馃攳
Bankroll Management for Sports Trading and Betting Using the Kelly Criterion for Stake Sizing - YouTube
Channel: Trademate Sports
[0]
I guess this is marios from trade mate
[2]
today we're going to talk about the
[3]
Kelly criterion and how we use it to
[6]
calculate your recommended bed sizing so
[10]
the formula for Kelly is equal to the
[14]
probability of the outcome of the game
[17]
you're betting on times the odds that
[20]
you're getting minus 1/2 outs that
[23]
you're getting minus 1 so let's let's
[28]
first have a look at what it looks like
[30]
in your trick feed so here you can see
[32]
that on the Norway Romania game there's
[36]
a recommended steak of 292 Kroners and
[39]
what will attempt to do in this video is
[41]
to show you how we come up with this
[43]
number so let's assume that we got on
[49]
the classical scenario of the coin toss
[51]
right now the probability of either
[54]
outcome is 50% so let's let's say that
[58]
you put money on tails but the odds
[61]
you're getting instead of 2.0 which is
[64]
equal to the 50% probability of tails
[66]
you get the outs of 2.1 meaning that you
[70]
have 5 percent or 5 percent edge right
[74]
so how can you then use the Kelly to
[79]
figure out how much money depending on
[82]
your bankroll that you should place on
[84]
this coin toss so what we have is the
[89]
probability how come
[90]
50% the else you're getting 2.1 and then
[94]
the kelly percentage using a formula so
[98]
basically what the Kelly criterion is
[100]
saying is that you should put for point
[104]
55% of your bankroll on this coin toss
[108]
so let's assume you have a bankroll of
[110]
$1000 then you recommend at stake would
[114]
be $45 now the thing to note about the
[118]
Kelly criterion is that it is a formula
[121]
for maximizing your profit growth this
[124]
also means that it is a very high risk
[128]
or high variance play to follow the
[131]
Kelly
[132]
completely so that is why we have given
[135]
you the option to choose between
[136]
following the Kelly criterion 100
[139]
percent 50 percent or 30 percent so
[143]
depending on your tolerance for risk you
[146]
can decide how how much your you're
[149]
willing to follow up Kelly so currently
[152]
I'm using 30 percent myself so that
[156]
means that I get more slower growth but
[160]
also less variance so Martin wrote a
[163]
great article back in the days of edge
[165]
pad about how this will place out
[168]
because one of the things when it comes
[172]
to the trading is that you're likely to
[175]
have multiple bets at the same time so
[179]
next we'll have a look at how this how
[182]
this influences different things so
[187]
let's assume that you have 20 open bats
[190]
on coin flips all the time you get a two
[195]
point one and outs at a 5% edge so you
[198]
should win 50% at the time right so
[200]
that's just steady now let's assume that
[203]
you have 20 open bats it's going to
[206]
randomly pick ten winners and ten losers
[209]
so basically it you're running bang on
[213]
Evie right but also it's a it's worth to
[217]
note that the the order of occurrence
[220]
whether or not you win a window when a
[224]
la win the coin toss or loser coin toss
[226]
doesn't really matter when you're using
[227]
to Kelly so for instance in the example
[230]
of you only using two coin tosses then
[234]
you can end up with same month whether
[236]
you win the first task or you lose the
[239]
second or whether you lose the first and
[242]
win the second okay so back to a better
[247]
example that we were talking about so
[250]
what you can see here on the x-axis the
[253]
distribution of all the different
[254]
possible combinations of picking can we
[258]
understand losers over 20 of the bets so
[260]
that's about 184 thousand different
[263]
combinations
[265]
now what we can see there is that if we
[269]
follow the hundred percent Calais kid
[270]
potentially end up a seventeen percent
[273]
growth from starting bankroll but at the
[277]
same time we could potentially lose 12
[280]
percent to those outcomes are very much
[282]
unlikely but on average you could expect
[286]
to have an EVs of three percent if you
[291]
followed the Kelly criterion for long
[295]
but to do so you're risking 60% of your
[299]
bankroll now compare that to using the
[303]
50 50 percent of Kelly as you can see
[305]
the variance is much smaller but also
[309]
your average decreases a bit from 100%
[311]
example so for the 50 percent Kelly
[316]
option you're risking about 37 percent
[318]
of your bankroll to in the best case
[322]
when about 6 percent and in worst case
[325]
you're down about at 2.6 percent now the
[331]
probability of you making a profit is
[333]
also much higher using a lower risk
[336]
Kelly alright so if we're following 30
[340]
percent of the Gallagher's area what we
[342]
can see here is that of variance is very
[344]
so much smaller but at the same time we
[348]
can expect two or less money so using
[352]
three percent means that you're more
[354]
likely to see a steady growth but
[357]
smaller profits than what you could
[359]
potentially have by following it percent
[362]
so using 30 percent of the Kelly you're
[366]
only risking 24 percent of your bankroll
[369]
to earn an expected of one point two
[374]
percent in the best case you're only
[377]
three percent but in the worst case
[378]
you're only the zero point five percent
[380]
so your probability of making a profit
[384]
is almost equal to one so that's that's
[388]
why I personally prefer to use the 30
[391]
percent of the Kelly now there are three
[394]
main reasons for why we
[397]
recommend using less than a half percent
[400]
Kelly and that is because one time
[405]
before kickoff so if you place your
[407]
trades a long time before kickoff then
[409]
it could both swing out of your favor
[411]
and swing in your favor so the edge
[413]
percentage could go up it could go down
[415]
and remember our goal is to beat the
[418]
closing line of the sharpest bookies
[421]
right mainly pinnacle pinnacles closing
[424]
line so this means that for instance if
[430]
the edge moves a lot out of your favor
[432]
then you've placed a very high
[434]
percentage of your bankroll on an
[437]
outcome when really you should have been
[439]
placing very small amounts so it becomes
[441]
very risky the second part is that Kelly
[445]
and myself as we saw on the graph is
[447]
very volatile so the variance of
[450]
following 100% Kelly is huge or you
[454]
could make you know anywhere from
[455]
negative 12% bankroll growth or positive
[462]
17% bankroll growth and also the third
[467]
thing is that at the South Brook case
[470]
they do place limits on winning players
[473]
so by betting larger amounts it could
[475]
make potentially make them ban you
[478]
sooner what's also important to know
[481]
about to Kelly criterion is that it is
[486]
optimized for placing a series of bets
[490]
it is not optimized for placing bats in
[493]
in parallel so what that means the
[498]
example that we were using we only had
[500]
20 open bets so if you were to have even
[504]
an even higher amount of pets open at
[507]
the same time let's say 50 to 100 on a
[510]
Saturday when there's a lot of action
[511]
going on then your variance is going to
[514]
be even larger than what you can she see
[517]
in this photo so that's that's also
[520]
another reason for why we recommend
[523]
using a lower Kelly percentage so the
[528]
downside of this is that it's so it's
[530]
harder to turn over
[531]
your bankroll with a lower Kelly because
[534]
your true Eevee is calculated by
[537]
multiplying your risk bankroll so
[540]
meaning your turnover at times your
[543]
average edge so if you're using lower
[547]
Kelly you will play slower bets so you
[550]
will have lower turnover and thus
[552]
potentially lower Eevee so what this
[558]
means for you it means that if you're
[560]
going to have a lot of open bets at the
[563]
same time then most likely you can
[566]
you're better off using a lower Kelly
[569]
percentage like for instance the 30 or
[571]
the 50%
[580]
you
Most Recent Videos:
You can go back to the homepage right here: Homepage