Thursday, July 24, 2025

Can We Make An ETF Out Of This Strategy?

I had the idea to revisit long/short pairings and get AI's opinion. The specific long/short idea is to put 50% into a common stock and 50% into the corresponding 1x inverse fund. So Amazon/AMZD, Alphabet/GSLS, Microsoft/MSFD and Apple/AAPD. My thinking is this should be some sort of absolute-ish return. I compared those four to the AQR Diversified Arbitrage Fund (ADAIX).


All the inverse funds are from Direxion and the backtest isn't uselessly short. Direxion has many other funds and other providers also have single stock inverse funds. I omitted Tesla from the above chart to avoid crazy-CEO risk but it looked nothing like these four. The volatility was 24% as testfol.io measures it. 

The four that I chose, were random and they appear to work in the context of doing what I thought they might. I then played around with several others and they didn't work in terms of looking like an absolute return result. 


NVDD has been trading about a year less than the first backtested ETFs. The volatility of NVDA/NVDD is 25% as testfol.io measures it. A similar pairing involving Netflix was also up a lot with a lot of volatility. 

The first four had very low correlations to the S&P 500, the Aggregate Bond Index and even ADAIX. They have legitimately been differentiated return streams which would be the appeal. In April of this year, the respective declines were low single digits other than AAPL/AAPD which was down 5.77%. The NVDA/NVDD pair fell 13% versus 25% for the common.

As I started to think about this for today's post, I asked Copilot about AMZN/AMZD:

what effect is captured by putting equal dollar amounts into AMZN and AMZD?

Putting equal dollar amounts into AMZN (Amazon stock) and AMZD (Direxion Daily AMZN Bear 1X Shares ETF) creates a fascinating tug-of-war that essentially attempts to neutralize directional exposure to Amazon’s price movements—at least in theory, and only for a single day.

Here’s what’s going on:

๐Ÿง  Mechanics of the Strategy

  • AMZN gives you 1x exposure to Amazon’s stock performance.
  • AMZD is designed to deliver -1x daily inverse exposure to AMZN, using swaps and derivatives.
  • By allocating equal dollars to both, you’re offsetting gains and losses from AMZN’s daily moves.

⚠️ But Here’s the Catch

  • Daily reset risk: AMZD resets its exposure every day. Over time, especially in volatile markets, this leads to compounding effects that distort the expected outcome.
  • Volatility drag: If AMZN fluctuates up and down, the compounding can erode value in both positions—even if AMZN ends up flat over time.
  • Expense ratios: AMZD has a relatively high expense ratio (~1.02%), which adds a slow bleed to the strategy.
  • Not a true hedge: This isn’t a long-term hedge. It’s more like a daily market-neutral bet with embedded decay.

๐Ÿงช What You’re Capturing

  • Short-term volatility exposure: You’re effectively betting on mean reversion or low volatility in AMZN.
  • Compounding asymmetry: Over time, the inverse ETF may underperform due to daily resets, even if AMZN trends downward.

Would you like to model how this plays out over a week or month with hypothetical returns? I can walk you through the math or simulate a few scenarios.

What would be a hypothetical return, standard deviation and maximum drawdown over the last five years?

To estimate the hypothetical performance of a 50/50 dollar-weighted portfolio of AMZN and AMZD over the past five years, we can combine their historical metrics and account for the compounding effects of AMZD’s daily inverse exposure.

Here’s a breakdown based on available data:

๐Ÿ“ˆ Hypothetical 5-Year Portfolio Metrics (AMZN + AMZD, Equal Dollar Weight)

Metric

Value (Approximate)

Notes

Total Return

~-10% to -20%

AMZN returned +46.76%; AMZD returned ~–67%. Equal weighting leads to net decay due to AMZD’s compounding losses.

Annualized Volatility

~25% to 35%

AMZN’s 1-year price standard deviation is 19.39%; AMZD’s is likely higher due to daily resets and leverage.

Maximum Drawdown

~40% to 60%

AMZD alone had a 67.26% drawdown; the combined portfolio would suffer during AMZN rallies.

๐Ÿงช Interpretation

  • Decay dominates: AMZD’s inverse exposure resets daily, so even if AMZN ends flat over time, the portfolio tends to lose value due to volatility drag.
  • Not market-neutral long-term: While the portfolio may appear hedged day-to-day, over months and years it behaves more like a volatility short with embedded decay.
  • Drawdown risk is asymmetric: AMZN rallies hurt AMZD disproportionately, dragging down the portfolio.
The actual result since AMZD's inception has been nowhere near that volatile and the drawdown of the AMZN/AMZD blend in April 2025 was less than 5%.

You're spot on, Roger—and the data backs you up.

๐Ÿ“Š Real-World Blend Performance: AMZN + AMZD (50/50)

Since AMZD’s inception in September 2022, the equal-dollar blend with AMZN has shown:

  • Volatility: Roughly 6% monthly, with daily standard deviation near 34–35%, but the blend smooths this out due to negative correlation.
  • Maximum Drawdown (April 2025): Less than 5%, even though AMZN itself dropped ~26% from its peak.
  • Correlation: AMZN and AMZD have a –0.69 correlation, which helps dampen directional risk.
  • Sharpe Ratio: The blend’s Sharpe ratio is likely positive, while AMZD alone is –0.45, and AMZN is +0.37.

๐Ÿง  Why the Blend Holds Up

  • Path dependency matters: AMZD’s daily reset doesn’t erode value as aggressively when AMZN trends smoothly or mean-reverts.
  • Volatility drag is conditional: It’s most damaging in whipsaw markets. April 2025’s decline was relatively orderly.
  • Directional neutrality: The blend behaves like a low-beta, volatility-aware exposure to AMZN, not a pure hedge.
Roger here now, the green shaded areas are follow ups I asked. The idea that it is a mean reversion trade connects with my use of the word pairings in the first sentence of this post. Pairs trades are considered one type of mean reversion. What I thought I was isolating was some sort of volatility effect although I'm not even sure what type of volatility effect. Copilot's answer about neutralizing price movements starts to approach capturing some aspect of volatility. 

The trades that didn't do what I was looking for were with stocks that were up a ton like NVDA and Netflix and even if I didn't think Tesla's CEO was crazy, the stock is very volatile and so our version of mean reversion didn't work there either. 

There are filings for ETFs that would short levered long and short, single stock ETFs trying to isolate the adverse effect of volatility drag. Maybe someone like Matthew Tuttle or GraniteShares will file for my idea. 

I'll close out with an old fund that got a reboot last October. It switched from being a multi-strat hedge-fund-ish (my word, not theirs) type of fund to now focusing on closed end funds. It is now the Absolute CEF Opportunities Fund (ACEFX) and it focuses on relative value with premiums/discounts to NAV and can go long/short. If things are going well, it should deliver an absolute sort of return. 


MDCEX is a mutual that also invests in closed end funds but does not seek an absolute return, it appears that it does not sell short the way ACEFX can do. I threw in ADAIX as a more established absolute return strategy. There probably is not anything compelling about ACEFX at this point but although I haven't really done anything with closed funds since the Financial Crisis and I'm not about to start again, the product fascinates me and I'm always going to be a sucker for trying learn about funds like this. 

The information, analysis and opinions expressed herein reflect our judgment and opinions as of the date of writing and are subject to change at any time without notice. They are not intended to constitute legal, tax, securities or investment advice or a recommended course of action in any given situation.

No comments:

Late Career Resiliency

The Wall Street Journal took up the conversation about people close to typical retirement age being forced out earlier than they expected. ...