Alright, so, I wanted to run some simulations to demonstrate the trade-off between short-term volatility and long-term certainty.

To start, I simulated out completely random* yield curve changes.

Below is an example evolution.

https://twitter.com/choffstein/status/1923061735805587858

Next, I used these random yield curves to price out the return of two portfolios.

First was a portfolio where I just keep rolling over 6-month Treasuries every single month.

Below is the spread of annualized returns over time.

(I'll address the negative drift in a moment.... but ignore it; it's not the point.)

The team at ReSolve is out with some new research on multi-asset carry when applied to Liquid versus Extended trading universes. In September 2021, ReSolve started trading a multi-asset carry strategy with a target volatility of 20%.

The universe includes ~70 markets (bonds, currencies, energies, metals, equities, grains, livestock, and softs).

Over the live period, they've realized a Sharpe of ~0.5

In celebration of another managed futures replication fund being launched today, here's a little insight into a problem that both top-down (returns-based) and bottom-up (process-based) replication face. Both of these approaches use the historical returns of an index.

The return of that index is driven by two factors:

(1) The underlying component weights
(2) The underlying component returns

Replication is trying to figure out (1), but (2) can be a significant problem.

“Levered portfolios are subject to variance drain!"

So are your long-only active stock funds.

Don’t believe me? Read on. Any long-only active stock strategy can be thought of as a combination of two parts:

- 100% exposure to the benchmark
- 100% exposure to a long/short portfolio capturing the over- and under-weight positions of the manager

(It's long/short portfolios all the way down...🐢)

Most folks are accustomed to evaluating investment products based upon their performance.

In the case of capital efficient strategies, I think this misses a huge opportunity. Consider a capital efficient strategy that gives you $1 of exposure to stocks and $1 of exposure to bonds for every $1 invested.

i.e. a 2x levered 50/50.

I would expect most people to look at a long-term graph and pontificate about the trade offs of leverage and diversification

One question I receive all the time is,

If Treasury futures don’t pay coupons, aren’t they just the price return of the bonds?

Let’s debunk it with a quick picture and then explain. The white line is Ultra 10-year US Treasury Note Futures.

The yellow line is the S&P U.S. Treasury Bond Current 10-Year Index (Total Return) divided by 1-3 Month T-Bills (Total Return)

Staying on trend
Designing an algorithmic trading strategy

Highlighting some new research from SocGen Diversification matters.

“Consider a hypothetical experiment. We construct a series of model portfolios, each containing between 2 and 100 randomly selected trend-following strategies, drawn from a pool of 120 such strategies. For simplicity, we assign the same notional value to each strategy within a portfolio. We then calculate the Sharpe ratio of each portfolio."

cc @investingidiocy

1/ People keep sending me this piece from AQR and asking for my thoughts regarding alternatives, cash rates, and return stacking.

At the risk of inviting the ire of Cliff… I disagree (a little) with the piece.

*ducks*

aqr.com/Insights/Resea… 2/ Actually, that’s probably not fair.

I disagree with the overwhelming interpretation I see people making, which is: higher cash returns imply lower forward asset class returns.

My company, Newfound Research, turned 15 today.

Coming up on this anniversary, I reflected quite a bit on my career.  I’m not sure why, but this milestone feels larger than I would've expected.

So I decided to write something.

15 Ideas, Frameworks, and Lessons from 15 Years This is going to be an obnoxiously long thread, with obnoxiously long posts, so if you'd rather read it as a blog or listen to an audio version, here are the links:

📝

🎙️ Spotify: blog.thinknewfound.com/2023/08/15-ide…
open.spotify.com/episode/0u2tBK…

TIL that the sum of the squared z-scores is always equal to the number of z-score values. Also, the shape of the distribution of z-scores will be the same as the original shape of the distribution of the underlying data.

This one seems more obvious, since you’re just shifting the scaling. But never thought about it before.

1/ A month-or-so ago, @GestaltU penned a piece about replicating the Societe Generale Trend index.

I spent the better part of last week doing a deep dive into the recent performance of these approaches.

investresolve.com/peering-around… 2/ Adam proposes two primary means of replication:

Top Down: Regress recent returns of the SG Trend Index on futures markets to identify the weights that best explain returns.

Bottom Up: Combine an ensemble of trend models selected and weighted to match the SG Trend Index.

pretty explosive recent episode of @TopTradersLive with @andrewdbeer1 and @AuspiceTim…

nobody asked for my opinion, but i’m going to share it anyway.

toptradersunplugged.com/podcast/246-sy… first, kudos to both of these gentlemen for pioneering access to CTAs in an ETF wrapper.

i liked niels's summary: everyone on the podcast can agree that investors should allocate more to the space.

1/ A "quick" summary of my Democratize Quant discussant role for @RA_Insights's paper "Reimagining Index Funds"

(The paper actually seems to have been taken down from Research Affiliates's website... but here's the tweet, I guess

https://twitter.com/RA_Insights/status/1646894358023802880

) 2/ The key contributions of this paper are:

👣 Economic footprint is a more stable measure than market capitalization

📈 Economic footprint avoids "buy high, sell low" behavior

🌊 Economic footprint should be proportional to market cap, retaining liquidity and capacity.

I think one of the best exercises for a young quant is to try to replicate QIS research / indices.

You can learn a ton.

Practically: gather/clean/store data, write non-trivial code, and interpret research.

But more importantly... You get to see how strategies perform out of sample.

Or how they perform with slightly different inputs.

Or how the different steps of the strategy impact the output in non-linear ways.

You develop more intuition for what will have an outsized impact.

What's worse than a twitter thread? A thread of threads.

Here's my 🧵² on Return Stacking™️

(i.e. threads you may have missed) 1/ What is Return Stacking™️?

https://twitter.com/choffstein/status/1437772693764550662?lang=en

On our way...

thinknewfound.com/structural-alp…

https://twitter.com/choffstein/status/1607721358187716610

The models are now available at Altruist, Amplify, Morningstar Model Marketplace, OneDigital, and SmartX.

Huge thank you to our partners at @SimplifyAsstMgt for their continued support in this endeavor.

this was a “fun” real life example of “limits to arbitrage” for me.

the limits I encountered:

1. taxes
2. liquidity
3. laziness

https://twitter.com/0xtuba/status/1639964219847110656

Assuming I timed things perfectly, a repeg would’ve realized a ~14% return.

With a bunch of ETH, I had two choices:

1. Sell ETH to buy USDC
2. Borrow against my ETH

1/ New paper from ReSolve explores methods for replicating trend-following managed futures, with simulated results and performance statistics. 📊
 
investresolve.com/peering-around… 2/ 💹 Trend-following managed futures is a hedge-fund strategy that involves trading futures contracts to identify and capitalize on trends in financial markets, producing returns with low correlation to traditional portfolios.

1/ 🥞 Return Stacking and Variance Decay 📉

Or, "Wait, I thought levered ETFs were bad?" 2/ First, what is "Variance Decay" (or "Volatility Decay")?

I'm certain you've heard this somewhere before...

To get back to break even:
- If you lose 10%, you need to make 11%
- If you lose 20%, you need to make 25%
- If you lose 50%, you need to make 100%

for those who were just listening to 🏴‍☠️PoF…

looks like the NFT waterfall we were discussing already started.

the bored ape floor plummeted from 70 to to a bunch of sales at 55 as @machibigbrother liquidated a bunch of his inventory.

https://twitter.com/punk9059/status/1629233786046271488

@machibigbrother waiting for @osf_rekt to announce he repurchased a bunch that he sold.

If you wanted to replicate Warren Buffett’s portfolio, there’s two ways you might do it.

1. Look at his returns and try to identify the stocks, and their weights, that best explain it.

2. Look at his returns and try to find the investment strategy that best explains it. In the first approach (which I call “top down”), you’re agnostic to how Buffett picks stocks.

But you’ve got some potential data limits. First, how constant is his portfolio? If not very stable, you can’t use much history of returns.

Two, you’ve got a *lot* of stocks.