hw#4
hw#4
hw#4
Recall the simulation of the Asian option that was modeled in class. Now Trusty’s client,
the petroleum refiner, is considering a related “lookback” option as an alternative to the
original Asian variety. In this problem we will consider the mechanics of evaluating the
lookback option.
The lookback option that the refiner wants to price works as follows. As before, at
the end of each day of the month, the bank records the spot price of crude oil. At the
end of the month the bank calculates two quantities: the average of the daily closing
prices, and the minimum closing price for the month. For the lookback option, the
payoff to the refiner equals the average minus the minimum.
Model the payoff calculation for (one sample of) a simulation that prices the lookback
option. As in the lecture notes, you should 1) describe all data and variables; 2)
explicitly define the type and parameters of all random variables that model the
uncertain elements of the system; 3) write out mathematical expressions that define
how the payoff is calculated at the end of the month. You may assume that the
dynamics of oil-price changes are the same as those described in class.
Use Crystal Ball to simulate the model developed in part (a). Run the simulation for
1000 trials using random number seed 1234. Print out a copy of the summary
statistical data and frequency chart generated by the simulation and include it in your
write-up. Using this information, briefly answer the following questions:
i) Of these 1000 trials, roughly what fraction of the time is the forecast payout at
least $2.75. That is, what fraction of the time is the average price per barrel at
least $2.75 above the minimum?
ii) Construct a 95% confidence interval for the expected payoff based on the
summary statistical data generated by Crystal Ball.
iii) What is the relationship between the frequency distribution analyzed in part (i)
and confidence interval you constructed in part (ii)?
iv) How will the frequency distribution of part (i) and confidence interval of part (ii)
change as the number of trials increases?
Medtronic designs and manufactures medical devices for a wide variety of purposes. In
many cases clinical medical trials may be required to ascertain the safety and
effectiveness of a new or redesigned device. In some cases, however, clinical trials may
not be required. In these cases, Medtronic nevertheless may run its own training tests,
which it calls Limited Market Release (LMR) protocols.
The attached document is an example of a Medtronic LMR Protocol, which tests a new
generation of medical device at a number of medical centers across Western Europe.
The device is to be tested at 17 centers, and Criterion 1 on page 4 describes the
minimum numbers of cases that must be run before the LMR Protocol can be
considered complete. (It is not necessary to read the LMR document to answer this
question. But feel free to read it if you are curious.)
Once an LMR protocol is successfully completed, Medtronic releases the new device to
the general market. To support this general market release, Medtronic ramps up
production of the new device and, at the same time, it phases out production of any
older generation device that may be replaced.
Because LMR cases arrive only sporadically to the medical centers performing the tests,
the time required to complete an LMR can be long and uncertain. This uncertainty, in
turn, drives uncertainty in the timing of the production ramp-up decision. Medtronic has
developed Monte Carlo simulations that it uses to assess and manage the uncertainty
stemming from its LMR protocols.
For this question, you will analyze a model we developed for a single medical center,
and you will extend the model to analyze cases run at two medical centers.
Given these assumptions, the spreadsheet simulates the number of weeks it will take
for the LMR protocol to complete. The model is defined as follows.
Time index
t = 1,…,20 for each of 20 weeks. (We model 20 weeks of potential cases.)
Random variables
Xt = number of potential cases encountered at the center in week t
Xt ~ Poisson with mean 1
Cumulative Cases
Ct = cumulative number of cases handled at the center through week t
C1 = X1
Stopping Condition
Nt = 1 if protocol not complete by the end week t
1, if C t 5
Nt =
0 if C t 5
i) Based on your results, roughly what is the probability that the number of weeks
required to complete the trials is 4, 5, or 6?
ii) Construct a 95% confidence interval for the expected number of weeks required
to complete the trials.
iii) If the sample size were increased from 10,000 to 40,000, what would you
estimate would be the new sample standard deviation? Explain.
b) Suppose now that the device is to be tested at 2 medical centers. In this case, to
complete the protocol, each center is required to use the device at least 4 times and,
together, the two centers must test the device 10 times. (Thus, it would be sufficient
for one center to complete 4 cases and the other 6; it would also be sufficient for
each center to complete 5 cases.) Again, assume that number of potential cases
encountered by each of the centers each week follows a Poisson distribution with a
mean of 1 (and that the numbers of independent of each other). How many weeks
will it take for the protocol to complete?
i) Construct a Monte Carlo simulation to evaluate the number of weeks it will take
for the LMR Protocol to complete. (Hint: construct separate simulations for each
of the two centers and then model the condition that, together, they reach 10
cases.) You should explicitly describe the formulas that define how your
simulation model works. (For an example, see part (a).)
ii) Run 10,000 trials of your simulation. Report the frequency distribution and
summary statistics for the number of weeks required to complete the protocol.
c) Compare your results from parts (a)(ii) and (b)(ii). Are the results of the two
schemes significantly different? Cite specific statistics that support your judgment.
What is driving the difference?
Revision A
January 6, 200X
Marketing XXXXX
Clinical Affairs XXXXX
Regulatory Affairs XXXXX
Core Team Leader XXXXX
Scope of Feedback:
During this LMR, our most experienced physicians will have earlier access to XXXXX than other physicians
so that they can give us feedback on product usability and provide feedback on any additional product in-
service and training requirements.
Because the focus of this LMR is physicians’ preference and any additional training requirements and does
not collect any patient-specific information, we anticipate this feedback to cover physicians’ experience
during their time in the operating room and do not anticipate needing Ethics Committee approval to complete
the questionnaire. Nevertheless, some hospitals may require Ethics Committee approval due to their own
internal processes.
During the LMR, we will conduct a Physician Preference Questionnaire in which we will monitor any
suggestions that the physicians have about our troubleshooting, product in-service, and product use.
This first group of physicians will have access to the products after LMR completion and pending Full
Market Release (FMR). Pending acceptance from them, we anticipate the following physicians to be
included in the LMR:
Timing of LMR:
The LMR will start right after Phase 4 review (March 10). Once the LMR is complete, we will return to
Corporate Review to ask permission to begin the full market release (FMR). We anticipate this occurring X
months after the start of LMR on approximately March 10.
Appendix:
1.) Acceptance Criteria
Acceptance Criteria
If any of the below criteria are not met for this Limited Market Release (LMR), all data will be reviewed in order to
consider the following scenarios: 1) continued LMR pending resolution of issues (and plan to do so), 2) Full Market
Release (FMR) (with re-evaluation of criteria, in view of all data), or 3) discontinuation of product until full resolution
of issues.
Criterion 2: Physician Training Recommendations Must be Synthesized into Documents and Practices
There will be several opportunities for physicians to offer suggestions for our training program. We will collect data
via the questionnaire during each case. In addition, we will get initial suggestions from them during the Train the
Trainer meeting. We will review all of their suggestions (XXXXX, head of European Training, will be responsible for
this role along with help from two physician proctors, XXXXX, and XXXXX) and make sure that we integrate
feedback that our training department determines to be valuable into our training materials and practices, and
Implantation and Troubleshooting Guide. We will consider this criterion met when XXXXX indicates that she and her
team have integrated the training suggestions that we have received.
In addition, we will plan to produce a Troubleshooting Guide that will be distributed through our sales force
and will complement the IFU. Suggestions made in the Troubleshooting Guide will also be rolled into our IFU during
its next worldwide revision. Only if there are recommendations in the IFU that we believe could cause serious injury to
a patient will we plan to revise the IFU prior to FMR.
Criterion 3: Summary of LMR Questionnaire Responses Provided to Participating Physicians, Corporate Review,
and Business Area Senior Management
XXXXX (XXXXX product manager) will be responsible for summarizing the major findings of the LMR
Questionnaire responses. This summary will be shared with at least a subset of LMR participating physicians to
confirm any conclusions drawn in the Questionnaire Summary. We will invite the leading implanters from the LMR
onto a conference call in which they can provide feedback on the Summary. The summary will then be reviewed and
approved by XXXXX Senior Management (XXXXX and XXXXX at a minimum) prior to Full Market Release.
We anticipate that physician preference for XXXXX will at least match that of the predecessor product. As a
result, if the score for questions 1 through 8 on the physician questionnaire does not have an average score of at least 3
out of 5, we will raise this issue with all of XXXXX Senior Staff before recommending continuance to FMR.
Criterion 4: Summary of Any Reported Adverse Events and Root Cause Analysis
While we think an adverse event is unlikely to happen given the LMR centers’ experience, we need to consider this
possibility during the 35 cases planned, given the overall XXXXX complication rates with current products on the
market. As a result, for any adverse events reported during the LMR, XXXXX will study the event and determine (as
well as possible), whether it was related to the patient condition, the procedure itself or any other cause. A report of
these events and any ways in which XXXXX may mitigate their occurrence in the future (such as via training or
patient inclusion criteria), will be presented along with the Summary of the LMR Questionnaire and will be provided
to both Corporate Review and XXXXX Senior Management. XXXXX will be responsible for making sure that we
generate this Summary.