Four New Graphs

Sally,
While I think it makes sense for this type of information to be available,
that it should be done by Operations, and that Eugenio has the skills and
knowledge to articulate it well, I have some concerns. First and foremost,
the process of creating a standard risk metric belongs to RAC, I believe. To
me this is no different than VAR, or the Capital Pricing methodololgy, or the
RAROC process. This has been published without even a review from our
group, which I think is inappropriate for two reasons: 1) we might disagree
and, as stated above, I think we have the final say, and 2) we can't quality
control it and all the reciepients think that I have. Second, this has been
developed in isolation when in fact we have been doing similar things for the
past 2 years. We report these to the Board and have published to the Desk
heads on a monthly basis returns by trader. Clearly, there could have been
some benefit in each direction if we were told that this was a goal of this
group, particularly when we fundamentally agree that this group should be
doing these things. Thirdly, it is a little irritating when someone sets up
a process in isolation for what may appear to be "show-time" when my group is
spending hours and hours analyzing the output of existing process because the
data integrity is so bad (double mapping of portfolios, books not
officialized......) - we have at least attempted to take both parts of the
equation, Eugenio has chosen to focus on the cool part. Now I would be a
little less concerned if on almost every day, Eugenio wasn't in my morning
meeting contributing virtually nothing, i.e., this doesn't appear to be
oversight on his part.

I would prefer to not make a big deal out of this, but rather take it as an
opportunity to clarify and shift some responsibilities. In the big scheme of
things it is a move in the right direction.

Penny for your thoughts
Ted
---------------------- Forwarded by Ted Murphy/HOU/ECT on 09/08/2000 09:23 AM
---------------------------

Eugenio Perez
09/07/2000 05:17 PM
To: John J Lavorato/Corp/Enron@Enron, John Sherriff/LON/ECT@ECT, Jeffrey A
Shankman/HOU/ECT@ECT, Ted Murphy/HOU/ECT@ECT, Mark Frevert/NA/Enron@Enron,
Rick Buy/HOU/ECT@ECT, Sally Beck/HOU/ECT@ECT, Shona Wilson/NA/Enron@Enron
cc: Michael E Moscoso/HOU/ECT@ECT, Maria Teresa
Aguilera-Peon/Corp/Enron@Enron, Jennifer Nguyen/Corp/Enron@ENRO
Subject: Four New Graphs

We have created four additional graphs to be published daily along with the
DPR:

Return on VaR (ROVAR)
Curve Shift P&L (Delta - Gamma) Sharpe Ratio
Risk - Return (Large)
Risk - Return (Small)

All of these graphs deal with the concept of risk-adjusted return. The
markets in which we operate have unlike risk characteristics. Furthermore,
Enron's appetite for risk also differs in each market. As a result, it is
difficult to compare returns in absolute terms. Therefore, the approach of
these graphs is to divide return by a measure of risk, in effect restating
return numbers into a standard base (how many dollars we earn for every
dollar we put at risk).

For each business unit or commodity group, the ROVAR graph shows an average
of daily total P&L divided by an average of daily VaR. Higher ROVARs
indicate better performance (relative to other business units). For each
business unit or commodity group there are two bars: the dark blue on the
left shows ROVAR on a 4 week (20 business day) rolling basis, and the light
blue on the right shows ROVAR on a 1 week (5 business day) rolling basis.
Because of the shorter term, the 1 week ROVAR is more sensitive: it can be
quite high or low, and it changes frequently.

The Sharpe Ratio graph is very similar in concept; it shows an average of
daily curve shift P&L (both delta and gamma) divided by its standard
deviation. This Sharpe ratio is uncomplicated by non-trading P&L (like new
deals, reserves, and so forth). Furthermore, whereas VaR is a forward
looking statistic (the most we are likely to lose in one day with a given
probability), standard deviation is backward looking (the actual volatility
of our returns over the period). This graph also has two bars for every
business unit or commodity group: a dark brown on the left showing the Sharpe
ratio on a 4 week rolling basis, and a light brown on the right showing the
ratio on a 1 week basis.

The Curve Shift P&L Sharpe Ratio and the ROVAR graphs complement each other
well. They measure different things, but if our P&L is composed mostly of
curve shift, and if our VaR closely predicts curve shift volatility, the two
statistics should be very close.

Sometimes it is interesting to see the actual magnitude of the P&L relative
to the risk. We therefore put together the Risk - Return graphs. On the
vertical axis, they show 4 week rolling average P&L. On the horizontal, they
show 4 week rolling average VaR. The various business units are plotted as
points on the graphs. These graphs are closely related to the 4-week ROVARs,
in that the vertical value of every point divided by its horizontal value is
the business unit's ROVAR (average P&L / average VaR). Another way of
looking at these graphs is that the slope of a line connecting the origin
(0,0) with the plotted point of a business unit is its ROVAR. Books with
small risks and returns (< $2 million) are plotted on the "Small" graph, and
the rest, on the "Large".

These graphs are based on the DPR Database that we built to standardize and
capture the information that is reported in the DPR. Most, but not all of
the information in the DPR currently flows from the database. As a result,
we do not have ROVARs or Sharpe Ratios for every business unit. As we
continue to set up exports into the database, we will fill in the missing
numbers.

We will e-mail you these graphs every day until we can publish them on the
intra-net. Please remember to launch the file in Excel, instead of using the
Lotus Notes viewer. The VaR Limit Usage graphs that we created last week are
already on the web. We hope that you find all of these useful.

Regards,

Eugenio Perez