►
Description
Learn how to maintain and improve market-making strategies.
*Important Disclaimer**
This video is for educational purposes only. We are not giving any legal, tax, financial, or investment advice. Every user is responsible for their use and configuration of Hummingbot.
A
Okay,
okay,
yeah,
hitting
the
joint
audio
button,
is
probably
a
good
thing
to
do
when
you're
giving
a
live
stream.
Sorry,
everyone
I'll
start
over
thanks,
whoever
told
everyone
that
we
don't
have
audio.
A
So
one
second,
okay,
all
right,
so
so
yeah.
So
as
I
was
saying,
my
name
is
michael,
I'm,
ceo
of
humminbot
and
we're
back
with
another
edition
of
humanbot
live,
which
is
our
weekly
web
show
where
we
basically
do
an
intro
into
some
topic
without
hummingbot,
and
today
I'm
going
to
present
basically
a
bit
of
history
on
how
we've
evolved
the
strategies
having
bought
over
time,
and
we
also
take
questions
of
the
audience.
A
So
if
you
have
any
questions
about
hummingbi,
you
know
crypto
in
general
or
or
really
anything
feel
free
to
post
it
on
the
chat
and
I'll,
try
to
read
it
out
and
answer
it
in
real
time.
So
so,
of
course
the
what
one
purpose
of
these
sessions
is
to
show
off.
You
know
live
running
of
the
bot
itself,
so
I
will
be
doing
some
of
that
during
the
session
as
well.
A
So
I'll
try
to
kind
of
like
start
a
bot
kind
of
like
you
know
as
at
the
very
beginning,
and
then
let
it
run
for
a
while
and
then
we
can
come
back
and
let's
see
how
it
goes.
So
let's
get
started,
I'm
going
to
share
my
screen.
Give
me
one.
A
Okay,
well
so
yeah,
so
we're
gonna
talk
about
the
hummingbike
kind
of
strategies.
I
guess
see.
You
can
still
see
me
needs
to
work.
Yes
looks
like
it
still
works
great.
So
first
I'll
give
a
little
presentation
about
strategies
past
present
and
future
first
as
a
disclaimer.
This
is
only
for
instructional
purposes,
only
just
education.
When
I
run
a
bot,
it
really
is
for
testing
purposes,
not
that's
not
really
giving
any
financial
advice
so
coming
by
now.
A
Today,
we're
at
open
source
trading,
client
and
the
main
purpose
is
to
allow
traders
to
trade
on
different,
centralized
and
decentralized
exchanges.
A
Without
without
having
to
build
connectors
into
all
of
these
different
exchanges,
now
when
we
started
up
when
we
first
started,
we
used
to
run
a
quant
trading
fund
and
back
then
we
would
spend
weeks
like
building
these
low-level
api
integrations
into
an
exchange
before
we
could
even
get
started
with
the
fun
stuff
and
many
other
people.
A
You
know
both
individuals
and
also
trading
firms
have
the
same
problem,
so
we
thought
we'd
better.
Instead
of
building
a
fund
to
launch
this
open
source
project
hummingbike
and
really
to
save
everyone
a
lot
of
time
by
building
connectors
all
these
exchanges,
and
so
today
we
support
over
25
connectors.
A
A
lot
of
them
are
both
centralized
and
decentralized,
and
what
we
enable
is
a
global
community
of
quant
traders
over
the
world
to
basically,
you
know,
run
all
types
of
strategies
on
these
exchanges
and
they
really
focus
on
market
making,
because
liquidity
provision
is
one
of
the.
You
know
the
biggest
kind
of
topics
of
interest
in
crypto
and
market
making
has
always
been.
A
You
know
a
lot
of
like
mystique
around
how
to
do
it,
but
you
know,
in
our
opinion,
it's
just
software
and
you
know
if
it's
software,
then
there
should
be
an
automated
solution
for
it
and
that's
exactly
what
it
is.
So
today,
a
lot
of
people
use
hummingbot
and
over
time
our
strategies
have
really
changed
a
lot.
You
know
we've
always
really
focused
on
market
making
and
arbitrage,
but
in
the
very
beginning
you
know
they
only.
They
only
were
available
on
spot
exchanges.
A
I
mean
only
for
spot,
centralized
exchanges
that
is,
and
they
were
really
basic
in
rudimentary.
You
know,
looking
back,
I'm
always
ashamed
of
how
basic
they
were,
and
today
our
strategies
are
can
run
on
both
spot
and
perpetual
exchanges,
both
centralized
and
decentralized
exchanges
in
crypto
and
overall,
I
think,
there's
a
lot
more
advanced
than
the
ones
in
the
past.
A
So
now
I'll
talk
about,
you
know
what,
where
we
kind
of
came
from
the
past
ones,
what
what
some
of
my
strategies
look
like
today
and
then
finally
I'll
talk
about
kind
of
like
you
know
what
we're
doing
in
the
future.
A
Okay,
so
we
first
started:
there
were
kind
of
three
main
strategies
we
had
at
humminbot
and
they
were,
we
call
them
peer
market,
making,
cross-exchange
market
making
and
then
arbitrage,
and
so
so
let
me
kind
of
show
off
the
the
pure
market
making
strategy.
That
was
really
the
one
that
a
lot
of
people
have
been
using.
You
know,
because
for
market
making,
that's
really
kind
of
like
the
the
in
the
past,
our
flagship
strategy.
A
So
I
have
honeybot
running
in
my
aws
instance
over
here,
so
I'm
gonna
start
it
up
enter
my
password
and
then
I
will
see
the
screen
here.
So,
let's
see,
yep
should
be
big
enough.
So
first
you
know
first
we're
coming
back,
there's
always
a
health
command
that
shows
you.
You
know
what
the
commands
available.
So
on
this
instance,
I
think
I
have
yep.
I
have
a
connector
into
binance,
so
I've
added
that
key
and
looks
like
I
was
also
testing.
A
You
know
the
the
the
key
for
huabi,
so
I
intentionally
put
in
an
invalid
key,
and
so
this
is
what
happens
when
you
put
in
an
invalid
key.
It
says
no
for
keys,
confirmed
and
you'll
see
an
error
message
here.
So
I'll
I'll
show
a
demo
at
test
bottom
binance
to
show
you
how
the
peer
market
making
strategy
works,
so
I'll
use
a
create
command
and
select
the
peer
market
making
strategy
and
then
for
the
exchange.
A
This
is
for
the
exchange,
so
I'll,
select,
finance
and
then
for
the
trading
pair.
So
I'll
pick
I'll
I'll
go
and
uict.
This
is
mainly
because
I
know
there
are
some
algo
tokens
and
uict
tokens
in
this
test
account
and
then
basically
the
peer
mechanics
strategy.
It
really
just
automated
a
set
of
kind
of
fairly
deterministic
and
simple
kind
of
processes
that
would
allow
the
bot
to
kind
of
update
your
orders.
So,
typically,
what
we,
you
know
typically
have
people
input
is
the
spread.
A
So
this
is
a
spread
on
kind
of,
like
both
sides,
of
the
the
order
book,
the
big
spread
and
the
aspirate,
and
then
also
how
many,
how
much
time
long
you
would
take
the
cancel
order,
the
amount
of
I'll
go.
You
want
to
have
the
order
and
then
and
then,
and
then
we
asked
this
question.
We
had
this
ping-pong
feature
that
was
a
very
simple
way
to
kind
of
manage
inventory
risk,
which
is
the
main
risk
faced
by
mark
makers,
which
will
just
allow
your
orders
to
kind
of
bounce
around.
A
And
so,
when
you
fill
a
buy
order,
you
only
show
a
sell
order
and
then,
when
you
fill
a
sell
order,
you
then
only
show
a
buy
order
and
this
pinpoint
feature
really
allowed
people
a
very
simple,
rudimentary
way
to
handle
the
risk
where
you
know.
If
the
market's
going
down
a
lot,
you
end
up
buying
and
then
when
the
market
goes
up,
you
end
up
selling
and
so
that
that
risk
is,
you
know,
mitigated
by
this
feature,
but
in
an
extremely
basic
way.
A
So,
oh
actually,
I
should
turn
on
ping
pong.
That
was
didn't
mean
to
turn
that
off
and
so
yeah,
because
you
can
always
use
a
commit
command
to
kind
of
you
know
edit
things
like
this
on
the
fly,
so
it's
gonna.
A
If
I
just
start
that
bot,
it's
gonna,
really
just
kind
of
you
know,
place
a
couple
orders
and
then
you
know
place
a
couple
orders
for
for
to
buy
and
sell
these
ago
tokens,
and
you
know
it's
just
really
update
them
every
30
seconds
or
faster.
If
you
want,
and
over
time
you
know,
we
have
to
add
a
lot
of
a
lot
of
com
parameters
to
this
strategy
to
help
users
deal
with
inventory
risk
and
other
risks
that
market
management
face.
A
So,
for
instance,
we
added
a
feature
called
hang
orders
which
allows
you
to
kind
of
like
leave
your
orders
on
the
books
and
have
them
out
there.
We
also
added
stuff,
like
we
have
something
called
inventory
sku,
which
allows
you
to
kind
of
target,
a
certain
inventory
percentage,
and
then
the
bot
would
adjust
the
sizes
of
orders
to
help
you
get
there
and
really
a
bunch
of
other
stuff,
and
you
know
to
help
users
manage
the
inventory
risk,
the
main
risks
that
come
with
market
making.
A
But,
to
be
honest,
I
I
think
I
think
that
the
issue
with
a
lot
of
this
stuff
is
that
it's
just
it's
kind
of
like
they're
all
lovers,
but
they're
kind
of
independent
lovers
that
that
are
they
don't
really
have
a
lot
of
like
you
know,
theoretical
kind
of
grounding
and,
and
so
they
do
work,
but
but
but
I
think
I
think
it's
kind
of
like
it's
a
little
bit
like
a
trial
and
error
in
order
to
find
the
right
combination
that
works
for
a
certain
market
condition
and
as
markets
change
as
a
market
maker.
A
You
really
need
to
kind
of
like
fine-tune
this,
so
what
we
found
was
the
users
who
had
the
most
success
with
the
strategy
were
really
the
ones
depending
the
most
amount
of
effort
who
can
watch
the
markets
adjust
to
the
parameters
in
their
bot
and
so
forth,
and
so
because
of
that,
we
we
eventually
added
a
script
of
feature
called
scripts.
That
would
allow
users
to
kind
of,
like
you
know,
customize
the
bot,
even
more,
and
so
so
with
scripts.
You
can.
A
We
actually
have
a
script
that
allows
users
to
kind
of
adjust.
The
spreads
when
the
markers
were
volatile,
and
so
that
was
really
popular
script
that
people
had,
but
you
know
the
issue
scripts.
They
do
still
require
you
to
kind
of,
like
you
know,
run
the
python.
I
think
they
also
didn't
work
with
their
binaries
because
they
required
like
a
different
type
of
like
module,
so
so
yeah
there's
some
issues
there.
So
that's
why?
A
Overall,
I
think
we
we
decided
to
kind
of
really
take
a
step
back
a
few
months
ago
and
and
really
kind
of,
like
you
know,
assess
how
how
we
wanted
to
kind
of
do
our
strategies
again,
because
you
know
with
because
crisis
change
market
making
average
charge
these.
These
two
strategies
are
still
quite
you
quite
useful,
but
because
they're
arbitrage,
they're
kind
of
arbitrage-like
strategies-
users,
you
know
don't
use
them
as
much
and
they're
more
complicated
to
use
and
they're,
not
as
kind
of
simple
to
use
as
market
making.
A
So
what
we
did
come
back,
we
start
we
started
really
last
year
in
the
last
year,
at
schengen,
this
year
was
really
developing
strategies
that
we
felt
would
be
kind
of
like
help
users
market
make
in
a
safer
in
a
more
diversified
way,
and
so
really
we
focused
on
92
in
particular.
So
these
are
the
two
I
focus
on
today,
which
is
liquidity,
mining
and
the
avalanche
strategy.
A
So,
as
you
may
know,
we
run
the
main
way
that
we
make
money
as
as
a
company
is
we
run
a
liquidity
mining
platform
called
hung
about
minor,
where
people
can
take
coming
by
and
run
and
earn
rewards
there.
So
on
that,
one
because
you're
earning
rewards
it
tends
to
offset
the
the
the
cost
of
market
making
and
many
of
our
users
have
actually
you
know,
used
it
to
earn
a
profit,
but
also
because
of
diversification,
a
lot
of
users.
A
They
run
a
multiple
instances
of
heinbot
in
order
to
earn
rewards
in
multiple
markets
and
allow
kind
of,
like
you
know,
kind
of
like
the
the
diversification
to
smooth
out
their
returns
profile
and
achieve
you
know
kind
of
like
a
more
consistent
level
of
profitability.
So
you
know
observing
their
behavior.
We
decided
to
kind
of
kind
of
like
just
create
a
strategy
that
kind
of
automates
this
a
bit
more
and
kind
of
just
gives
you
more
control,
and
so
it
mean
really
the
main
value
is
it.
A
It
allows
you
to
run
multiple
multiple
markets
and
earn
rewards
and
multiple
markets
without
having
to
run
multiple
instances
and
then
and
then,
and
the
other
thing
it
does,
is
the
volatility
kind
of
like
you
know,
script
we
use
have
in
hummingbot.
We
we,
you
know,
we
basically
put
it
into
equity
mining.
So
let
me
let
me
let
me
run
a
recruiting
mining
strategy
right
now
and
then
I'll
pause
and
take
some
questions,
because
we
do
have
some
questions
in
the
comments.
A
One.
Second,
as
I
show
you
guys
how
pretty
mine
works,
so,
let's,
let's
check
in
our
stats
with
the
peer
market.
We
can
buy
so
it's
been
running
for
a
while,
but
because
these
spreads
are
actually
pretty
wide
for
this
market.
You
know
we
we
haven't
had
any
orders
hit
yet
so
I'm
gonna
stop
the
block
and
I'm
gonna
set
up
a
liquidity,
mining
bot.
That's
meant
to
show
you
how
how
that
works.
A
A
So
so
recall
that
before
I
had
I'll
go
unct
as
my
first
as
my
market
and
now
I'm
going
to
add
a
new
one,
so
I'm
going
to
add
avalanche
unct
just
as
another
market
and
and
this
this,
the
second
one
actually
is
one
that
where
we
have
a
reward
pool.
A
So
this
is
a
one
where
miner
has
rewards
so
I'll
put
these
two
markers
in
and
then
I'll
use,
because
usat
is
a
common
token
there
I'm
going
to
use
uit
to
provide
liquidity
for
these
markets,
so
so
the
size
of
basically,
you
know
size
I'll
put
in
in
usat
as
well
so
size
of
it
to
say,
20
and
then
here
this
the
simple
set
option.
A
This
is
the
spread
0.5
percent
and
then
target
inventory
percentage,
and
so
this
is
applied
to
every
market
that
you're
in
and
so
and
then
then.
Finally,
just
the
configuration
name,
so
what
this
does
is.
Essentially,
if
I
start
this
strategy,
it
basically
makes
a
market.
A
It
does
the
same
thing
as
per
market
making,
but
it's
just
doing
so
on
two
markets
instead
of
one
and
so
and-
and
we
also
we're
also
kind
of
like
you
know,
in
addition
for
markets
where
it
is
awarded
by
a
minor,
if
you
run
the
status
command,
it
shows
you
the
market,
the
reward
per
week
and
some
of
the
some
other
stats
to
help
you
can
decide
which
markets
you
want
to
participate
in
and
and
so
forth.
A
So
this
one
and
really
the
advantage
of
this
one,
I
think,
is
that
it
has
that
volatility
adjustment
feature
from
before
and
you
can
set
kind
of
a
multiple
list
of
markets.
In
fact,
I'm
going
to
stop
this
bot
and
I'm
going
to
I'm
going
to
kind
of
like
config
the
the
markets
to
to
even
more
and
kind
of
like
see
how
many
I
can
get
in
here.
A
Yeah,
I
know
these
are
kind
of
like
markets
where
we
have
some
campaign
running
so
that
way
I'll
be
able
to
kind
of
earn
rewards
in
all
these.
A
A
Yep,
so
so
now
now
so
now
that
I
have
essentially
a
market
making
bot
writing
five
different
markets
and
then
basically
what
the
box
has
done
is
kind
of
it's
taken,
my
total
of
amount
of
inventory
in
uict
any
of
these
other
five
tokens
and
then
really
just
kind
of
divide
them
up.
A
So
there's
an
equal,
roughly
equal
amount
of
you
know
of
of
value
in
each
one,
and
it's
trying
to
essentially
make
sure
that
the
the
the
balance
is
like
around
it's
targeting
50
percent
inventory
rate
or
50
allocation
across
in
each
market.
A
A
Here's
the
list,
improvements,
I'd
like
to
see,
expose
all
pmm
parameters,
scripting,
more
events,
especially
partial,
fills
and
user
update
parameters
and
then
and
then
and
finally,
number
three
interface
to
expose
user
settable
parameters
to
scripts
in
the
hangout
ui,
yeah,
sean
actually
yeah.
These
are
really
really
good
suggestions
and
actually,
but
let
me
let
me
first
kind
of
explain
kind
of
like
our
the
history
of
scripts
and
kind
of
what
we're
planning
to
do
with
them
going
forward.
A
So
so
scripts
were
something
we
introduced
middle
of
last
year
in
response
to
kind
of
like
the
demand
for
users
to
kind
of
customize
their
homework
strategies.
So
and
so
initially,
we
we
kind
of
thought
of
it
as
an
add-on
to
a
strategy,
and
so
that's
why
the
currently
it's
built
as
a
as
a
different
sub
process
and
because
it's
a
different
sub
process,
that's
kind
of
what
that's.
Why?
A
Even
if
you
have
a
bug
in
your
script,
you
know
it
doesn't
block
the
kind
of
operation
in
hummingbird
you
know,
and
so,
but
because
it's
a
different
sub
process.
We,
like
we,
have
to
kind
of
manually,
expose
certain
variables
to
that
sub
process,
and
so
like
everything
we
can't
we
could
expose
more,
but
every
single
one
we
expose
just
requires
more.
You
know,
engineering
time
to
push
them
all
there.
A
To
be
honest,
I
I
think
that
what
what
I
think
people
really
want
to
do-
and
I
think
this
is
what
you
want
to
do
as
well
sean,
it's
really
customizable
strategies
and
I
think
what
we've
realized
is.
I
think
the
scripts
was
like
kind
of
like
you
know.
A
It's
served
I
need,
but,
but
I
think,
there's
actually
a
better
way
to
serve
it
holistically,
which
is
to
enable
easier
strategy
creation
and
to
kind
of
allow
strategies
to
just
basically
to
basically,
you
know
people
to
create
all
types
of
to
customize
these
ones
more
easily
because
like
because,
because
really
hyperlinks
is
python
and
a
lot
of
the
code,
even
for
the
on
the
strategy
level,
it's
actually
pretty
simple
and
overall,
I
think
I
think
we
can
actually
create
more
examples
of
easier
strategies
and
kind
of
reduce
all
the
bullet
plate
and
strategy
and
just
allow
users
the
full
access
to
full
power
on
bond.
A
So
it's
that's.
That's
a
bigger
change,
that's
kind
of
why
we're
not
doing
much
with
scripts
right
now,
because,
ultimately,
we
feel
like
it's
more
limited
as
a
vehicle
for
users
to
customize
strategies,
and
you
know,
I
think
a
better
long-term
solution
is
to
just
make
strategy,
customization
and
creation
a
lot
easier
and
to
you
know
basically
start
doing
that
so
overall
we're
planning
to
do
that
starting
this
summer.
You
know,
if
not
earlier
and
and
and
and
kind
of
like
do
that,
so
so
this
should.
A
Overall,
I
would
say
that
the
trend
with
hummingbird
is
for
the
strategy
that
we
create
ourselves.
I
think
they're
going
to
be
kind
of
almost
more
complex
in
terms
of
like
they're
they're,
doing
avalanada,
which
I'll
show
you
after
this
and
then
they're
kind
of
based
on
academic
papers
and
then
on
the
other
side.
I
think
we're
gonna
expose
more
kind
of
customization
and
creation
tools
for
users,
so
they
can
kind
of
like
go
and
create
their
own
strategies
and
kind
of
using
the
templates
we
create.
A
So
so
that's
why,
overall,
what
what
I
think
we
want
to
try
to
do
is
kind
of,
like
you
know,
for
the
strategy
we
create
just
focus
on
really,
like
you
know,
spend
time
on
the
research
part,
whereas
for
the
engineering
aspect
basis
open
source,
it
really
should
be
really
easy
for
users
to
basically
do
whatever
they
want,
but
I
think
right
now,
the
code
base
just
needs
some
refactoring
in
order
for
us
to
enable
that
without
kind
of
like
you
know,
without
would
that
make
it
almost
impossible
for
users
to
change
it.
A
So
that's
actually
a
good
segue.
Let
me
if
you
have
more
questions,
please
post
them
in
the
chat
and
then
in
the
meantime.
Let
me
let
me
demo
the
other
new
strategy.
I
just
mentioned
the
avogdano
strategy,
because
that
one
is
really
kind
of
like
the
that's
really
the
other
direction
we're
going
to,
because
you
know,
with
liquidity
mining.
A
It
was
like
taking
the
peer
marketing
strategy,
making
it
simpler
for
users
to
do
that
across
the
queen
mining
and
then
avanada
is,
is
really
about
kind
of
making
the
core
like
foundational
math,
underlying
the
strategy.
More
rigorous
because,
basically
there's
a
whole
host
of
academic
literature
in
the
market,
making
industry
there's
been
a
lot
of
kind
of
you
know.
Innovation
academically
in
that
area
in
the
last
20
years
and
and
the
alveoli
stoichov
paper
was
really
what
really
kicked
it
off
it
published
about
12
years
ago.
A
It
really
defined
a
model
for
how
a
market
maker
kind
of
should
think
about
inventory,
risk
versus
spread
and
so
and
we
actually
just
publish
a
blog
post.
That
explains
how
it
works
and
then
we've
kind
of
basically
created
strategy.
A
That
kind
of
replicates
that
paper,
but
also
kind
of
takes
some
of
the
parameters
and
makes
it
easier
for
users
to
use.
A
So
this
is
about
the
easy
mode
where
you
can
kind
of
like
let
it
do
its
thing
for
you
or
there's
an
advanced
mode
where
it
basically
replicates
exactly
what
how
the
strip,
how
the
paper
works,
but
then
you
have
to
kind
of,
like
you
know,
kind
of
inject
your
own,
you
know
like
lower
level
like
variable
parameters,
account
gap
on
cam,
gamba,
kappa
and
gamma,
which,
which
might
be
you
know
harder
for
new
users
to
use,
but
does
give
professional
users
a
lot
more
control.
A
So
so
here's
how
it
works.
Let
me
let
me
show
that
works,
and
meanwhile
we
actually
did
have
a
couple
fills
here.
You
know
for
for
these
two
tokens,
so
so
yeah
so
because
these
two
markets
are
ones
where
there
are
there
are
liquidity
rewards
available.
You
know
I
am
earning
rewards
as
well
by
by
running
the
spot
in
these
markets.
A
So
let's
stop
this
that
bot
and
then,
let's
kind
of
like
take
a
look
at
the
the
new
avalanche
strategy.
So
the
other
strategy
is
really
in
our
in
our
mind,
intended
to
kind
of
eventually
replace
the
peer
marketing
strategy,
as
kind
of
like
a
the
entry
point
for
new
market
makers.
A
So
and
the
reason
is
that,
because
it's
it's,
you
know,
we
realized
that
with
pre-market
making,
adding
more
and
more
deterministic
parameters,
even
though
it
gave
people
more
levers,
it
also
made
it
more
confusing
so
there's
kind
of
like
a
decreasing
level
of
returns,
you're
getting
to
get
by
doing
so,
and
so
with
albanada.
A
We
we
felt
it
would
be
better
to
kind
of
almost
kind
of
like
really
build
a
better
foundation,
and
so,
instead
of
kind
of
deterministically
placing
spreads
and
changing
order
amounts,
you
know,
although
not
it
kind
of
uses
your
inputs
to
to
basically
kind
of
like
set
your
spreads.
You
know
more
intelligently,
so
so,
let's
let
me
run
it
and
now
I'll
kind
of
go
back
and
I'll
kind
of
explain
how
how
it
works
so
similar
to
premarket
making.
A
First
you
select
the
exchange,
and
then
you
select
the
trading
pair
so
like
before.
I
will
take
I'll
go
uict
as
a
trading
pair.
You
know
and
then
and
then
use
that,
and
so
this
time
I'm
gonna,
let
it
let
the
bot
automatically
automate
the
kind
of
the
the
lower
level
kappa
and
gamma
parameters
based
on
the
surprise
I
select
so
then
I
I
just
need
to
kind
of
enter
kind
of
like
my
minimum
spread.
This
is
the
lowest.
A
I
want
the
bot
to
set
the
spread
and
then
I
can
also
set
the
max.
What
is
the
maximum
spread
that
I
want
the
bot
to
set
so
you
know
and
then
and
then
and
and
so
and
then
finally,
I'm
going
to
set
two
variables
that
allow
the
bots
to
kind
of
control
how
sensitive
it
is
to
volatility
and
then
inventory
so
the
first
one
is
a
volatility
threshold
multiplier.
So
essentially
you
know
basically
yeah
so
I'll
just
set
it
to
like
one
one
point,
one
actually
one
yeah.
A
So
it
has
to
be
a
little
greater
than
one
point
zero.
So
I'll
pick
one
point
one
there
in
the
final
and
then
finally
I'll
pick
inventory
version
so
I'll
pick
something
like
1.5
over
here.
A
So
and
then,
and
then
finally
I'll
set
the
inventory
percentage.
So
so
let
me
explain
what
this
is
doing
before
I
start
the
bond.
Basically,
I
set
the
minimum
spread
in
the
maximum
spread.
A
I've
set
my
inventory
target,
which
is
50,
and
then
I've
set
two
variables
that
help
the
bot
kind
of
more
intelligently,
use
my
inventory
risk
and
into
diffuse
inventory
risk,
and
so
what
it's
doing
is
basically
it's
going
to
try
to
keep
the
inventory
level
at
around
50,
so
half
I'll
go
and
have
to
zt
and
then,
when
it
deviates,
it
will
basically
calculate
something
called
a
reservation
price
which
is
kind
of
like
the
price
that
you
know
that
takes
in
account
where
you
want
to
get
to
from
an
inventory
perspective.
A
So
so,
and
then
it'll
place
a
spread
around
the
reservation
price,
not
the
actual
market
price
and
then,
in
addition,
it
will
use
volatility
to
adjust
your
spreads
based
on
this
threshold
multiplier,
but
there's
a
bunch
of
math
behind
the
scenes.
That
is
that
it
kind
of
allows
everything
to
work
in
sync
with
one
another,
and
so
what
we've
done
is
we've
basically
kind
of
you
know
kind
of
derived,
some
of
that
into
easier
kind
of
like
parameters.
A
They're
they're
basically
easier
for
users
to
set,
and
so
that's
the
volatility
threshold
multiplier
and
the
inventory
risk
parameter.
You
know
kappa
and
gamma
respectively,
so
we
should
actually
kind
of
maybe
change
these
directions,
but
kind
of
make
that
a
little
more
obvious.
But
you
know
that's
something:
we'll
probably
try
to
do
in
the
next
one
or
two
releases.
A
So
let
me
start
at
the
bottom.
So
the
first
thing
you'll
notice,
is
that
it's
gonna
take
a
little.
It's
gonna,
it's
not
going
to
stop
the
bot
right
away
and-
and
the
reason
is
because,
let's
see,
the
reason
is
because
it
has
to
calculate
the
volatility,
because
it's
going
to
it's
going
to
be
it's
volatility.
It
volatilities
inputs
into
the
spread
into
what
kind
of
spreads
you
want
to
set.
The
bot
first
has
to
accumulate
enough
samples
and
volatility
in
order
to
kind
of
calculate
volatility
before
setting
the
market
setting
starting,
but.
A
Okay,
so
as
we're
waiting,
let's
take
a
few
more
questions
from
the
audience.
Saucy
asks
any
plans
to
add
an
ios
app
for
a
hummingbit,
also
any
plans
to
add
more
exchanges,
I'm
the
u.s
and
I'm
a
huge
disadvantage
since
it
won't.
Let
me
use
regular
violence,
anymore
yeah.
I
totally
sympathize.
You
know
I
yeah,
it's
being
the
us
myself,
we're
also
very
limited
in
terms
of
what
we
can
access
and
you
know
like,
I
think
we
kind
of
like
we
have
like
test
accounts
set
up.
A
You
know
like
from
our
overseas
offices
in
order
to
do
something
like
this
so
yeah.
So
the
first
thing,
I
would
say,
is
basically
how
many
buy
the
hummingbird
out
of
the
package
actually
supports
a
lot
of
exchanges,
including
a
lot
of
us
ones
like
buying
side
us,
bittrex
and
others.
I
think
you
might
be
referring
to
our
liquidity
mining
platform,
the
miner
platform,
the
minor
platform
right
now
at
sportsfinance.com,
which
is
not
available.
A
You
guys
people
we
just
added
kucoin
recently,
so
ku
klux
markets
are
available
to
people
in
the
us
and
we're
adding
bitmaps
or,
as
they
relabel
themselves,
ascendx
soon.
Ascendx,
I
think,
is
also
available
to
people
in
the
us.
Overall
I
would
say,
for
both
hummingbike
is
also
a
minor.
We're
gonna,
be
constantly
adding
more
exchanges
on
the
honeywell
side,
we're
probably
going
to
be
adding
more
decentralized
exchanges
for
time
being,
but
for
minor
we'll
be
at.
A
I
think
we
plan
on
being
six
exchanges
by
him
this
year,
which
concludes
you
know,
comes
which
should
include
some
of
the
larger
ones
like
you
know
that
will
be
adding.
You
know
the
end
of
the
year.
A
So
and
so
now
coming
back
to
this,
the
bot.
So
although
nada
is
now
running
and
so
now,
if
we
take
status,
you
can
see
that
there
are
some
strategy,
specific
parameters,
that
avanade
shows,
in
particular
the
risk
factor,
the
order
book
depth
factor.
This
is
basically
gamma
and
kappa,
and
then
finally,
what
the
observed
volatility
is
as
well
as
the
time
to
the
end
of
the
trading
cycle.
A
So
note
that
avenatti
has
a
kind
of
has
a
trading
cycle
kind
of
kind
of
like
a
concept
and
and
and
it's
basically
to
it's-
it's
necessary
in
in
there
and
and
kind
of
how
the
strategy
works.
So
it's
it's
mainly
because
the
model
itself
is
kind
of
based
on
traditional
stock
markets,
which
they
do
have
like
a
end
of
a
trading
session.
A
So
that's
why
we
basically
in
replicating
the
strategy,
as
is
we
basically
kind
of
like
made
it
as
is,
but
in
the
future
we
might
create
a
new
version
and
and
kind
of
like
try
to
kind
of
make
that
time
frame
infinite,
but
there's
some
kind
of
like
additional
math
view
there.
So
for
now
I
think
you
know
we're
still
kind
of
modeling
the
traditional
paper.
A
Let's
see
okay,
so
so
I
think
it
should
of
at
least
a
feel
for
how
the
avenue
strategy
works.
It's
overall,
I
would
say
this.
This
should
be
just
better
than
pure
market
making.
Overall,
you
know
so
and
over
time
you
know
as
more
people
use
this
strategy
and
as
it
gets
more
baked,
we'll
probably
be
adding
more
of
the
like
order.
Levels
and
some
of
the
other
kind
of
like
convenience
type
features
from
pure
market
into
this
one.
A
And
so
so,
finally,
where
are
we
going
next?
We're
talking
about
strategies?
You
know,
as
I
mentioned,
one
of
the
first
things
we're
going
to
do
is
incorporate
oblanada
into
more
of
the
market,
making
strategies
which
we
include
it's
kind
of
like
the
liquidity
mining,
as
well
as
probably
the
perpetual
marketing
strategy.
A
In
addition,
we're
also
going
to
add
the
kind
along
that
same
vein,
because
there
are
many
papers
around
market
making
we
plan
on
kind
of
incorporating
more
of
them,
so
there's
a
number
of
paper.
I
think
this
oliver
gwent
is
another
author.
That's
written
a
bunch
of
great
papers
on
market
making,
and
is
it's
really
a
whole
host?
Rather
so
so,
ultimately,
I
think
for
whether
it's
us-
and
we
also
want
to
encourage
you-
know
kind
of
like
clients
and
data
scientist
community
who
are
interested
and
kind
of
like
modeling.
A
These
academic
papers
to
to
you
know,
raise
your
hand
on
our
disc
floor
and
let
us
know
we're
happy
to
work
with
you
guys
on
on.
Maybe,
like
you
know,
like
you
know,
on
incorporating
external
distributors
into
these
strategies
as
well.
A
In
addition,
we
we
we've
made
some
inroads
into
building
connectors
that
support
perpetual
exchanges,
so
we
support
binance
features.
You
know
we
also
support
dydx
as
well
as
perpetual
protocol.
So
I
think
now
that
we
have
a
few
perpetual
connectors.
I
think
we
can
really
kind
of
sort
of
standardize
those
connectors,
because
you
know
for
users
who
are
arbitraging
funding
rates
or
doing
spot
perpetual
arbitrage
between
them.
A
I
think
there
are
a
number
of
differences
between
these
different
exchanges,
and
so
I
think
I
think
right
now,
we've
done
some.
You
know
right
now
we
kind
of
added
the
connectors,
but
I
think
we
kind
of
really
looked
at
the
strategy
a
bit
more
to
trying
to
make
them
more
snare.
A
And,
finally,
you
know
we've
done.
We've
done
some
work
in
terms
of
adding
unity
swap
balancer
and
other
decentralized
exchange
connectors,
but
I
think
there's
a
lot
more.
We
can
do
there
and
so
currently,
when
the
process
of
kind
of
you
know
refactoring
that
gateway
component
to
make
it
you
know
better
for
for
developers
use
and
for
even
for
our
only
term
developers
use
but
heading
to
you
know.
A
You
know
over
the
summer
I
think
we'll
be
adding
a
lot
more
decentralized
exchange,
connectors
there,
oh
and,
as
I
mentioned
finally,
you
know
enabling
easier
strategy.
Creation.
Customization
is
definitely
one
area
that
that
we're
going
to
do
a
lot
more
of
this
year.
A
You
know
last
year,
I
think
our
whole
team
was
really
focused
on
building
the
minor
platform
and
kind
of
like
just
kind
of
like
adding
the
first
few
connectors
to
decentralize
exchanges,
but
now
this
year
I
think
another
amount
of
platform
is
really
up
and
running
now
you
know
we
can
really
turn
attention
back
coming
by
and
and
kind
of,
like
you
know,
allow
the
user
base
community
to
really
you
know,
involve
and
really
involve
them,
a
lot
more
in
the
development
and
the
upkeep
of
this
project.
A
So
I
hope
that
was
helpful.
This
is
just
like
a
you
know.
I
just
wanted
to
just
just
jump
on
this
live
stream,
and
you
really
just
you
know,
offer
more
information
into
the
community,
because
I
know
that
you
know
in
the
past
few
months
it's
been.
We
really
did
a
heads
down
trying
to
accommodate
how
much
growth
the
minor
platform
has
had.
A
So
it's
been,
you
know
I
I
just
haven't
had
the
time
to
kind
of
do
things
like
this,
but
hopefully
I
think
going
to
the
summer
as
we're
enabling
more
kind
of
strategy
creation
we'll
also
be
announcing
a
hackathon.
You
know
soon,
so
hopefully
that
will
allow
us
to
kind
of
you
know,
be
more
informative
and
and
kind
of
you
know,
just
be
better
stewards
of
the
community
than
we've
been
in
the
past.
A
A
Oh,
so
actually
so
sean
has
one
more
comment.
One
thing
I
like
to
see
is
balanced
limits
persist
between
runs
or
visually
display,
the
change
to
balance
from
the
initial
limits.
Yeah,
that's
a
that's
a
that's
a
great
comment.
I
I
do
think
that
yeah
we
introduced
this
bounce
limit
feature
in
order
to
kind
of
allow
users
to
like
have
more
control
over
what
they're
doing.
But
I
do
agree
that
yeah
having
it
not
persist
between
bronze
is
is
is
is
doesn't
seem
like
a
problem.
A
So
I
think,
if
you
have
any
suggestions,
I
would
I
would
yeah
you
should
put
them
in
a
get
up
issue
or
you
know
talk
to
one
of
our
support
team
members
because
overall,
I
think
I
think
we
added
that
feature,
because
I
think
certain
users
requested
the
ability
to
kind
of,
like
kind
of
you
know,
have
more
control,
but
I
think
I
don't
think
we've
really
kind
of
you
know
put
a
lot
of
thought
into
how
to
kind
of
make
that
feature
more
useful,
useful
and
usable
for
everyone.
A
So
if
you
have
any
suggestions,
please
let
us
know
and
we're
always
looking
for
ways
to
kind
of
just
make
our
lab
make.
The
lives
of
you
know,
make
the
trading
easier
for
our
users.
A
Well,
all
right!
Well,
I
hope
that
that
was.
It
was
helpful
for
everyone.
You
know
it's,
I
know
we're
all
pretty
busy,
so
I'm
gonna
get
on
with
my
day
and
hope
you
have
a
good
day
as
well,
happy
trading
and
hope
to
see
you
at
the
next
version
of
coming
about
live,
bye.