Description

What if the key to unlocking consistent profits in algorithmic trading lies in the short-term momentum of bonds? Join us in this compelling episode of "Papers With Backtest," where we delve deep into the groundbreaking research paper titled "One Month Momentum in Bonds," authored by Adam Zaremba, Huigang Long, and Andreas Karthenasopoulos. This episode is a must-listen for algorithmic trading enthusiasts eager to expand their understanding of market behavior across various asset classes.

Our hosts dissect the intriguing concept of short-term momentum, contrasting it with the widely recognized reversal phenomenon typically observed in individual stocks. The findings presented in the paper reveal a surprising trend: winners in asset classes, particularly bonds, tend to maintain their winning streak in the short term, defying the expected reversal behavior seen in stocks. This revelation opens up a new dimension for algorithmic trading strategies, challenging conventional wisdom and inviting traders to rethink their approaches.

Spanning over two centuries of data across multiple asset classes—including equities, government bonds, T-bills, commodities, and currencies—this research offers a comprehensive analysis that sheds light on the mechanics of short-term momentum. Our hosts break down the trading strategies employed within the research, revealing a significant momentum in commodities and currencies, while government bonds exhibited no such momentum. This distinction is crucial for traders looking to refine their algorithmic trading strategies.

As we explore the implications of these findings for algorithmic trading, listeners will gain valuable insights into how short-term momentum can inform investment decisions across diverse asset classes. Whether you’re a seasoned trader or just starting your journey, this episode promises to equip you with the knowledge and tools to enhance your trading strategies. Tune in to discover how the principles of momentum can be leveraged to optimize your algorithmic trading approach and achieve better results in today’s dynamic financial markets.

Don't miss out on this opportunity to deepen your understanding of the intersection between academic research and practical trading strategies. Join us on "Papers With Backtest" as we navigate the complexities of algorithmic trading and uncover the secrets to harnessing short-term momentum in bonds and beyond!

Hosted by Ausha. See ausha.co/privacy-policy for more information.

Transcription

• Speaker #0

  Hello, welcome back to Papers with Bechtest podcast. Today we dive into another algo trading research paper.

• Speaker #1

  Yes, and this one's quite interesting. It's titled One Month Momentum in Bonds.

• Speaker #0

  Right, by Adam Zaremba, Huigang Long, and Andreas Karthenasopoulos. Looks like they're from Poznan University, Zijian University, and the University of Dubai.

• Speaker #1

  And it mentions funding from the National Science Center of Poland. We're looking at the version from February

• Speaker #0

  22nd, 2019. OK, so what's the... the big question they're tackling here.

• Speaker #1

  Well, you know how in individual stocks, there's this well-known idea of short-term reversal, where last month's winners tend to lose next month and vice versa.

• Speaker #0

  Yeah, that's pretty standard stuff.

• Speaker #1

  They're basically asking, does that same reversal pattern happen when you look beyond individual stocks at broader asset classes?

• Speaker #0

  Oh, okay. And I guess the title Momentum and Mons kind of gives away the answer or maybe complicates it.

• Speaker #1

  It definitely complicates it because what they found was actually the opposite. surprisingly. Not reversal, but short-term momentum across several major asset classes.

• Speaker #0

  Momentum. So winners keep winning in the short term. That is contrary to the stock level evidence.

• Speaker #1

  Exactly. The assets that did best last month tended to keep doing well, at least for the next month.

• Speaker #0

  And they tested this across quite a range of things, didn't they?

• Speaker #1

  Yeah. The scope is huge. They went back over two centuries, like 1800 to 2018. Wow.

• Speaker #0

  1800.

• Speaker #1

  Yep. And across five big asset classes. equity indices, government bonds, treasury bills, commodities, and currencies.

• Speaker #0

  And lots of markets within each.

• Speaker #1

  Tons. We're talking 45 equity markets, 54 bond markets, 52 T-bill markets, 48 commodities, 62 currencies. It's incredibly broad, both historically and globally.

• Speaker #0

  Okay, that's some serious data. So our mission for this deep dive is to really get into the weeds on this short-term momentum. What are the actual trading rules and how did the back... tests stack up across these different areas.

• Speaker #1

  Precisely. Let's start with the core trading rule. It's actually fairly simple. Each month for every asset class, they just sorted all the assets based on how they performed the previous month. They call it SMOM, short-term momentum.

• Speaker #0

  SMOM. Got it. Previous month's return.

• Speaker #1

  Then they formed portfolios. Equal weighted quintile portfolios.

• Speaker #0

  Quintile. So dividing them into five groups, top 20%, bottom 20%, and so on. Equal weighted means each asset gets the same slice of the pie within its quintile.

• Speaker #1

  Exactly. And the main strategy they tested was long-short. Buy the top 20% performers from last month, short the bottom 20%.

• Speaker #0

  Right. Standard way to try and isolate that momentum effect.

• Speaker #1

  Okay. Let's get to the results. How did this play out? Maybe start with equities since that's where the reversal idea usually comes from.

• Speaker #0

  Okay. Equities. Interestingly, even there, their long-short SMM strategy showed a positive average return, 1.32%

• Speaker #1

  per month. 1.32%. That's quite high. And was it statistically significant?

• Speaker #0

  Very. t-statistic of 5.82. So even using these broad global indices over this long history, they found short-term momentum, not reversal.

• Speaker #1

  Huh. That's already a counterintuitive finding. Okay. What about bonds? The paper's title mentions bonds specifically.

• Speaker #0

  Yeah, you'd think. But actually, for government bonds, they found basically nothing. No evidence of short-term momentum.

• Speaker #1

  Nothing. Like, flat. Pretty much. Mean return was negative 0.01% per month. t-stat of negative 0.25. So statistically indistinguishable from zero. No momentum signal there. OK.

• Speaker #0

  So a clear difference between equities and bonds. What about T-bills? Very short-term debt.

• Speaker #1

  Now, T-bills were different again. They found a positive effect, 0.68% per month average return.

• Speaker #0

  And significant.

• Speaker #1

  Highly significant. TSAT of 8.61. So definite short-term momentum in T-bills.

• Speaker #0

  Interesting. Equities and T-bills show it. Bonds don't. What's next? Commodities.

• Speaker #1

  Ah, commodities. This is where it got really strong. The effect was most pronounced here. How strong? Average return on the longshore portfolio, 2.37% per month.

• Speaker #0

  Wow. Over 2% a month.

• Speaker #1

  Yeah. And the T statistic was massive, 13.81. So very strong, very reliable short-term momentum in commodities according to their back test.

• Speaker #0

  Okay. Commodities are the standout winner so far. What about the last one, currencies?

• Speaker #1

  Currencies also showed significant short-term momentum, average return of 1.03% per month.

• Speaker #0

  Also significant, I assume.

• Speaker #1

  Yep. T stat of 9.4. So another strong signal.

• Speaker #0

  Right. So. Let's recap. Momentum and equities, T-bills, commodities, currencies, no momentum in government bonds. That's quite a specific pattern.

• Speaker #1

  It really is. And they didn't just look at them in isolation. They also combined them.

• Speaker #0

  Okay. How did that work?

• Speaker #1

  First, they created a combo strategy by just taking an equal weight of the long-short returns from each of the five asset class strategies.

• Speaker #0

  Averaging the results,

• Speaker #1

  basically. Pretty much. And that combination yielded 0.84% per month. with a very high T-stat of 13.18. It shows the effect as kind of broad-based, not just driven by one oddball asset class.

• Speaker #0

  Makes sense. Did they also try just lumping everything together, all assets, one big sort?

• Speaker #1

  They did. The all-assets strategy pooled everything sorted by last month's return, top quintile long, bottom quintile short. They gave 1.3% per month with an even higher T-stat, 15.40. Again, suggesting this is a pretty pervasive effect across the whole universe they studied.

• Speaker #0

  Those are impressive numbers. Did they check if it was just . . . the extreme winners and losers driving this or was it a smoother pattern?

• Speaker #1

  Good question. They used that monotonic relationship test, the MR test. Ah,

• Speaker #0

  Patton and Timmermans test.

• Speaker #1

  Exactly. To see if returns increased steadily from the bottom quintile to the top. And they did find significant monotonicity for T-bills, commodities, currencies, and those multi-asset strategies.

• Speaker #0

  So it's not just the extremes, it's more of a gradual trend across the rankings. That adds confidence.

• Speaker #1

  Definitely. it strengthens the argument that it's a real consistent effect in those classes.

• Speaker #0

  Now, the big question for any factor, is this just, you know, something else in disguise? Is it just regular 12-month momentum or value or beta or something else we already know about?

• Speaker #1

  They looked into that quite thoroughly, and largely the answer seems to be no. This short-term one-month momentum appears distinct.

• Speaker #0

  So it's not explained by standard momentum? Nope.

• Speaker #1

  Or value, beta, idiosyncratic vol, skewness. Even seasonality didn't explain it, except maybe a bit in currencies. For the most part, SMOM added something new.

• Speaker #0

  OK, so it seems like a potentially unique factor. What about commonality? Does this short-term momentum tend to happen at the same time across the different asset classes where it exists?

• Speaker #1

  They checked correlations between the strategy returns. They found weak but still statistically significant positive correlations between the strategies and equities, bonds, bills, and currencies.

• Speaker #0

  Weak but positive. Suggesting maybe some underlying common driver.

• Speaker #1

  Possibly, yeah. But interestingly, commodities were the exception again. The commodity momentum strategy didn't seem very correlated with the others.

• Speaker #0

  Commodities marching to their own beat again. Okay, what about over time? You said the data goes back to 1800. Was this effect always there?

• Speaker #1

  Not really, no. The superior analysis was quite revealing. Before about 1880, the effect wasn't really apparent.

• Speaker #0

  Ah, so it's more of a modern phenomenon, relatively speaking.

• Speaker #1

  It seems that way. Post-1880, it became much stronger and more consistent across most classes. They suggested maybe fewer assets or smaller return differences back in the early 1800s might explain the lack of signal then.

• Speaker #0

  That makes sense. Less dispersion, harder to pick winners and losers reliably. Did they look at market conditions like high versus low volatility periods?

• Speaker #1

  They did. Generally, the effect was robust, but it did tend to be stronger when there was higher return dispersion among assets within a class.

• Speaker #0

  Which fits with the idea that You need some spread in performance for the momentum signal to really show up clearly. And did they try tweaking the strategy, different ways to implement it?

• Speaker #1

  Yep. They looked at alternatives like time series momentum, just going long assets with positive pass returns, shorting those with negative ones.

• Speaker #0

  Instead of ranking them against each other.

• Speaker #1

  Right. And also volatility adjusted momentum. The results were broadly consistent, suggesting the core finding wasn't just down to the specific quintile sorting method. OK,

• Speaker #0

  pretty robust then. And finally, how sure are we about the data itself? Could it be specific to that data set?

• Speaker #1

  Always a fair question with historical data. Yeah. They did try to replicate using other sources, DataStream, Bloomberg for commodities, even some exotic currencies from GFD.

• Speaker #0

  And did it hold up?

• Speaker #1

  Largely, yes. They confirmed a short-term momentum in equities, commodities, and currencies in these other data sets. Bonds still looked weak, consistent with their main finding.

• Speaker #0

  Right. OK, this has been really interesting. It sort of flips the script on short term effects, moving from reversal in single stocks to momentum in asset classes.

• Speaker #1

  It really does. The paper presents, I think, quite compelling evidence for this one month momentum across diverse asset classes. The trading rules seem straightforward and the backtested results, especially in commodities, currencies, T-bills and even global equities, look strong.

• Speaker #0

  So the key takeaway seems to be that. Contrary to single stocks following last month's winners might actually be a viable strategy when looking at broader asset classes.

• Speaker #1

  That's certainly what the evidence in this paper suggests. It definitely gives you something to think about regarding short term dynamics beyond just the stock market.

• Speaker #0

  Thank you for tuning in to Papers with Backtest podcast. We hope today's episode gave you useful insights. Join us next time as we break down more research. And for more papers and backtests, find us at https.paperswithbacktest.com. Happy trading.