Asset Selection / MPT Advanced
The MPT (Modern Portfolio Theory) is a fantastic method of selecting assets in a
rational manner.
However, there is a problem:
The flaw is in the way we measure risks (We used the standard deviation of
returns)
o The standard deviation punishes volatility in both directions.
However, we don’t want to punish volatility upwards, we want to
reward it.
Improving the MPT:
The Sharpe Ratio punishes upward variability as much as downward variability ,
which we don’t want.
We only want to punish variabilty in downward direction.
This would be the ideal version of this risk measurement:
The Sortino Ratio
It takes a measurement of your reward / expected returns relative to your risk.
But the risk is measured as semi-deviation measurement of only your downwards
variability.
Sharpe Ratio punishes both upward and downward variation smooth curve, when
you maximize it.
Sortino Ratio only punishes the downward variation very sharp moves (price or
equity, whatever you’re measuring)
Semi-deviation:
It only gives you a one-dimensional measurement of risk. (Distance from its mean to
the first negative standard deviation)
A measurement to consider the overall nature of risk, not only the one-dimensional
measurement of semi-deviation.
You may have a semi-distribution of 1, but the kurtosis and skewness might be
completely different. This will change the nature of the risk. Wouldn’t it be better if
we had a measure that took into account ALL the behaviour of the risk?
What about a ratio that considered all the behavioural information about all the
risks and all rewards? It would be the ultimate ratio; no data would be wasted.
Because of the unusual skewness neither the semi-deviation or the standard
deviation would give us a good concept of the risk and reward relationship of this
distribution of returns.
In this example, neither standard deviation nor semi-deviation would give us useful
information.
The Omega Ratio
It gives us the optimal way of looking at this distribution by dividing it up into two
area densities.
\__> Probability density of negative returns and the probability density of positive
returns.
It’s “converting” them into a surface area compares the 2 surfaces and gives a
Ratio
\__> Overall nature of reward to the overall nature of risk.
Now, the shape doesn’t matter anymore: Now it doesn’t matter what the
distribution is shaped like! Every possible data point is now accounted for! This is
good because crypto often behaves extremely weird.
Because no other ratio can possibly contain more information on the nature of risk
and reward, it is known as ‘omega’ because it is the ultimate.
You will often find the behavior quite similar to the sortino ratio:
The Omega Ratio is the ultimate form of risk and reward measurement.
What can go wrong when using the Omega ratio
> You may encounter results with this method that don’t make sense, or are
counterintuitive.
> Results from the omega ratio might end up being biased due to extreme gains
(10,000%+) and the limit of an asset to ‘only’ go down -100%.
> Infinite upside skew.
> Never forget that much like all data, its backwards looking and tells you nothing
about the future, just about the historical behavior.
> It might be worthwhile considering the results in the context of some qualitative
analysis to temper extreme results.
Adapt this to the MPT:
The Ultimate Portfolio Theory (UPT) The best possible asset selection
model in the universe.
Problem:
Sometimes it will select assets that only have massive upsides because the
ratio was overpowered by the upsides.
E.x. Shitcoins that go up 1000% but never really went down, it still doesn’t
mean that it’s the perfect asset.
Always look at the results critically, think about if they’re realistic or not.
In summary:
MPT uses the Sharpe Ratio (overall returns / overall standard deviation of returns)
Punishes volatility in both directions improve by using the
Sortino Ratio (expected returns / semi-deviation) only the negative proportion.
The Omega Ration improves on all of this. Always analyse the outcomes
Asset Selection Agenda (MPT Advanced)
> If MPT is good, then how can we improve it?
> Failure of Standard Deviation in favor of Semi-Deviation (“PMPT”)
> Failure of Semi-D (does not take into account the ‘nature’ of the risk)
> Improve using a probability density function
> “Ultimate portfolio theory”
> How I go about selecting assets (screencap of TV & Spreadsheet)
Problem?
> We already know MPT is useful for selecting the optimal asset. But it has a major
flaw.
> The sharpe ratio is a wonderful measurement for giving context to risk and
reward, but there is a problem in the way it measures risk.
> Risk according to the standard deviation of returns actually measures volatility in
both directions by definition.
Summary
> MPT uses the sharpe ratio, but it punishes both upwards variation and downwards
variation
> The sortino ratio improves on this and only punishes negative deviations
> The omega ratio improves on this again and forms a theoretically perfect
probability density ratio of all wins and all losses, and because of this, it is the best
ratio for asset selection.
> Conclusions stemming from the omega ratio might end up being biased due to
extremely skewed gains.