bilateral in nature so would not go through the exchange settlement process. Secondary pricing is provided on request by the SD in the form of a close-out price, settlement at the end of the deal would also be bilateral.
This introduces significant bilateral counterparty risk which is now non-transparent - which reflects the issues faced in 2008/9. This will require a form of waterfall that would enforce settlement with the exchange first before bilateral agreements are considered.
The only condition is that if the package is executed delta neutral (for example a 160 call might be priced at 26 points together with a short position in the traded contract at 164 at a ratio of 10:6) then the contract must settle through the exchange.
This is an interesting additional layer of complexity that has yet to be tested. The ability to price call options throughout the life of the game has yet to be developed. however I did manage to price a series of calls and puts at different strikes based on a simple average of the payoffs of a sample of 10^6 games. The results are as follows:-
Strike__ Call___ Put
100 ____65____ 1.0
110 ____55____ 1.5
120 ____46____ 2.5
130 ____38____ 4.0
140 ____30____ 6.5
150 ____23____ 9.5
160 ____18____ 14
170 ____13____ 19
180 ____9.0____ 25
190____ 6.0____ 32
200____ 3.5____ 40
210____ 2.5____ 48
220____ 1.0____ 57
More analysis on binaries and implied deltas to follow.
Monday, December 20, 2010
The role of the Market-Maker
The Market-maker is central to the smooth operation of the market and hence is key for making the game as realistic as possible.
The key aspect is that the MM has no incentive to make money, only to avoid losing a lot. Any money the MM makes goes to the Bar Bill that is expected to pay for the drinks after a long afternoon session. This is not a position to give to a junior member of the group.
Given that approx 60% of total volume goes through the MM, the key function is to move the market to the point of balance. It is therefore better to lose a little money by moving (for example) the spread beyond the original bid-ask and flushing out a couple of quick profits for the other dealers than to take a wrong direction view, in particular between trading sessions. The tails of the distribution in the second pack include numbers that can move the market by +/- 20 points.
Compensation for the MM only comes from the bid-ask spread which at a fixed 3 point spread so long as he can find a balanced market and get sufficient volume done.
Unlike the LT who can shout out Bid or Ask prices at any time, the MM also is required to make a two-way price at all times. This is tough to maintain but is an important component to maintain a functioning market. This is particularly true during the last trading session where the Hedge funds have to unwind all of their positions. sometimes with size that is multiple the 5 that the MM is good for.
To keep the noise level up, the MM should continue to shout out the market levels during the trading sessions.
One last aspect of the MM is to keep track of he open interest, less on the PL which should not be a distraction, but more on the net position. Target is always to be close to home when the closing bell calls. It is surprisingly difficult to maintain the bid-ask price, deal, create deal tickets and keep track of open position at the same time. Hence the pads!
The key aspect is that the MM has no incentive to make money, only to avoid losing a lot. Any money the MM makes goes to the Bar Bill that is expected to pay for the drinks after a long afternoon session. This is not a position to give to a junior member of the group.
Given that approx 60% of total volume goes through the MM, the key function is to move the market to the point of balance. It is therefore better to lose a little money by moving (for example) the spread beyond the original bid-ask and flushing out a couple of quick profits for the other dealers than to take a wrong direction view, in particular between trading sessions. The tails of the distribution in the second pack include numbers that can move the market by +/- 20 points.
Compensation for the MM only comes from the bid-ask spread which at a fixed 3 point spread so long as he can find a balanced market and get sufficient volume done.
Unlike the LT who can shout out Bid or Ask prices at any time, the MM also is required to make a two-way price at all times. This is tough to maintain but is an important component to maintain a functioning market. This is particularly true during the last trading session where the Hedge funds have to unwind all of their positions. sometimes with size that is multiple the 5 that the MM is good for.
To keep the noise level up, the MM should continue to shout out the market levels during the trading sessions.
One last aspect of the MM is to keep track of he open interest, less on the PL which should not be a distraction, but more on the net position. Target is always to be close to home when the closing bell calls. It is surprisingly difficult to maintain the bid-ask price, deal, create deal tickets and keep track of open position at the same time. Hence the pads!
Suggestions for Materials
The game should be played as paper-based as possible. The action of settling the trades should be done manually but the record-keeping can be via a spreadsheet. OK Materials:-
1) Cards for the two packs - ideally use cards similar in size and thickness to a business card - they need to be shuffled, dealt and not be transparent. Therefore you need 40 for the first pack and 100 for the second pack.
2) Trading cards. There should be a deal ticket for each buy and each sell, so again the size of a business card works well, enough to write the dealer, the counterpart, the buy or sell size and price. Card is better as the MM needs to stand in the pit with a pack and deal quickly. In a recent game we closed just over 100 deals, so you need north of 200 cards or ones that can be reused
The best way to deal in the pit is for one member of the IB to come to the pit members, ask for a price, deal, enter the details on their ticket, then take the pit members deal ticket and hand it over to the person who is running the settlement. Presenting both sides of the deal massively reduces settlement burden.
3) ID tags. Each group of HF, IB and Pit members should come up with a two letter identifier and display on a badge when dealing, this speeds up the settlement process.
4) Pads etc, very useful for each group to keep a tab on their outstanding positions. Ultimately the prime record of each deal lies with the settlement person. More on their role in following blogs.
5) Spreadsheet for settling trades, once entered the deal size and economics are binding.
6) Alarm for the MM to sound the end of each trading session (pen on a glass works!)
1) Cards for the two packs - ideally use cards similar in size and thickness to a business card - they need to be shuffled, dealt and not be transparent. Therefore you need 40 for the first pack and 100 for the second pack.
2) Trading cards. There should be a deal ticket for each buy and each sell, so again the size of a business card works well, enough to write the dealer, the counterpart, the buy or sell size and price. Card is better as the MM needs to stand in the pit with a pack and deal quickly. In a recent game we closed just over 100 deals, so you need north of 200 cards or ones that can be reused
The best way to deal in the pit is for one member of the IB to come to the pit members, ask for a price, deal, enter the details on their ticket, then take the pit members deal ticket and hand it over to the person who is running the settlement. Presenting both sides of the deal massively reduces settlement burden.
3) ID tags. Each group of HF, IB and Pit members should come up with a two letter identifier and display on a badge when dealing, this speeds up the settlement process.
4) Pads etc, very useful for each group to keep a tab on their outstanding positions. Ultimately the prime record of each deal lies with the settlement person. More on their role in following blogs.
5) Spreadsheet for settling trades, once entered the deal size and economics are binding.
6) Alarm for the MM to sound the end of each trading session (pen on a glass works!)
Thursday, December 16, 2010
Distribution of the second pack
Using the example below, the histogram of the pack is some distance from looking like a normalised distribution.
Note the forced lack of a zero which was put in to make the game more interesting, but looks like the resultant distribution has shifted to the right a bit. Not sure it works. The distribution of the sum of the six cards chosen at random has a mean of zero (no surprise) and a standard deviation of 25.25. The histogram looks like based a sample of 100k.
Distribution of the first pack
The distribution for the sum of eight cards to be picked out of a list of cards numbered 1 to 40 has an average of 163.9 and standard deviation of 29.5. The histogram shows its distribution based on a sample of 100,000.
Subscribe to:
Posts (Atom)