►
Description
In this AMA, Primoz (Risk Core Unit Facilitator) and Jan (Risk Core Unit team member), we discussed a wide range of Risk parameters within the DAO. The team also presents some of the updates and progress on their dashboard, as well as general team priorities.
Contact Info:
Risk Team:
Forum - https://forum.makerdao.com/g/Risk-Core-Unit
Discord - https://discord.gg/z3Xu7uQQ
Primoz:
Forum - https://forum.makerdao.com/u/primoz/summary
Discord - Primoz#9314
Jan
Forum - https://forum.makerdao.com/u/josolnik/summary
Discord - Josolnik#6878
Payton:
Forum - https://forum.makerdao.com/u/jerryag/summary
Discord - @Jerry.G#6157
LinkedIn - https://www.linkedin.com/in/jerry-g-188a8a210/
*Disclaimer: This update is assembled by the MakerDAO community. This does and should not serve as financial advice or an inducement to vote in any specific way. This update is purely for informative and for communications purposes exclusively.*
A
Hello,
everybody
and
welcome
here
my
name
is
peyton.
I
go
by
pros11
online
and
I'm
happy
to
be
hosting
this
ama.
Today,
we've
got
pramosh
the
facilitator
of
the
risk
core
unit
over
at
make
it
out
here
to
answer
some
questions
about
what
the
unit
does
and
how
all
that
goes.
So
hi
everyone,
hey
thanks
for
joining
us
bourbon,
just
as
some
housekeeping
matters
that
we
will
ask
people
to
keep
their
video
off
unless
they're
asking
a
question.
A
This
is
just
for
the
recording,
no
worries
just
for
later
on,
just
to
keep
it
a
little
clearer
and
it
is
an
ama.
So
we
do
want
you
guys,
asking
questions.
I've
got
some
predefined
ones
here,
but
it's
more
interesting
if
we
get
some
live
interaction.
So
if
you're,
not
in
a
spot,
where
you
can
ask
it
yourself,
throw
something
up
in
the
chat
and
I'm
happy
to
ask
it
for
you,
otherwise,
we'll
kind
of
make
it
happen
here
so
yeah
perish.
A
B
Yeah
sure
so,
hey
everyone,
I'm
pretty
much,
I'm
a
risk
facilitator,
so
facilitator
of
risk
or
unit
which
was
officially
established
in
march
this
year.
I
think
we
were
one
of
the
first
core
units
to
establish,
but
essentially
I've
been
working
on
risk,
since
I
think
it's
2019
february,
and
before
that
I
mean
in
that
time,
cyrus
unesc
was
the
was
the
risked
lead,
but
then
last
year
I
think
in
september
I
took
the
the
risklet.
B
So
at
that
time
it
was
only
basically
me
and
marco
who
were
working
on
risk,
so
it
was
quite
quite
intensive.
Now
it's
actually
seven
full-timers,
so
the
team
actually
grew.
It
was
much
easier
to
scale
because
you
know
we
got
our
budget.
We
have
our
milestones
so
yeah.
It's
it's
much
easier
to
do.
Work
right
now,
and
essentially
our
team
actually
consists
of,
is
split
in
two
there's
actually
four
of
analysts
and
there's
four
developers.
B
So
I
would
say
this
is
essentially
our
core
part
of
of
our
business
or
our
core
unit.
It's
it's
really
about
consultancy
right
about
proposing
certain
parameters
so
ask
for
actually
you
know
do
this,
but
in
order
to
you
know,
have
good
arguments
and
good
underlying
data.
We
use
other
part
of
the
team
who's
actually
involved
in
in
data
science
in
tooling
and
dashboards,
and
this
kind
of
stuff
yeah.
So
that's
that's
about
introduction.
I
would
say.
A
Yeah,
that's
excellent
and
it
is
kind
of
neat
going
from
you
know
being
with
us
before
the
core
units
to
being
one
of
the
first
core
units
on
board
at
maker
kind
of
gave
you
the
chance
to
really
define
your
your
vision
and
mandate
and
and
what
you'd
like
to
see
for
maker
dow.
So
I
was
curious
if
you
wanted
to
talk
about
that
a
little
bit,
maybe
just
in
terms
of
how
you
went
about
setting
your
mandate
and
and
how
you
view,
risks
just
just
more
broadly.
Within
the
protocol.
B
Yeah,
so
the
vision
of
our
core
unit,
the
way
I
see
it
so
there's
there's
like
operational
vision
that
we
have
and
then
there's
like
organizational
house
settings
so
from
operational
perspective.
What
the
end
goal
would
be
or
kind
of
already
is,
is
that
we
we
want
to
utilize
this
concept
of
automated
governance.
B
It's
actually
gauntlet
introduced
one
of
the
first,
which
is
also
another
defy
risk
team,
and
we've
been
essentially
doing
that
for
time.
But
you
know
there's
so
much
more
to
be
built
around
this,
and
what
this
exactly
means
is
that
you,
essentially
you
build
this
risk
model
for
all
kind
of
things
at
maker
right,
and
you
really
focus
on
the
methodology
itself
and
then
you
have
a
set
of
inputs
and
you
get
outputs
and
the
whole
idea
is
the
outputs
should
be
related
to
gardens
parameters.
B
So
there
should
be
garnish
parameters
and
the
inputs
are
sometimes
taken
directly
from
unchained
data,
but
sometimes
also
not
right,
so
we're
essentially
young
later
on
we'll
do
a
demo
on
our
dashboards
and
we'll.
You
know
explain
how
we
do
certain
calculations
and
you'll
see
that
we're
using
a
bunch
of
inputs,
while
we
calculate,
for
instance,
risk
premium
of
certain
world
type
and
yeah.
Essentially,
this
is
like
the
core.
You
know
mandate
how
we
wanna
do
things.
B
The
whole
idea
is
that
there
is
total
transparency,
so
all
the
risk
models
are,
the
methodology
is
all
known.
The
inputs
are
also
all
transparent,
and
you
know
community.
If
they
have
some
expertise,
can
actually
check
by
themselves.
You
know
how
we're
when
we're
defining
certain
there
is
premium,
they
can
actually
see.
B
You
know
what
kind
of
inputs
we're
using
and
with
this
there
are
certain
challenges
and
that's
because
measuring
risk
is
can
be
sometimes
really
subjective
when
it
comes
to
inputs,
because
you
know
you
can
do
you
can
have
some
empirical
evidence
on
on
some
inputs
users,
but
then
it's
always
a
bit
subjective,
and
why
is
this
like
really
important?
That
maker
is
is
because
of
the
nature
of
risk.
I
usually
like
to
say
so,
if
you
think
about
it,
how
banks
usually
do
business,
so
you
know
they
do
certain.
B
B
B
You
know
just
something
technically
wrong
and
you
get
a
loss
from
it
right,
and
this
is
something
it's
really
hard
to
to
evaluate,
and
and
this
is
why
our
all
our
financial
models
are
very
conservative,
so
that
we
try
to
account
for
that
other
part
of
unknown
risk,
which
is
impossible
to
predict
yeah.
So
this
is
this
is
what
I
would
say
regarding
vision
and
challenges
and
the
other
part
of
the
vision
is
yeah.
It's
just
operational,
just
few,
maybe
a
few
more
words
on
that.
Sorry,
organizationally!
B
It's
it's
now,
seven
of
us.
So
it's
much
easier
to
do
work
you
know
compared
to
one
year
ago,
but
but
still
how
how
I
see
a
risk
team
evolving.
Of
course,
you
could
have
many
different
risk
teams.
That's
that's!
Definitely
one
one
of
the
options.
The
other
way
of
scaling
within
our
core
unit
was,
in
my
mind,
has
always
been
that
we
could
have
specialized.
B
You
know
sub-core
units
or
domains
within
core
unit
where
you
essentially
have
a
team
of
you
know
specialized
analysts
who
were
just
focused
on
auctions,
for
instance
right
when
you
do
auctions,
you
need
more.
You
know.
Maybe
it's
a
bit
more
of
a
statistical
or
quant
knowledge,
so
you
would
have
this
kind
of
expertise
within
the
core
unit.
It
would
cover
just
this,
and
why
is
this
good?
B
I
think,
because
why
this
hierarchical
structure,
I
think,
is
still
good,
because
all
these
sub
domains
within
the
risk
are,
you
know
interrelated
and
they
need
to
share
stuff
between
each
other.
So
that's
that's
one
of
the
key
and
then
the
challenge
of
this.
What
it
is
is
that
it's
just
hard
to
find
talent.
So
that's
that's.
Definitely
something
on
my
radar.
Lately
to
attract
experts,
you
know
who
know
defy
who
know
maker,
but
on
top
of
it
they're
specialized
in
on
on
certain
fields.
So
we
can
use
them.
A
Yeah,
that's
definitely
a
challenge.
I
see
a
lot
with
us
over
gov
alpha
2
right,
it's
like,
in
addition
to
kind
of
the
subject
knowledge
you
need,
then
you
need
this
whole
other
intimate
knowledge
of
of
the
protocol
and
how
it
relates
to
other
things.
So
definitely
not
a
large
talent
pool
you're
you're,
drawing
from
there.
A
A
If
nana
could
just
be
zoom
cool,
zoom,
b
and
zoom,
I
was
gonna.
Ask
that
frozen.
I
feel
like
it
might
be
hard
for
the
average
like
community
member
to
envision,
like
what
all
does
the
risk.
Your
unit
do
like
in
terms
of
the
bulk
of
your
work,
how?
How
is
that
time
spent
like
you
mentioned,
that
you
do
split
between
like
kind
of
the
development
side
and
the
actual
risk
analysis
side,
but
I
was
wondering
if
you
could
maybe
give
a
little
insight
into
between
those
groups.
B
Yeah
sure,
so
most
of
the
time
that
we
spend
is
is
really
related
to
all
sorts
of
stuff,
but,
like
the
the
most
obvious
is
you
know,
we
have
a
risk
model
for
every
collateral
type
and
you
all
know
that
each
collateral
type
has
a
bunch
of
parameters.
Definitely
depth.
Ceilings
are
one
of
the
core
parameters
that
we
need
to
set
and
propose
and
again
we're
relying
on
the
risk
model.
B
Yeah,
so
there's
there's
definitely
depth
sealing
and
there
are
risk
premiums
that
we
calculate.
So
this
is
usually
covered
within
the
open
market
committee,
so
each
month
maker
has
this
market
committee
that
actually
actually
proposes
new
rates
and
new
debt
ceilings.
If
it's
not
like
needed
even
earlier
right
at
some
point,
we
need
to
react
earlier,
but
like
on
monthly
cadence,
we
have
this
group
of
people
which
we
risk
is
part
of
where
we
essentially
share.
B
You
know
whether
the
liquidity
of
an
asset
changed,
or
you
know
something
fundamentally
changed
or,
or
you
know,
volatility.
You
know,
whatever
related
financially
to
an
asset
changed
and
we
need
to
react
with
the
lower
depth
ceiling.
For
instance,
or
you
know,
we
see
that
volts
are
keeping
low
collateralization
ratio
and
therefore
we
might
want
to
increase.
B
The
first
parameters
that
we
started
calculating
already
some
time
ago
and
then,
of
course,
there
are
collateral,
onboarding
evaluation.
This
is
the
other
core
part.
Again.
Both
parts
of
our
team
are
actually
cooperating
on
this.
So
analysts
do
the
assessment,
they
write
it,
but
they
you
they
need
bunch
of
data,
so
they
can.
You
know,
justify
any
kind
of
parameters,
and
here
there's
a
lot
of
time
spent
on
just
you
know,
there's
like
probably
at
least
one
newly
onboarded
collateral
per
month.
Sometimes
it's
not
even
collateral.
B
B
B
B
I
won't
go
into
details
here,
but
there
are
certain
duration
risks
at
maker
right.
If
you,
if
you
offer
fixed
trade
products,
you
need
to
make
sure
that
you
have
enough
capital
available,
so
you
don't
affect
the
pack
and
by
capital
available
these
days
it
usually
means
how
much
stablecoin
packing
you
hold.
B
A
A
B
B
Is
that
if
we
have
a
bad
debt
event
and
we
needed
to
raise
capital
to
to
cover
the
the
shortfall,
it's
always
good
to
have
you
know
liquid,
and
you
know
order
books
or
being
better
than
you
know
would
be
otherwise,
and
that's
why
you
would
support
any
market
making
activity
regarding
the
deal
with
winter
mute.
I
just
know
that
the
revised
deal
now
looks
much
better.
I
think
the
call
option
is
is
now
at
a
much
higher
price.
B
B
B
I've
heard
like
this
is
a
standard
practice,
although
I'm
not
sure
if
it
really
is
because
most
of
the
teams,
I
know
they
have
some
kind
of
entity
right.
So
even
if
it's
uncollateralized,
they
can
still,
you
know,
claim
you
know
if
they
don't
return
it
in
some
sense,
or
you
know
something
goes
wrong.
This
couldn't
be
done
at
maker
because
we
don't
have
an
entity,
so
maybe
there's
like
additional
risk
but
like
in
essence
this
these
counterparties
are
trusted
until
they're,
not,
but
I
could
see
like
in
d5.
B
You
know
I
have
you
know.
I
have
a
bunch
of
products
which
actually
do
uncollateralized
loans,
so
there's
trufi
and
I
think
maple
finance
and
I've
seen
winter
mutes
there.
So
people,
actually,
you
know,
believe
that
they
have
so
much
reputation
on
the
line
that
yeah
they
would
there's.
There's
there's
close
to
zero
counterparty
risk
and
they
do
trust
them.
So
that's
I
mean
that's
one
angle:
how
to
look
at
it,
but
yeah.
A
Very
neat
and
kind
of
cool
to
hear
how
how
you
go
about
approaching
something
like
that.
I
do
want
to
make
sure
we
have
plenty
of
time
for
like
the
dashboard,
demo
and
stuff.
I
don't
know
now
it's
like
a
natural
time
or
if
there's
some
more
stuff
you
wanted
to
cover
before
we
hopped
into
that,
but
I
just
find
the
dashboard,
so
cool
would
love
to
ask
some
questions
about
it.
A
A
C
Okay,
perfect,
so
this
is
basically
the
landing
page
of
the
dashboard,
and
the
first
thing
that
we
can
see,
let's
say,
is
the
is
the
aggregate
metric
for
the
capital
at
risk.
This
is
something
that
let's
say
pretty
much
can
talk
more
about
later,
but
the
most
important
thing
here
is
also
the
relationship
with
the
surplus
buffer.
So
that
means
the
multiplier
of
capital
risk
and
comparing
that
with
the
surplus
buffer,
which
is
at
the
moment,
220
percent.
C
C
So
that
means,
in
the
background
we
are
scoring
each
of
the
votes
with
the
whole
protection
score,
whether
it's
a
high
risk,
medium
risk
or
low
risk
and
there's
a
horrible
methodology
behind
this,
for
which
we
posted
a
post
on
the
maker
forum
recently,
and
we
can
see
that
60
of
of
total
debt
is
low
risk
and
about
five
or
six
percent
is
high
risk
and
other
than
that.
We
can
also
see
when
we
look
at
the
debt
exposure
and
given
a
certain
clutter
price
drop.
C
We
can
see
how
much
that
would
liquidate
it
if
none
of
them
was
would
protect
themselves.
This
is
clearly
a
very
conservative
scenario,
so
but
it's
more
of
a
mental
model.
Okay,
how
well
are
the
high
risk
volts
protected,
as
you
can
see
here,
there's
a
bunch
of
different
tabs,
so
I
will
focus
just
on
the
on
the
on
the
essential
parts
and
if
you
have
any
questions,
just
filter
feel
free
to
drop
a
question
here.
C
C
The
only
difference-
let's
say
we
see
as
an
outlier-
is
that
if
b
volt,
which
has
a
quite
extreme
risk
premium,
which
takes
which
contributes
to
the
capital
risk
intent
right
and
then
we
can,
then
we
can
dive
deeper
into
this
and
see
that
quite
quite
a
big
chunk
of
the
that
exposure
is
either
high
risk
or
medium
risk.
The
red
one
is
high
risk
and
the
yellow
one
is
medium
risk,
and
if
we
compare
this
to,
let's
say
if
a
volt
type,
we
can
see
that
these
are
basically
the
next
volts
here.
C
We
can
see
that
much
a
much
larger
share
of
that
exposure.
It's
it's
basically
low
risk
right
and
that's
also
impacts
the
risk
premium
in
the
end,
in
the
background,
when
we're
taking
into
account
a
lot
of
factors
right
and
then
in
the
risk
part,
there
are
also
other
charts
which
I
can
also
go
deeper
into.
But
for
the
sake
of
time
I
will
not
at
the
moment,
but
one
one
interesting
aspect
is
also
that
we
are
tracking
the
capital
at
risk
over
time,
and
then
we
can
see.
C
Let's
say
if
for
a
certain
collateral
for
a
certain
wall
type,
if
it
increases
or
decreases,
we
can
look
deeper
into
that,
because
we
have
all
the
way,
let's
say
from
the
portfolio
level,
to
the
wall
type
level
to
default
level.
We
can
see
how
much
this
is
affected
and
this
can
be
obviously
tracked
continuously
over
time.
C
So
that's
that's
it
for
now.
So
if
anyone
has
any
questions,
we
can
also
dive
into
that,
or
maybe
just
another
thing.
Let's
say
here:
we
can,
we
can
see
per
volt
per
individual
volt
in
the
if,
if
b,
volt
type,
we
can
see
that
the
the
risk
premium
is
quite
high.
That's
something
that
we
basically
looked
at
ourselves,
because
the
biggest
the
largest
volts
are
have
a
quite
high
that
have
a
high
that
exposure
are
either
medium
risk
or
high
risk.
C
And
then
we
can
see
that
certain
are
low
risk
because
they
are
attacked,
as
if
I
say
saver,
they
subscribe
to
their
services
and
and
what
we?
What
would?
What
we
could
see
is
that
the
most
of
the
volts
that
are
ideal,
medium
risk
or
high
risk
are
those
that
that
have
basically
been
opened
recently.
So
that's
why
they're
they
are
defined
as
as
either
medium
range
or
high
risk
and
over
time,
when
they're
they
would
be
more
stress
tested.
C
They
will
basically
turn
into
low
risk,
and
this
will
also
lower
the
the
capital
at
risk
echogate
metric
on
the
portfolio
level.
So
that's
something
some
logic
behind
all
these
dashboards,
but
there
there's
also
other
components
such
as
the
the
d3
model,
which
we
built
recently,
which
I
will
not
go
deeper
into
right
now
and
then
there
is
also
the
defy
monitor
which,
where
we
check
across
different
landing
protocols,
what
is
the
market
share
in
the
locked
assets?
C
Let's
say
if
nwbtc,
basically
the
the
main
assets,
and
then
we
track
the
d
ferrates
and
so
on,
and
that's
that's
quite
informative.
But
let's
say
when
we
set
the
the
rates
yeah
so
and
the
other
thing
that
also
pretty
much
mentioned
is
that
the
community
can
can
play
with
with
different
parameters.
We,
our
team,
sets
the
defaults
and
then
you
can.
A
Just
so
neat
there's
so
many
neat
features
there
lots
of
threads.
I
can
pull.
I
will
pause
a
moment
do
see
a
question
here
in
chat
asking
if
the
dashboard
recommends
a
new
premium
like
like
if
a
collateral
grows
up
or
tells
you
that
should
go
up
or
down.
Do
you
guys
have
to
monitor
that
all
and
then
draw
the
conclusion
manually
so
kind
of
asking
I
guess
kind
of
about
automation
there.
So
let's
say
you
know
there
is
a
severe
change
in
in
risk
premium.
C
Yeah,
so
our
approach
is
basically
to
always
have
a
human
in
the
loop,
so
so
that
means
like
we.
We
want
to
move
towards
some
some
kind
of
automated
governance,
but
there
are
always
qualitative
aspects
and
and
things
that
that
just
cannot
be
completely
automated
right
because
of
the
large
governance
service
at
maker.
It's
very
difficult
to
automate
everything
right.
C
A
B
B
But
then
there's
additional
metric
that
we
just
recently
added
is
the
die
capital.
We
call
it,
you
can
even
see
it
here
on
the
page
and
the
text
so
why?
Why
is
this
useful?
So,
even
if
we
know
that
somebody
is,
you
know
actually
protecting
himself.
It's
also
good
to
know
whether
he's
die.
I
simply
call
it
whether
he's
die
long
or
short.
B
So
if
you
think
about
it,
when
somebody
is
borrowing
die,
he's
normally
short
right,
but
because
of
you
know,
all
this
yield
farming
going
on
and
we
know
a
lot
of
our
users
are
basically
yield
farming,
so
they
put
heat
on
in
the
world
and
they
borrow
dye
and
then
the
deposits
die
on
either
ave
or
compound.
You
know
they
make
like
10
percent
yield
or
whatever
it's
it's
good
to
have
this
information
on
each
world,
because
if
you
know
he
has
die
capital
at
hand,
it's
much
easier
for
him
to
react
in.
B
You
know
that
one
hour
or
you
know
basically
at
any
time
he
could
get
liquidated,
it's
good
to
know
that
he
can
react
quickly.
You
know
he
just
withdraws
die
from
you
know
whatever
deal
farming
platform
and
replacing
his
depth.
So
that's
that's
really
useful
and
it
really
helps
us
knowing
you
know
it
even
helps
us
with
other
aspects
of
maker.
B
B
Then
we
can
see
how
you
know
if
we
go
this
high,
we're
going
to
basically
lose
this
user
right,
so
this
is
also
something
so
what
I'm
saying
it's
not
useful
just
for
measuring
risk.
It's
also
useful
for
other
business
and
and
growth,
metrics
and-
and
I
just
recently
made
the
twitch
storm
why
our
approach
to
measuring
risk
is
unique.
A
Yeah,
that
is
really
neat,
and
you
can
see
pretty
directly
how
it
plays
into
things
like
d3m
or
the
the
monthly
rates
changes
right.
You
know
when,
when
you're
actually
able
to
track
and
see
alright.
Well,
where
are
the
pain
points
here?
How
much
room
do
we
have
that's
really
neat?
There's?
Maybe
one
mic
coming
on.
I
don't
know
if
someone
had
a
question,
but
if
you
want
to
take
the
money,
take
go
for
it.
A
C
Yeah
definitely
so
we
have
for
the
vault
protection
score.
We
have
three
different
tiers
kind
of
the
low
risk,
medium
risk
and
high
risk
and
for
the
low
risk
is
when
the
when
a
certain
volt
has
a
collateralization
ratio.
Above
three
times
of
the
of
the
liquidation
ratio
of
the
specific
wall,
type
or
is,
is
subscribed
to
device,
saver
services
or
basically
has
survived
or
basically
was
protecting
itself
during
at
least
three
unique
days
in
the
last
year
that
that
we
chose
as
as
large
drops
in
that
collateral
asset
right.
C
So
that's
basically
the
low
risk,
and
then
we
have
the
medium
risk,
which
means
that
that
it's
a
bit
less
protected
and
it
has
to
do
with
the
the
qualitization
ratio
as
well.
But
now
the
multiplier
is
one
and
a
half
and
not
three
relative
to
the
liquidation
ratio.
That's
the
first
thing
and
then
the
second
thing
is
that
the
median
qualitization
ratio
needs
to
be
again
the
multi
above
the
multiplier
of
1.5.
C
And
then
we
have
another
thing,
which
is:
if,
because
we
saw
with
the
new
volts,
we
introduced
this
another
signal.
Kind
of
that
that
it's
not
a
high
risk
but
a
medium
risk
volt.
If
it's,
because
we
notice
with
new
volts
that
they
become
quite
that
certain
volts
become
quite
active
early,
even
though
they
have
a
low,
a
low
qualitization
ratio.
C
And
that's
why
we
picked
also
I'd
say
like
a
threshold
that
if
they're,
if
they
made
a
lot
of
different
vault
actions
in
in
the
last
month,
then
we
scored
them
as
as
medium
risk.
So
that's
a
voltage
either
low
risk
or
medium
risk,
and
and
if
none
of
these
conditions
are
satisfied,
then
it's
it's.
A
high
risk
goal.
B
C
So
here
the
the
inputs
are
essentially
the
the
jump,
severity
drop,
jump,
frequency
and
and
the
keeper
profit
which
we
can
change,
but
then
the
the
biggest,
the
biggest
thing
that
that
impact
is
are
essentially
the
the
the
the
liquidity
right.
How
much
liquidity
is
there
in
the
cluster
asset
and
then
also
the
the
quantization
ratio?
The
signals
that
that
I
basically
mentioned
before
right?
Basically,
these
two,
these
don't
affect
the
the
risk
premium
directly.
C
A
A
You
know,
based
on
the
the
tools
available
like
with
the
dashboard,
for
instance
like.
Are
there
certain
things
we
should
be
paying
attention
to
when,
when
trying
to
contemplate
like
all
right,
what
should
we
do
with
the
surface
buffer
like
what?
What
values,
I
guess
are
we
wanting
to
look
at
to
kind
of
determine
if
the
surplus
buffer
is
too
high
or
too
low.
B
Yeah
sure
so
actually
should
be
contacted.
Me
actually
contacted
him
to
address
this
topic,
and
this
is
because
I
think
we're
going
soon.
You
know,
hit
the
surplus
buffer,
the
limit,
which
is
now
at
60
million,
and
this
is
because
you
know,
we've
seen
a
lot
of
dimenting.
Potentially
rates
are
also
going
to
be
increased
and
I
think,
like
in
two
weeks
we're
going
to
hit
the
surface
buffer
and
then
you
know
the
maker
burning
starts,
and
you
know
I
this
topic
is
a
bit
controversial.
B
How
do
you
say
so
you
have
people
who
are
pro
pro
burners
or
they
just
want
to
see.
You
know,
make
her
burn
and
of
course
then
there's
the
other
angle,
which
is
you
know
the
the
you
know,
making
sure
that
maker
is
well
capitalized
when
it
comes
to
potential
shortfalls,
and
you
know
this
is
what
we
do
and
like
it's
normal
that
we
push
towards.
You
know
higher
surplus
buffer,
like
this,
just
normal.
B
That
risk
would
would
do
if,
if
it
thinks
that
we're
not
properly
compensating
risks
right
now
and
that's
why
it
was
really
useful
that
we
came
up
with
this
capital
at
risk
metric.
You
know
actually
what
one
or
two
weeks
ago,
because
this
actually
gives
us
some.
You
know
some
guidance
of
what
the
surplus
buffer
should
be.
You
know,
assuming
our
methodology
and
inputs
are
correct,
and
this
this
becomes
quite
complex
because,
as
I
said,
when
we
measure
risk,
you
know,
what's
the
expected
shortfall,
which
is
essentially
capital
at
risk.
B
So
now
it's
at
130.
You
saw
that
you
know
a
bunch
of
capital.
Risk
is
being
contributed
from
it
be,
and
why
is
that?
Because
you
know
there's
a
bunch
of
new
worlds
that
haven't
been
tested
and
we're
not
sure
how
well
are
they
going
to
protect
themselves?
If
you
know
tomorrow,
there
is
a
40
drop
in
eater
price,
which
is
possible,
so
there's
a
bunch
of
inputs
that
we
use
and
it
becomes
problematic
because
some
are
subjective.
You
know
even
saying
you
know
it
can
drop
40
twice
a
year.
B
You
know
like
we
do
test
this
empirically,
and
I
think
this
is
like
you
know
something
that's
possible,
but
you
can
see
again
that,
like
this
is
like
a
tail
scenario
right.
You
can't
really
say
like
it's
gonna
happen
two
times
or
three
times,
you
know,
and
it's
completely
you
get
completely
different.
You
know
risk
premium
then,
but
the
the
point
I
want
to
make
is
that
when
we,
when
we
say
you
know,
the
capital
at
risk
is
130
million
and
we
believe
the
surplus
should
be.
B
B
and
because
inputs
are
in
some
cases,
so
subjective
and
ideally
would
want
to
crowdsource
them.
We
try
to
make
this
reversible.
I
mean
try
to
make
reverse
calculation
when
you
know
when
community
asks
us,
you
know
what
should
be
the
surplus
number,
because
it's
not
like
one
finite
number.
We
say
like
this:
like
you
know,
130
million
basically
means
we
are
protected
if
you
assume
this
kind
of
inputs
and
scenario.
So
at
the
end,
what
we
actually
want
to
ask
community
is,
you
know,
is:
is
your
risk
tolerance?
B
You
know
in
this
range
right,
you
know
if
you're
or
put
it
otherwise
put
it
differently.
If
you're
gonna
have
130
surplus
buffer,
we
are
protected
against
this
kind
of
scenario,
and
you
know,
are
you
you
know
fine
by
accepting
it
or
do
you
think
you
know
it's?
On
average,
it's
going
to
be
not
as
bad,
and
you
know
in
this
case
we
can
tweak
the
inputs
and
we
get
the
lower
surplus
now.
So
this
is
how
you
know
we
would
want
to.
You
know,
explain
things
to
community,
but
essentially
you
know
just
like.
B
If
somebody
asked
me,
you
know
what
should
be
the
surplus
buffer.
I
think
our
our
inputs
are
quite
conservative
but,
as
I
said,
we
don't
account
for
some
unknown
black
swan
risks
right,
such
as
technical
risk,
and
they
do
believe
that,
like
100
million
surplus
is,
is
appropriate,
which
is,
you
know,
almost
twice
as
high
as
now,
and
some
people
won't
like
it.
B
But
you
know
if
you
just
think
at
how
much
100
million
is
in
terms
of
you
know,
all
that
from
risky
collateral,
which
is
now
you
can
see
here
at
4.3
billion
and
130
million
is
what
is
three
percent
of
this
debt
like
this
is
not
like
a
huge
you
know
if
you
look
at
banks
today.
This
is
this
is
not
like
a
huge
ratio
right
and
we're
currently
sitting
at.
B
I
think
one
point:
three
percent
of
of
risky
depth
so
surplus
buffer
currently
is
about
one
point,
three
percent
of
risk
tap
and
that's
why
I
believe
we
should
increase
it,
but
I
also
get
the
you
know
pro
burning
aspect,
and
I
also
think
it's
actually
good
to
burn
some
maker.
Why
is
that?
Because?
It's
also
related
to
you
know
our
basically
our
ability
to
catch
against.
You
know
shortfalls,
and
why
is
that?
B
But
it
has,
you
know,
positive
pr
effects
and
so
on
and
if
you
do
some
burning
and
you
are
able
to
increase
the
mkr
price,
that's
actually
good,
because
if
you
have
a
shortfall-
and
you
cover
some
part
from
the
surplus
buffer
and
then
you
have
dilution
event,
at
least
you
know
you
were
able
to
increase
the
mkr
market
cap
and
then
it's
much
easier
to
do
a
dilution.
You
know
it's
much
more
effective,
so
I
I
do
get
the
the
burning
aspect
as
also
being
useful
yeah
in
that
regard.
A
Definitely
thanks
for
the
the
input
there.
I
remember
this
had
to
been
a
while
ago,
maybe
like
january
or
something
I
was
still
pretty
pretty
green
to
make
her.
But
andy
came
on
the
g
r
call
and
did
presentation
on
value
at
risk,
and
it's
definitely
eye
opening
to
see
like
how
much
things
like
ethp,
you
know,
really
do
contribute
with
the
lower
collateralization
ratio,
and
especially
once
you
add,
in
the
osm,
delay
and
stuff.
B
You
get
into
bed
depth
really
soon,
and
this
is
just
the
you
know
the
negative
side
of
of
this
osm
delay.
I
would
say,
but
it
also
has
many
benefits
right.
Our
oracles
are
more
robust
because
we
can
see
if
you
know
something
goes
wrong
with
oracles
we
can
react
and
the
other
major
benefit
is
just
that.
D
Yeah
sure
I
I
I
thought
that
was
great.
I
was
totally
following
along
with
what
you
were
saying,
but
you
kind
of
gave
it
as
a
as
a
a
kind
of
an
abstract
way.
You
said
you
said
what
we'd
like
to
ask
is:
if
we
are
at
a
certain
surplus
buffer,
we
are
protected
against
this
scenario
and
then
you
went
on
to
say
that
you
would
suggest
100
million.
So
I'm
kind
of
curious.
If
you
put
those
two
together,
if
at
100
million,
we
are
protected
against
what
scenario
yeah.
B
Okay,
so
here
you
can
see
for
130
million
capital
at
risk.
You
can
see
below
the
contribution
of
of
each
world
type
right
right,
so
we
have
capital
at
risk
and
then
jan
a
few
ten
minutes
ago
or
whatever
he
explained
how
this
capitalist
risk
for
each
world
type
is
calculated.
So
young,
can
you
click
on
risk
model
type
please.
B
So
here
you
can
see
so
it
capital
at
risk
is
44
million.
For
the
current
exposure
right
and
on
the
left,
you
can
see
you
can
see
the
inputs,
but
these
are
not
all
the
inputs,
so
the
the
core
inputs
are
jump
severity.
So
here
we
are
testing
and
jump
frequency.
So
here's
the
thing
like
loss
at
maker
usually
comes
from
a
crash
right.
It's
not
like
20
percent
drop,
it's
like
40
50
percent
raw,
and
this
this
has
happened.
For
instance,
this
year
happened
almost
twice,
so
we
we
simulate
something.
B
That's
you
know
something
that
we
think
is
good
enough
or
conservative
enough,
which
is
two
drops
per
year
of
50
percent,
and
but
there
are
a
bunch
of
other
inputs,
so,
for
instance,
we
use
the
current
quartilization
ratio
of
each
volt.
So
what
we
basically
do,
we
apply
this
price
drop
to
each
volt
on
it
a
and
we
use
protection
score,
which
is
now
you
can
see.
B
Volts
protected
chair
here
is
another
input
is
at
87
percent,
and
we
say
87
of
the
depth
of
these
volts
are
going
to
save
themselves,
because
we
can,
you
know
we
can
measure
this
like.
We
historically
obtain
this.
You
know
by
looking
at
these
faults
and
then
the
13
percent
gets
liquidated
right
and
you
get
the
total
liquidation
amount.
It's
not
shown
here,
and
then
you
have
another
simulation
right
when
you
have
particular
share
of
that
liquidated,
you
need
to
look
at.
B
B
B
This
is
very
important
because
all
the
auctions
happen
on
chain,
so
the
liquidity
on
chain,
the
current
liquidity
matters
a
lot
and
we
say:
okay,
we're
liquidating
300
million
of
it.
The
slippage,
according
to
this,
would
be
what
we
have
here
like
six
percent
and
we
apply
this
slippage
to
you
know
to
the
actual
auction
process.
B
On
top
of
it,
we
apply
also
keepers
profit,
because
no
keepers
would
do
this
unless
you
earn
something
right
and
let's
say
the
keeper's
profit
is
five
percent
which
we
have
here
as
as
as
a
standard,
so
six
percent
slippage.
Five
percent
from
keepers
makes
like
eleven
percent
slippage
and
you
apply
this.
B
You
know
on
the
liquidated
amount,
and
here
we
actually
measure
whether
this
pushes
you
know
whether
we
are
able
to
collect
all
the
depth
that
we're
looking
for
or
whether
there's
a
loss
and
if
there's
a
loss,
you
know
we
get
this
risk
premium,
which
is
the
expected
loss
so,
and
we
do
this
for
every
volt
type
and
it's
all
automated.
It
assumes
all
all
these
inputs,
like
liquidity,
slippage,
collateralization
ratio.
This
is
all
automated
most
of
stuff,
is
on
chain
yeah.
Basically,
all
the
data
are
from
on
chain
and
and
yeah.
A
D
D
B
Yeah,
like
I
said
like
we
can
tweak
those
inputs.
You
know
just
by
a
small,
it's
like
making
a
discounted
cash
flow
evaluation
right,
yeah.
Why
I'm
saying
100
is
because
I
already
know
there's
an
opposition
from
people
not
willing
to
increase
the
surplus
buffer.
I
think
100
would
be
a
great.
You
know
compromise
right
now,
but
like
essentially
it's
truly
like
this,
like.
I
won't
lie
like
if
you,
if
you
drop
the
jumpsuit,
very
quite
conservative,
on
this
inputs
like
if
you
drop
the
jump
severity
to
45
percent.
B
We
don't
have
it
here
like
we
have
40
like
everything
changes
you
see
like
capital
at
risk
drops
for
from
44
million
to
seven.
So
we
are
already
aware
that
we
are
conservative
here
and
like
100
surplus
buffer
for
me
would
be
you
know,
good
enough,
I
would
say
for
the
for
knowing
how
we
use
the
inputs
and
everything.
A
B
B
These
premiums
that
you
get
usually
from
the
markets
here
at
maker
we
can
but
like
in
traditional
finance,
you
get
risk
premiums
from
markets
right,
you
see
where
they're
trading
and
you
can
calculate
it
reversely
and
it's
basically
all
relies
on
crowdsourcing
there's
so
many
actors,
thinking
about
how
risky
something
is,
and
you
get
a
good.
You
know
hard
bit
of
risk
here.
We
don't
have
any
kind
of
similar
product
trading,
so
we
could
say
you
know
this
is
the
risk.
B
D
A
B
Sure
so
we
have
our
own
channel
on
discord.
What
it's
called.
I
think
it's
risk
public
right
so
probably
best
place
to
start.
Is
there
we're
also
active
on
forum?
You
can
find
me
primoz
primos.
I
know
we
have
a
good
name
but
yeah.
You
can
find
me
on
forum
or
discord
to
have
the
same
tech
should
reply
or
on
twitter.
A
Right
on
well,
thanks
to
you
and
and
jian
for
the
awesome
presentation
today,
I
appreciate
everyone
who
came
and
participated
as
well,
we'll
keep
the
discussion
going
in
the
forums
and
believe
we'll
probably
have
some
discussion
too,
from
risk
on
the
governance
risk
call
a
little
later
today.
That's
that's
in
a
couple
hours,
if
you
all
happen
to
be
free,
so
thanks
for
coming
I'll,
go
ahead
and
shut
off
the
video
now
and
appreciate
all
the
time
everyone.