Transcription

• Speaker #0

  Hello, welcome back to Papers with Backtest podcast. Today, we dive into another algo trading research paper. We're going to explore a fascinating study that combines two really intriguing concepts in the stock market, market sentiment and the overnight anomaly.

• Speaker #1

  You know how everyone always says buy low and sell high? Well, what if I told you one of the best times to buy might be while you are sleeping?

• Speaker #0

  Okay, now that's a strategy I can get behind. Tell me more.

• Speaker #1

  That's the intriguing idea behind this thing called the overnight anomaly. Basically, it's this curious tendency we see for U.S. stocks to do surprisingly well when the markets closed overnight. We're talking higher returns during those, you know, sleepy hours compared to the hustle and bustle of the regular trading day.

• Speaker #0

  So while most of us are catching Zs, the markets are quietly making moves. It seems counterintuitive. You'd think the big gains would happen when everyone's trading.

• Speaker #1

  Right. And for a long time, it was just a theory, you know, something traders would whisper about. But here's the thing. It's actually been confirmed by multiple academic studies. The overnight anomaly is a real thing. And what makes it even more compelling is that a big chunk of those market beating returns, what we call the equity risk premium, seems to be driven by these overnight gains.

• Speaker #0

  So this isn't just some statistical quirk. It's potentially a very lucrative pattern.

• Speaker #1

  Exactly. And the study gives a very concrete example. Looking at the SPY EPF, which tracks the S&P 500, they showed that the SPY EPF is a very lucrative pattern. And it's a very lucrative that from 2018 to 2021, the SPY delivered significantly higher cumulative returns overnight than during the day.

• Speaker #0

  That's a compelling illustration. But let's get to the why. Why this nine-hour advantage? What is actually driving these after-hours gains?

• Speaker #1

  Well, while there isn't a single definitive answer, there are some compelling theories out there that attempt to explain this phenomenon. One is this idea of order accumulation. Imagine buy orders kind of piling up overnight like a queue at your favorite coffee shop before it opens.

• Speaker #0

  So like a pre-market rush, but for stocks.

• Speaker #1

  Exactly. And then when the market opens, boom, there's this surge in buying pressure, which leads to a price jump. This surge often fades as the trading day goes on and those accumulated orders get filled.

• Speaker #0

  Kind of like a mini flash sale every morning, only it's for, you know, a slice of the S&P 500 instead of discounted shoes.

• Speaker #1

  Precisely. Another possibility is what's called the illiquidity premium. Basically, overnight, the trading volume tends to be much thinner, right? And so with fewer participants, those who do trade might demand a slightly higher return for holding those stocks during these quieter, less liquid periods. It's like a small reward for taking on a tiny bit of extra risk.

• Speaker #0

  OK, so we've got this intriguing overnight anomaly, this potential for outsized returns while we sleep. But then this study throws in another layer, market sentiment.

• Speaker #1

  And that's where things get really, really interesting. You see, market sentiment, that overall mood of investors, whether they're feeling bullish or bearish. can be a really powerful force. The researchers wanted to see if they could juice the overnight anomaly even further by factoring in sentiment indicators. Could they predict, in a sense, when those overnight gains might be even stronger?

• Speaker #0

  So instead of just blindly buying at the close every night and hoping for the best, they thought, let's see if we can use the market's mood to tip us off.

• Speaker #1

  Exactly. They wanted to see if they could pinpoint specific conditions where the overnight anomaly was more likely to really kick in.

• Speaker #0

  I love that. So for all of our learner listeners out there who crave those nitty gritty details, walk us through how they set up this study. What were the trading rules?

• Speaker #1

  So they kept it really focused, using the SPY ETF as their testing ground. Now for gauging market sentiment, they actually used three different indicators. First, they looked at the SPY's price trend itself. You know, was it above or below its 20-day moving average? This is a simple but surprisingly effective way to see if the SPY was in an upswing or a downswing.

• Speaker #0

  A classic trend following signal, a quick gut check on the SPY's short term momentum.

• Speaker #1

  Exactly. Next, they brought in the VIX, the CBOE Volatility Index. Now, the VIX is often called the market's fear gauge because it tends to spike when investors are uncertain or scared. So they wanted to see if the VIX was above or below its own 20 day moving average. Generally, a lower VIX suggests less fear, more complacency in the market, which could be a good sign for those overnight gains.

• Speaker #0

  Got it. So the VIX... acts as a counterpoint to the SPY's price action. It gives them a sense of the market's emotional temperature, so to speak.

• Speaker #1

  Precisely. And then for their third sentiment indicator, they got a bit more sophisticated. They decided to incorporate something called the brain market sentiment or BMS indicator.

• Speaker #0

  DMS. Okay, that sounds intriguing. Give us the breakdown.

• Speaker #1

  So it uses the power of natural language processing, a form of artificial intelligence, to analyze thousands and thousands of financial news articles. Think of it as an army of tireless interns reading the Wall Street Journal and the Financial Times and then summarizing the overall market sentiment.

• Speaker #0

  An AI-powered sentiment analyst. That is incredibly efficient. But how do you get an actual sentiment score from that?

• Speaker #1

  So the BMS processes the language used in those articles, looking specifically for positive or negative tones related to the market. Based on its analysis, it spits out a daily score ranging from 0 to 100. A higher score, of course, indicates more positive overall sentiment.

• Speaker #0

  So a high BMS score means the financial media is buzzing with good vibes, potentially setting the stage for a very strong overnight session. Exactly. Okay, so they've got their trading instrument, the SPY ETF. And there are three sentiment indicators, the SPY's trend, the VIX, and the BMS. What was the game plan? How did they actually use this information to trade?

• Speaker #1

  They kept their trading strategy remarkably straightforward. They would buy SPY at the close each day. But here's the catch. They would only hold it overnight if all three sentiment indicators were flashing green, meaning they were all pointing toward bullish sentiment.

• Speaker #0

  It's like they were looking for a trifecta of bullish signals before making their move. No half-hearted entries here.

• Speaker #1

  Exactly. All systems had to be go SPY trending up, VIX trending down, and the news sounding positive according to the BMS. Only then would they place their overnight bet. It was a high bar to clear, but as we'll see, the results were really quite interesting.

• Speaker #0

  All right, enough with the suspense. Did it work? Did filtering for sentiment supercharge the overnight anomaly?

• Speaker #1

  Here's where things get really, really interesting.

• Speaker #0

  I love when you say that.

• Speaker #1

  They backed. Black tested this strategy, meaning they tested it on historical data to see how it would have performed. What they found was that each sentiment filter on its own actually improved the returns of simply buying and holding SPY overnight.

• Speaker #0

  So even just using one of those indicators, the SPY trend, the VIX or the BMS, gave them an edge over just blindly buying every night.

• Speaker #1

  That's right. Each indicator seemed to offer some predictive power. But here's where things get really exciting. The magic truly happened when they combined all three indicators.

• Speaker #0

  Creating sort of sentiment powerhouse.

• Speaker #1

  You could say that. They created what they called an equally weighted portfolio, which basically means they allocated an equal amount of capital to each individual sentiment strategy.

• Speaker #0

  So spreading their risk and letting the combined wisdom, so to speak, of those sentiment indicators guide their trades.

• Speaker #1

  Exactly. And this combined approach, this demand for all three indicators to align. actually led to even higher risk-adjusted returns and lower drawdowns than any of the individual signals in isolation.

• Speaker #0

  So the hole was greater than the sum of its parts. By being selective, by waiting for that confluence of bullish signals, they managed to isolate the most potent periods for the overnight anomaly. That's impressive. But what kind of returns are we actually talking about here? What did the numbers look like?

• Speaker #1

  Well, remember, they were trading the SPY ETF, which obviously comes with some inherent market volatility. Without any filtering, just buying and holding SPY overnight during this period delivered an annualized return of about 13.6%, which is pretty impressive on its own.

• Speaker #0

  Not too shabby. But what happened when they added in the sentiment filter?

• Speaker #1

  When they applied their triple sentiment filter, those returns jumped to 15.58%. That's a pretty significant improvement, especially when you consider it came with potentially lower risk as well.

• Speaker #0

  Wow. That's a substantial increase. Yeah. But you mentioned lower risk. Did their strategy- also reduce those drawdowns, those stomach churning dips that can really test even the most seasoned traders' nerves.

• Speaker #1

  That's the best part. Not only did the returns improve, but their maximum drawdown, meaning the biggest peak to trough decline they experienced during the backtest, was significantly smaller when using the sentiment filter. We're talking about cutting the maximum drawdown by more than half.

• Speaker #0

  So less risk and higher returns. That's the holy grail of investing. This almost sounds too good to be true. Are there any caveats to this strategy? What should our listeners keep in mind before they start backtesting this themselves?

• Speaker #1

  That's an important point. While this research offers some really compelling insights, it's super important to remember that market dynamics are constantly evolving. They're fluid, always in motion. This study focused specifically on the SPY ETF and the U.S. stock market. Trying to apply these exact methods to other asset classes or global markets well, you might get very, very different results.

• Speaker #0

  So no blindly copying and pasting this strategy to, say, emerging markets or cryptocurrencies?

• Speaker #1

  Definitely not. Every market has its own unique, you know, rhythms and drivers. What works in one context might not translate to another. Also, it's super important to remember the market conditions during this study period.

• Speaker #0

  And the period they analyzed, 2018 to 2021, while not without its hiccups, was relatively calm and bullish overall for U.S. equities.

• Speaker #1

  You're spot on. We had a few pullbacks, but nothing like a sustained bear market or a major economic crisis. So it's entirely possible that the effectiveness of this overnight anomaly, even with the sentiment filtering, might diminish or even disappear during different market regimes.

• Speaker #0

  Like in a prolonged downturn where fear and uncertainty are the dominant emotions?

• Speaker #1

  Exactly. That's where rigorous backtesting and perhaps even more importantly, forward testing become absolutely critical.

• Speaker #0

  For those who might be unfamiliar, backtesting is looking at how a strategy would have performed historically under different conditions, right? And forward testing is taking that strategy and actually testing it in real time with real money on the line.

• Speaker #1

  Precisely. And even then, there's no guarantee of future results. Trading, especially when you're trying to capture these more subtle edges, is a game of probabilities, not certainties.

• Speaker #0

  That's a crucial point for our listeners to remember. It's about tilting the odds in your favor. Not finding some foolproof system that just prints money.

• Speaker #1

  Exactly. And on that note, another factor they didn't explicitly account for is slippage and commissions.

• Speaker #0

  Ah, yes. The real-world costs of trading that can really eat into your profit.

• Speaker #1

  Exactly. Their backtests assumed you could buy and sell at the closing and opening prices. But in reality, prices fluctuate, especially in those after-hours and pre-market periods.

• Speaker #0

  So you might not get the exact price you want, especially if you're dealing with larger orders.

• Speaker #1

  Exactly. And then you have those pesky commissions. Those costs can add up, especially if you're trading frequently, as this strategy might suggest.

• Speaker #0

  So it's essential to factor in those trading costs into your calculations to get a true sense of the strategy's profitability. But speaking of profitability, did the researchers provide any concrete guidance on how traders could implement this strategy beyond the high-level concept? Did they offer any specific entry or exit rules?

• Speaker #1

  They didn't. And that was deliberate. Their goal wasn't to hand over a ready-made trading system. They were exploring market behavior, trying to see if there were exploitable patterns related to sentiment and this overnight anomaly.

• Speaker #0

  It's like they've given us a treasure map, but they haven't marked the exact spot to dig.

• Speaker #1

  Precisely. They've provided a framework, the evidence that there might be something here worth exploring further. Now it's up to individual traders to experiment. to test and to see if they can turn these insights into a profitable strategy that fits their own risk tolerance and trading style.

• Speaker #0

  And I imagine that process of adaptation and optimization would look different for everyone.

• Speaker #1

  Absolutely. For some, it might mean incorporating additional indicators or filters, maybe looking at different asset classes or exploring how this overnight effect plays out in different market conditions.

• Speaker #0

  For others, it might mean adjusting the holding period, perhaps holding for multiple nights if the sentiment signals remain strong. Or implementing a more nuanced exit strategy to lock in profits.

• Speaker #1

  Exactly. The beauty of this research is that it really opens up more questions than it answers, inviting further exploration and experimentation. It's a fantastic starting point for anyone who's intrigued by the idea of harnessing market sentiment to potentially enhance returns from the overnight anomaly.

• Speaker #0

  It's like they've handed us this powerful new lens through which to view the markets, encouraging us to explore, to question. and ultimately to find our own unique edge.

• Speaker #1

  And that's part of what makes trading so fascinating, right? It's a blend of art and science where you're constantly learning, adapting, and seeking out those profitable opportunities.

• Speaker #0

  Absolutely. Now, before we shift gears and dive even deeper into the specifics of their findings, let's take a quick break. When we come back, we'll unpack the result in more detail and explore those nuances that could help you turn this academic insight into actionable trading strategies. Stay with us.

• Speaker #1

  Welcome back. Now let's dig into some of the more intriguing details of their findings, particularly around those sentiment indicators. Remember, they actually tested a few different moving averages, 10-day, 20-day, and 50-day, to kind of assess the trends in the SPY, VIX, and BMS.

• Speaker #0

  Yep, testing different timeframes to see what worked best for capturing those sentiment shifts.

• Speaker #1

  Exactly. And interestingly, the 20-day moving average really emerged as like the sweet spot across all three sentiment indicators. Using that 20-day look-back period... consistently produce the best results in their back tests.

• Speaker #0

  Interesting. So not too short, not too long. The 20-day moving average seemed to strike that balance for capturing the sentiment signals that mattered most for this overnight play. Any idea why that might be the case?

• Speaker #1

  It's tough to say for sure, but remember, market dynamics are always in motion, right? So those shorter-term moving averages, like the 10-day, might be a little too sensitive to those just day-to-day fluctuations in price and sentiment, right? Right. Whereas longer ones, like the 50-day, might be too slow to react to more meaningful shifts. So the 20-day moving average might be capturing that sweet spot, you know, that time frame where those sentiment shifts, whether bullish or bearish, have had enough time to really play out and influence that overnight session.

• Speaker #0

  Makes sense. It's about finding that rhythm in the market, that sweet spot where sentiments influence is most pronounced.

• Speaker #1

  Precisely. And there's another fascinating little nuance they uncovered. This one relates to when those overnight gains were most pronounced. They found that this overnight anomaly, even with the sentiment filters, tended to be much stronger after periods when SPY had actually experienced negative overall returns.

• Speaker #0

  Interesting. So if SPY had a rough night, the odds of a bounce back were higher, especially if the sentiment indicators were aligned, like the market was overreacting to the downside. And then with those bullish sentiment signals in place, it kind of self-corrected.

• Speaker #1

  That's what their data suggests. Yeah. It speaks to the potential for mean reversion, even in the very short term. If sentiment is broadly bullish, you know, those down nights might be seen as buying opportunities by other, you know, savvier market participants. leading to that overnight surge when the market reopens.

• Speaker #0

  Almost like the market's taking a breather, reassessing, and then resuming its upward course.

• Speaker #1

  Exactly. It's a fascinating interplay between sentiment, price action, and those who are seeking to capitalize on these patterns.

• Speaker #0

  It is fascinating. This is all incredibly insightful. But I'm also very aware that every backtest, every study like this has its limitations. What are some things we should keep in mind about this research? What didn't it cover that our listeners should be aware of before they start? you know, applying any of this.

• Speaker #1

  You're right. Context is key when you're evaluating any sort of trading strategy or market anomaly. And one really important thing to remember is that this study focused on a specific time period. Right. And a very specific market environment. The period they analyzed, 2018 to 2021, was generally pretty favorable for U.S. equities. So it's important to consider how this strategy might perform in different market conditions, such as a bear market or a period of, you know. heightened volatility.

• Speaker #0

  So past performance is not necessarily indicative of future results, especially when those market dynamics change.

• Speaker #1

  Exactly. Additionally, while the researchers did test different moving average periods, they did keep other variables constant. For example, they only looked at the SPY ETF. They didn't explore how the strategy might perform with, say, other assets or in different sectors.

• Speaker #0

  So there's room for further research and exploration, potentially uncovering even more nuanced applications of this concept.

• Speaker #1

  Exactly.

• Speaker #0

  What about the trading costs? I know we touched on slippage in commissions earlier, but how significant a factor were those in their backtests?

• Speaker #1

  Yeah. They acknowledged that trading costs could impact the profitability of the strategy in real world trading, but they didn't actually factor those costs into their backtests. So it's super important for traders to actually consider these costs because they can vary a lot depending on the broker, the trading platform, the order types used, all sorts of things.

• Speaker #0

  And they're not going to be able to do that with seemingly small fees. can add up, especially if you're making frequent trades, potentially eroding those overnight gains.

• Speaker #1

  Precisely. It's always this delicate balance, right, between identifying those potentially profitable opportunities and then managing your trading costs to make sure you're actually coming out ahead in the long run.

• Speaker #0

  Absolutely. Now, shifting gears a bit, I'm curious about the practical application of these findings. The researchers didn't really delve into specific entry and exit points for trades. They were much more focused on this overall concept and potential of combining sentiment indicators with this overnight anomaly. Do you think there are ways for traders to kind of build upon this framework and develop more concrete trading rules?

• Speaker #1

  Absolutely. And that's, you know, that's where the real excitement lies, right? Taking these academic insights and really crafting practical, actionable trading strategies. Remember, the researchers were really exploring the sort of broad market behavior. They weren't designing like a ready to trade system. So they've given us a fantastic foundation. But it's up to individual traders to really experiment and refine and personalize these concepts to fit their own risk tolerance, their trading style, their market outlook, all of that.

• Speaker #0

  It's like they've given us the building blocks in a blueprint, but we get to decide on the layout. The finishes really make it our own.

• Speaker #1

  Exactly. So for some traders, that might involve incorporating additional technical indicators, you know, maybe confirming those sentiment signals with, say, momentum oscillators or volume based indicators. Others might experiment with different entry and exit techniques, maybe using limit orders to try to mitigate slippage or setting trailing stops to protect profits.

• Speaker #0

  Right, right. It's about finding that balance between, you know, following a set of rules, but also being able to adapt to the ever changing dynamics of the market.

• Speaker #1

  Precisely. And remember, even with, you know, even with a well-defined strategy, risk management is still paramount. You know, things like position sizing, setting stop loss levels. And just having a very clear understanding of your own personal risk tolerance are crucial elements of successful trading.

• Speaker #0

  Absolutely. Don't bet the farm on any single trade, no matter how promising it seems. Now, before we wrap up our deep dive into the world of overnight anomalies and sentiment driven trading, I want to touch on one more aspect of their findings. They highlighted that this strategy performed particularly well during certain market conditions, specifically after periods of negative overnight returns for the SPY. What are your thoughts on this? Could this be a potential trigger, so to speak, for traders looking to capitalize on this strategy?

• Speaker #1

  It's definitely an interesting observation and one that speaks to, I think, the cyclical nature of markets, right? When we see these periods where the market, as represented by the SPY in this case, kind of experiences a pullback, especially during those quieter overnight sessions, it often creates what traders call, you know, a dip to buy, right? And if that underlying sentiment is still broadly bullish, those dips are often seen as opportunities to maybe enter the market at a more favorable price.

• Speaker #0

  So it's almost like the market is resetting itself, you know, shaking off some of the excess exuberance or fear and then potentially resuming that upward trajectory.

• Speaker #1

  Exactly. And those who are able to kind of identify those moments, especially when they're supported by other confirming indicators, you know, like the sentiment filters used in the study, they might be able to position themselves to benefit from those potential rebounds. However, and this is a big application, however, it's crucial to remember that mean reversion. You know, it's not a guaranteed outcome. Markets can remain irrational much longer than you might expect. What appears to be a dip can very quickly turn into a prolonged downturn.

• Speaker #0

  That's why it's so crucial to have that well-defined strategy, manage your risk, and just be prepared for any eventuality. Now, for our listeners who might be inspired to explore this strategy further, are there any resources or tools you'd recommend?

• Speaker #1

  Absolutely. I mean, one of the best ways to really deepen your understanding is to go straight to the source. We'll be sure to include a link to the original research paper in the show notes.

• Speaker #0

  That's a great idea. Reading the original research allows you to really delve into their methodology, analyze the data for yourself, and draw your own conclusions.

• Speaker #1

  Precisely. And then beyond that, you know, there are tons of online resources and trading platforms that offer backtesting tools and sentiment indicators. So experimenting with those tools, running your own simulations, you know, really testing out different variations of the strategy, that can be incredible. incredibly valuable.

• Speaker #0

  It's about taking those theoretical concepts and putting them into practice, seeing what works for you in your own trading with your own risk tolerance. Exactly. Now, before we move on to our final thoughts on this fascinating topic, let's let's jump right back in. So we've been talking about this idea of using sentiment to potentially boost returns from the overnight anomaly. Big picture. What are the main takeaways for traders? What should they really be thinking about as they. maybe explore these concepts further?

• Speaker #1

  I think, you know, first and foremost, this research really reminds us that market anomalies do exist. You know, this overnight anomaly we've been talking about, it's not just some myth. It's a statistically validated pattern that has actually persisted for decades. And you know, in a way, understanding exactly why it exists is kind of less important than just recognizing that it represents this potential opportunity for those who are willing to explore it.

• Speaker #0

  Right. It's about challenging those, you know, conventional assumptions about when and where profits are made in the markets.

• Speaker #1

  Precisely. Second, you know, I think the study really highlights the power of combining different approaches. They didn't just rely on the overnight anomaly itself, right? They layered in that sentiment analysis using a mix of those technical indicators and that new sentiment we talked about to try to pinpoint those periods when the anomaly might be even stronger.

• Speaker #0

  So it's like finding those confluences of evidence, those moments when multiple signals are aligning to increase your odds of success.

• Speaker #1

  Exactly. And then finally. you know, this is key. No trading strategy is foolproof. We have to remember that. Markets are dynamic. They're constantly evolving. And what worked in the past might not necessarily work in the future. So rigorous backtesting, forward testing, prudent risk management, those are all essential elements of really any successful trading approach.

• Speaker #0

  Right. It's about approaching the markets with this healthy balance of like curiosity and skepticism and discipline.

• Speaker #1

  Well said. You know, this research isn't about blindly following signals. It's about using these signals to inform your own trading decisions.

• Speaker #0

  To enhance your understanding of how the markets work and potentially gain an edge over time.

• Speaker #1

  Exactly. It's about exploration, experimentation, and just continuous learning.

• Speaker #0

  I love that. Well, I think we've taken a pretty deep dive into this fascinating topic. We've explored that intriguing overnight anomaly, unpacked the potential of sentiment analysis, and hopefully sparked some new ideas for our listeners as they navigate the markets. So for those who want to learn more, dig into the details, we'll be sure to include a link to the study in the show notes. Anything else you'd add?

• Speaker #1

  Nope. I think you covered it all. It was a great discussion.

• Speaker #0

  Awesome. Well, thanks for joining me on this deep dive. Until next time, happy trading, everyone.