Hello, welcome back to Papers with Backtest podcast. Today we dive into another Algo trading research paper. Hi there. Yeah, today we're looking at a really interesting one called Alpha Momentum in Country and Industry Equity Indexes. It's by Zaremba, Umutlu, and Karathanisopoulos. Okay, Alpha Momentum. So unpack that for us. What's the central question they're tackling here?
Well, they're essentially asking if a country or an industry has performed really well in the past, And even after you adjust for risk, that's the alpha part. Does that predict it'll keep doing well or does it mean it's maybe due for a fall? Right. Past performance, but specifically risk adjusted performance predicting the future. And they used a pretty big data set for this, I gather. Oh, absolutely massive. They looked at equity indexes from, get this, 51 stock markets.
And the data goes from 1973 all the way to 2018. Wow. Okay. And on top of that, 887 industry indexes. So yeah, really comprehensive. That's some serious data crunching. So what were the headline findings? What should we be paying attention to from all that analysis? They found two main patterns, really. First, in the short term, there's this thing they call alpha momentum.
Basically, if a market has shown strong risk adjusted returns, you know, strong alpha recently, it tends to keep outperforming for a bit longer, kind of like a hot streak, but measured properly for risk. Okay, that makes a certain intuitive sense. Momentum, but for Alpha, what was the second pattern? The second one is kind of the opposite over the longer term. They call it alpha reversal. Reversal.
Okay. Yeah. So countries or industries that have had really high alpha over a longer period, say several years, well, they actually tend to underperform later on. It's like they get overextended and then snap back. Huh. Short-term continuation, long-term mean reversion based on alpha. And the paper suggests you could actually trade based on this, build strategies. Exactly. That's the practical angle.
The idea is you could potentially use these patterns for, say, international equity allocation, knowing which markets might be poised to rise or fall based on their past alpha. OK, let's get into the nuts and bolts then. How did they actually measure this alpha? It sounds crucial. Right. They didn't just pick one method. They used four different pretty standard factor models, you know, ways to strip out market wide effects and see what's left. Like CAPM. Yep. CAPM was one.
Then the Fama-French three-factor model that adds size and value factors. Also the Carhartt four-factor, which brings in momentum itself. Okay. And another three-factor model from Asness, Frasini, and Peterson. And a key thing, they scaled the calculated alpha by volatility. Ah, volatility scaling. So looking at the consistency of that excess return, not just the raw number, makes sense for comparing diverse markets. Precisely. A smoother, more comparable alpha signal.
So with this volatility adjusted alpha calculated, how did they build the actual trading rules? Let's start with the momentum one, AM. Okay, AM. That uses the alpha calculated over the trailing 12 months, but they lagged it slightly. So months T12 back to T1. Right, skipping the very last month. Yeah. And the strategy was a classic long-short approach. They ranked all the countries or all the industries based on this recent alpha momentum. And then? They went long the top 20%.
the ones with the highest AMOM and shorted the bottom 20%, the ones with the lowest AMO. Betting on the winners continuing and the losers lagging. Makes sense. So how did that perform in their back tests? Did this AMOM strategy actually generate alpha itself? It did. And the results were pretty striking, especially for the equal weighted portfolios. For countries using the three factor model alpha, they reported an average monthly alpha of 0.92%. 0.92% per month. That's significant.
Yeah. With a sharp ratio of 0.58. And for industries, it was even better. Alpha of 1.41% per month, sharp ratio of 0.93. Wow, nearly 1% sharp for the industry strategy. That's impressive on paper. But what about real world friction, like trading costs? Good question. They looked into that. The MM strategy held up surprisingly well. For countries, it was profitable, even assuming one-way trading costs of up to 1%. 1%, that's quite a buffer. It is. And industries could handle even higher costs.
Plus, the effect wasn't super short-lived. It were make profitable even if you held the positions for, say, 10 or 12 months. OK, robust. Did it work better in some markets than others? Like small versus large? That's where the evaluated results come in. When they weighted portfolios by market size, the alphas were lower, which suggests, yeah, the effect is maybe weaker or harder to capture in those bigger, perhaps more efficient markets. Interesting.
So maybe more juice to squeeze in the smaller pots. OK, let's flip to the other strategy, the reversal one. AREV, you called it. Yep. ARRIV for alpha reversal. This one looked at alpha over a much longer window, 60 months, so five years, but again, lagging the most recent year. So months T60 down to T13. Five years of alpha history, skipping the last one. Yeah. And the strategy here is the opposite of momentum. Exactly. Betting on mean reversion.
They went long the bottom 20% of the countries or industries with the lowest long-term alpha and shorted the top 20%, the ones with the highest long-term alpha. So buying the long-term alpha losers and selling the long-term alpha winners. What did the backtests show for AREV? Well, as you'd expect of reversal holds, the strategy generated negative alpha. For the equal-weighted country portfolios using the three-factor model, the alpha was negative 0.52% per month. Negative 0.52.
Okay. And for industries, similarly, negative 0.49% per month. The Sharpe ratios were lower, too, compared to the AOM strategy. So evidence for reversal, but perhaps not as strong or consistent as the short-term momentum effect. That seems to be the case. And again, when they looked at value-weighted portfolios for AREV, the results kind of washed out. They became statistically insignificant for both countries and industries.
So both alpha momentum and alpha reversal seem stronger or more exploitable in equal-weighted universes, perhaps hinting at effects in smaller or less efficient markets. That definitely seems to be a recurring theme, yes. Now, one really interesting comparison was alpha momentum versus plain old price momentum. How did those stack up? This was a key finding. They found that alpha momentum basically subsumes price momentum. Subsumes, meaning?
Meaning when they controlled for alpha momentum in their statistical tests, the predictive power of simple price momentum just disappeared. It became insignificant. Wow. But crucially, alpha momentum remained significant even when they controlled for price momentum. So the risk-adjusted performance is the real driver, not just the price trend itself. It strongly suggests that yes, it's a more fundamental signal, apparently. Fascinating. What about the reversal side?
Alpha reversal versus price reversal. Was the same story? It was a bit more complicated there. Alpha reversal did show explanatory power over price reversal in some situations, but it wasn't quite as clean cut as the momentum side. Price reversal sometimes still held significance. Okay, so maybe a bit more going on with the long term reversals. Did they look at why these effects might be stronger in certain markets, like limits to arbitrage? They did.
They explored whether things that make arbitrage harder, like smaller market size, higher company-specific risk, idiosyncratic volatility, or lower correlation with the global market affected the profitability. And did those factors matter? Yes, significantly. Both the AMO and ARV strategies were much more profitable in markets with higher limits to arbitrage. Okay, so where it's harder for big players to create these effects away, the anomalies persist more strongly. Exactly.
They showed some clear examples. The alpha spreads between the long and short portfolios were much wider in, say, smaller markets or markets with higher volatility. It was also stronger in emerging markets compared to developed ones. Makes sense. Arbitrage constraints letting these inefficiencies linger. Did the effects hold up over time? Did they look at different superiors? They did. The momentum strategy seemed pretty stable across different time periods, which is encouraging.
Good. Interestingly, they found the alpha momentum effect was actually stronger following bull markets. Huh. Any theories why? Maybe behavioral biases getting amplified or trends persisting more strongly when sentiment is high. Hard to say definitively from this. On the flip side, the reversal effects seem to get a bit weaker in the later years of their sample. OK, so momentum holding up, reversal perhaps fading slightly. Did market conditions matter, like high stress periods?
Yes, they checked that too. The reversal effect, AREV, was stronger during periods associated with high limits to arbitrage, like when the VIX was high or credit spreads were wide or the TED spread spiked. Also during times of high investor sentiment. So reversal gets amplified when markets are stressed or perhaps overly optimistic. That's another layer to consider. Absolutely. It suggests these aren't just constant effects.
Their strength can ebb and flow with broader market conditions and sentiment. This has been a really insightful look at alpha momentum and reversal. It definitely moves beyond simple price trends. Yeah, it suggests that looking at that risk-adjusted performance history might offer a more refined edge for global allocation, both short-term and long-term. 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.