Risk and Reward: A Quant Tragedy

Through careful research, you have assembled a collection of alphas that are correlated with future asset returns.

There's some conventional stuff (momentum, ST reversal, valuation, quality, short interest etc)...

1/7
There's some totally unique stuff (you think)...

And there's some stuff you're still clinging onto because you don't want to admit you wasted all that time and money on alternative data.

You take these alphas and you combine them into an expected return for each asset.

2/7
You run some simple sense checks. Each period you sort the assets by expected return, long the top and short the bottom.

It looks good. You are encouraged.

But you wouldn't trade it like that.

Some assets are more volatile than others, many are driven by similar risks

3/7
You want to minimize your exposure to most risk sources.

And you don't want the same basic-bitch return drivers like everyone else. Please. You could have ditched your research team and cribbed the Moskowitz papers for that.

You want that sweet unique uncorrelated alpha.

4/7
So you identify risk factors, either economically or statistically.

You create a covariance matrix of these risk factors using EWMA probably.

You fit a model to explain each asset's returns in terms of these risk factors.

5/7
Now optimizer go brrr...

You maximize expected risk-adjusted returns whilst keeping your exposure to those risks and conventional factors close to 0.

You look at the results.

6/7
Your portfolio is market neutral.

It is neutral to all your identified risk factors.

It is also, you note, perfectly alpha neutral too.

7/7 Fin.

• • •

Missing some Tweet in this thread? You can try to force a refresh
 

Keep Current with Robot James

Robot James Profile picture

Stay in touch and get notified when new unrolls are available from this author!

Read all threads

This Thread may be Removed Anytime!

PDF

Twitter may remove this content at anytime! Save it as PDF for later use!

Try unrolling a thread yourself!

how to unroll video
  1. Follow @ThreadReaderApp to mention us!

  2. From a Twitter thread mention us with a keyword "unroll"
@threadreaderapp unroll

Practice here first or read more on our help page!

More from @therobotjames

21 Mar
Most beginner traders don't realize just how variable the p&l of even a very high-performing trading strategy is.

I simulated 10 5 year GBM processes with annual return 20% / annual vol 10%.

(Simulating a strategy within known Sharpe 2 characteristics.) Image
I plotted the path with the highest ending equity (green), median (black) and lowest (red).

All paths are from exactly the same process, with the same known return distribution.

You might think of the green line as trading a strategy with a known large edge and being lucky.
You might think of the red line as trading a strategy with a known large edge and being unlucky.

Even when you were really lucky, you were underwater for 130 days.

When you were unlucky, you were underwater for 508 days (about 2 years)
Read 8 tweets
21 Mar
@InBraised If you can trade for free, then your optimal trading strategy (given reasonable return estimates) would be incredibly hyperactive.

You would continuously change portfolio weights according to your latest return estimates.
@InBraised In the real world, this would kill you, because trading frictions would eat away at your PnL.

So, one way to avoid hyperactive rebalancing is to only calculate your return estimates periodically (say once every day, or every week, or something).
@InBraised But this isn't optimal because, if your alpha is good, you want to be calculating it as often as possible. You just only want to be trading when the increase in expected returns from the new position is much better than the old position.
Read 4 tweets
19 Mar
Examples of "elevator pitches" for retail-friendly trades, that I would find reasonable 👇👇👇

1/4
"Wealth management equity/bond rebalance flows are massive and, due to their size, may not be fully dispersed when performance differences (and therefore rebalance trades) are very large.

We might get paid for buying what they're selling around month-end"
"Institutional yield enhancement programs are massive and tend to be info-insensitive sellers of volatility on an up-tick in vol.

This may keep IVs depressed in the short-term, leading to trend effects in IV on significant bad news, which we could profitably trend-follow"
Read 4 tweets
18 Mar
How do I know if I have an edge?

A thread... 👇👇👇👇

I've been helping a family friend with his trading. I've given him a simple systematic strategy to trade by hand.

We can plot the distribution of historic trade returns from past trading or a backtest as a histogram.

1/n
The trade P&L is on the x-axis and the frequency (# of trades with that P&L) on the y-axis.

This is useful because it gives us a hint as to what the "edge" of our strategy might be - if we could ever truly *know* such a thing.

2/n
In this case, our strategy had positive mean and negative skew.

We saw winning trades about 58% of the time but losers were bigger, on average, than winners.

(As many things that make money tend do, regrettably)

3/n
Read 15 tweets
9 Mar
On "stationarity"...

When we talk about something being "stationary" we mean that the observations look like they could be drawn from the same "bag of observations" (distribution), regardless of what time we choose to look at.

1/5
You can observe this by eyeballing charts.

VIX (blue) stays within a range the whole time. If it gets to extreme values, it's likely to revert back to moderate values.

By contrast, the SPX price (orange) just seems to drift away. It doesn't appear anchored to any range.

2/5
We can also see this by sampling from the distributions at different times.

Let's divide our sample roughly in half (2004-2012 vs 2013-2021).

From the histogram, it's clear that SPX prices are not drawn from the same distribution in the first and second periods

3/5
Read 6 tweets
9 Mar
Let me try to be helpful.

If you have some kind of factor that you think predicts future stock returns (or similar) and you are making charts like below, then here are some tips...

We'll go through an example of trying to "time" SPX with the level of VIX.

A thread 👇👇👇👇
You get daily SPX index prices and daily VIX close data

You align them by date and plot them on dual axes, in true RealVision style.

"SPX tends to go down when VIX is high. I can therefore time an SPX allocation based on VIX. Let me share this on twitter" you say.

No. Very no.
There are two main problems with what you did:

1. The SPX price drifts. We can't directly compare the price of SPX in 2004 with its price in 2021

2. As traders, we are more interested in whether high VIX is *followed by* decreasing SPX prices, not *coincident with* them.
Read 23 tweets

Did Thread Reader help you today?

Support us! We are indie developers!


This site is made by just two indie developers on a laptop doing marketing, support and development! Read more about the story.

Become a Premium Member ($3/month or $30/year) and get exclusive features!

Become Premium

Too expensive? Make a small donation by buying us coffee ($5) or help with server cost ($10)

Donate via Paypal Become our Patreon

Thank you for your support!

Follow Us on Twitter!