🔍
Introduction to Statistical Arbitrage | Quantra Courses | MCX certified course - YouTube
Channel: Quantra
[2]
In this video,
[3]
we will understand
[4]
what statistical arbitrage means,
[6]
and the steps involved in implementing
[8]
statistical arbitrage strategies.
[12]
Statistical Arbitrage (or Stat Arb)
[16]
is based on the statistical
[17]
mispricing of one or more
[19]
assets compared to the expected
[21]
value of these assets.
[23]
In simpler words, if the
[24]
quantitative analysis using current and
[27]
historical market data suggests that
[29]
prices are off from expected
[31]
value, then it provides an
[33]
“arbitrage” opportunity.
[35]
This will be detailed out
[36]
further in the video.
[40]
The birth of Stat Arb
[41]
can be dated back to
[42]
the mid-1980s.
[44]
A small group of researchers
[46]
working at Morgan Stanley created
[48]
a strategy to buy and
[49]
sell stocks in a pair.
[51]
This strategy quickly earned a
[52]
reputation and given the name
[54]
“Pairs Trading”.
[56]
By the early 1990s,
[58]
analysis techniques became more sophisticated and
[61]
models deployed were more tech-savvy,
[63]
this is when the term
[64]
“statistical arbitrage” was first used.
[68]
Stat Arb suggests to take positions
[70]
against the general norms of the market,
[72]
short the overperforming asset,
[74]
go long the underperforming one.
[78]
We will now discuss the
[79]
basic steps in implementing a
[81]
Stat Arb Strategy on a universe of stocks.
[85]
We start with a universal set of Stocks.
[88]
Consider all the stocks being
[90]
traded on the exchange.
[93]
We filter down to a subset.
[95]
Filter this universal set of
[97]
stocks into a subset by
[98]
carefully matching them by region
[100]
and sector.
[101]
Co-integrated stocks are identified to
[104]
take advantage of the temporary
[105]
widening of spread.
[108]
We will discuss co-integration in
[109]
detail in the next section.
[112]
Then rank the subset.
[114]
Each stock in the subset
[115]
is then given a rank
[117]
generally using the mean reversion principle.
[120]
According to the Mean reversion
[121]
principle, it is expected that
[123]
individual assets or their linear
[126]
combination tend to move around
[128]
the mean, hence it is
[129]
expected that underperforming stocks will
[132]
rise and over performing ones
[134]
will fall.
[135]
Although the scoring formula vary
[137]
depending on which financial parameters
[139]
are considered for ranking the stocks.
[142]
Generally underperforming stocks will
[144]
receive higher scores and outperforming
[147]
stocks receive lower scores.
[150]
The higher scored stocks are
[151]
to be held long whereas
[153]
lower scored stocks are to be shorted.
[157]
Take positions
[158]
Long the top X percent and
[160]
short the bottom X percent
[162]
of the ranked subset.
[164]
Based on the risk taking
[165]
capacity of the system, the
[167]
numbers X and Y will
[168]
be decided.
[170]
We continue to rebalance the portfolio.
[173]
Rebalancing involves periodically
[175]
buying and selling assets in
[178]
a portfolio to maintain an
[179]
original desired level of allocation.
[182]
Rebalancing can have a large
[184]
impact on the portfolio in
[186]
terms of trading costs and
[187]
taxation.
[188]
The optimal frequency for rebalancing
[191]
depends on the assets in
[193]
the portfolio and must be
[194]
carefully evaluated to minimize risks.
[198]
To recap, the primary steps
[200]
involved in a Stat Arb
[202]
strategy are
[204]
First, start with a universal set
[207]
of stocks.
[208]
Second, filter down to a subset
[211]
Third, rank the stocks within the subset.
[215]
Fourth, take positions by going long
[218]
on higher ranked and shorting
[220]
the lower ranked stocks.
[223]
Fifth, rebalance the portfolio as required
[227]
That’s all to show you in this video.
[231]
In the upcoming unit,
[233]
you will learn different types of
[235]
statistical arbitrage strategies.
Most Recent Videos:
You can go back to the homepage right here: Homepage