Buyers are more and more exploring various indicators to realize an edge in monetary markets. Conventional indicators, reminiscent of earnings stories or macroeconomic knowledge, usually include delays or could already be priced in. Consequently, unconventional metrics have attracted consideration. For instance, current focus has been on building stock statistics, the place massive stockpiles have been interpreted as a sign of weakening demand for building exercise. This, in flip, could mirror broader shopper and company hesitancy to spend, implicitly suggesting a possible decline in asset costs. On this article, we look at a unique sort of different indicator: the Producer Worth Index (PPI) for the Corrugated and Stable Fiber Field Manufacturing trade, together with corrugated bins and pallets. Our motivation is to judge this index’s effectiveness as a predictive sign for the S&P 500 ETF, sector-specific ETFs, and particular person shares reminiscent of Amazon (AMZN), one of many largest customers of supplies tracked by this index. We current a number of funding methods that incorporate this indicator and assess whether or not it might improve risk-adjusted returns.
Motivation
Relying completely on conventional prediction inputs reminiscent of inventory costs, quarterly earnings stories or ahead steering comes with a structural limitation. Monetary markets often incorporate expectations lengthy earlier than the knowledge turns into public. Consequently, a lot of the anticipated efficiency is already mirrored within the value by the point earnings are launched. Coming into positions shortly earlier than these bulletins primarily turns right into a guess on whether or not the outcomes will outperform what’s already priced in. This introduces a degree of event-driven threat that’s tough to handle and presents little basis for constructing systematic methods with steady threat profiles.
This creates a transparent incentive to discover indicators that seize actual financial exercise with out being instantly absorbed by market expectations. With the PPI for Corrugated and Stable Fiber Field Manufacturing, we construct on an intuitive thought. Rising manufacturing or consumption of packaging supplies indicators larger demand for these inputs. Corporations buy extra corrugated bins and pallets when they should ship a bigger quantity of products. Elevated packaging demand subsequently serves as a proxy for rising order flows, stronger shopper exercise and better throughput throughout provide chains. These situations usually accompany durations of financial growth, which are likely to help company revenues and earnings. If this relationship holds, an upward pattern in packaging-related PPI knowledge could precede rising costs in broad market ETFs, sector-specific ETFs or particular person shares reminiscent of Amazon.
Knowledge overview
The evaluation depends totally on month-to-month knowledge. This frequency is dictated by the traits of our main indicator, the PPI for Corrugated and Stable Fiber Field Manufacturing, which is launched solely as soon as per 30 days and sometimes not on a set schedule. Consequently, all technique choices are applied with a one-month delay and evaluated at month-to-month intervals.
Our predictor sequence is out there from March 1980, offering an extended historic window for testing. For comparability with market devices, we embody a number of broadly used exchange-traded funds and particular person equities, every with its personal begin date based mostly on knowledge availability. Amazon (AMZN) enters the evaluation from February 2000. Sector ETFs are included starting in December 1998, particularly Client Discretionary (XLY), Utilities (XLU), Well being Care (XLV) and Client Staples (XLP). These funds characterize distinct segments of the economic system and provide a structured technique to observe how completely different sectors reply to modifications in underlying financial exercise. The broad market proxy SPY, monitoring the S&P 500, is included from December 2004. For a low-risk asset for instances of uncertainty, we use BIL, a short-term Treasury ETF, obtainable from June 2007, which serves as a cash-like various inside our technique exams.
A number of further ETFs have been reviewed in the course of the exploratory part, together with Industrials (XLI), Retail (XRT) and the On-line Retail ETF (ONLN). Their outcomes didn’t differ meaningfully from the sectors already offered, and in some circumstances the obtainable historical past was too quick to offer sturdy conclusions. For readability and relevance, they’re subsequently not included within the remaining set of devices mentioned on this article.
AR fashions and their limitations
One pure thought was to discover autoregressive fashions, since it isn’t the extent of the indicator that seems most related, however its change. This naturally results in working with first variations. If month-to-month actions seize significant shifts in financial exercise, one may assemble a rule through which deviations of a sure dimension generate purchase or promote indicators. Conceptually, this resembles making a band across the differenced sequence and reacting at any time when the index strikes exterior that band.
Though interesting, this method carries an necessary implicit assumption. It really works provided that the variability of the sequence stays roughly fixed over time. In statistical phrases, the tactic assumes homoskedasticity. In our case, nonetheless, the variability of the PPI variations doesn’t stay steady. Over the lengthy pattern, the fluctuations develop into bigger and the amplitude of actions will increase. When the variance grows, the differenced sequence begins to provide a big quantity of noise, making the indicators unstable and lowering the reliability of any band-based technique. As an alternative of capturing significant modifications in financial exercise, the mannequin more and more reacts to the altering scale of the info itself.
Idea of heteroskedasticity
A time sequence is heteroskedastic in our setup when variance of its first variations will not be fixed throughout time. Intervals of calm are adopted by durations of bigger swings, and the general dispersion grows or shrinks (un)predictably. For monetary or financial indicators, this can be a frequent phenomenon. The primary drawback arises when a mannequin assumes fixed variance however the knowledge don’t fulfill that assumption. Methods that depend on mounted thresholds or mounted bands can develop into distorted as a result of what appears like a big deviation in a single interval could merely be routine volatility in one other. In follow, heteroskedasticity results in unreliable indicators, extreme turnover and a decrease signal-to-noise ratio, all of which cut back the effectiveness of strategies based mostly on easy autoregressive dynamics or static thresholds.
Incorporating this time-varying volatility right into a predictive mannequin considerably complicates its design. The presence of heteroskedasticity implies that fixed-threshold guidelines or static bands now not translate constantly throughout all the pattern, and the mannequin should account for the altering scale of fluctuations. Even makes an attempt to explicitly introduce time-dependent variance changes don’t essentially enhance predictive energy, as a result of the signal-to-noise ratio should stay low and the underlying financial relationships will not be captured just by scaling the sequence. Consequently, methods based mostly on naive autoregressive frameworks or static thresholds can lose reliability when utilized to knowledge with evolving variability.
MA fashions
Following the destructive expertise with autoregressive fashions, consideration turned to shifting common (MA) approaches. In contrast to AR fashions, MA fashions are much less delicate to heteroskedasticity as a result of they easy the sequence domestically, making the long-term construction of the variance operate largely irrelevant. Two pure instructions emerge when working with MA fashions. The primary is to trace N-month maxima, whereas the second is to check values in opposition to the N-month shifting common.
Methods based mostly on each approaches have been examined, however N-month maxima carried out comparatively poorly. This seems to be as a result of maxima primarily seize traits fairly than short-term deviations, that are the main focus of our predictive efforts. By the point a neighborhood most is reached, the chance for early sign detection is already partially misplaced. In distinction, evaluating the present worth to the N-month shifting common permits us to determine vital deviations from the pattern, successfully capturing uncommon actions relative to the anticipated degree. From a broader perspective, this is smart, as smoothing the sequence gives a clearer view of the underlying dynamics. Primarily based on these concerns, we undertake the moving-average-based method as the inspiration for the methods offered on this article.
Change fashions and financial uncertainty
In a number of of our earlier articles (for instance, about BTC ETFs or VIX-based property), we’ve employed what are generally known as change methods. These methods are based mostly on the precept of dynamically adjusting portfolio allocations in keeping with the indicators generated by a number of indicators. Reasonably than sustaining a set allocation throughout all property or sectors, the technique “switches” between completely different portfolios relying on the noticed state of the indicator.
Change technique motivation
The change technique can be understood when it comes to sector sensitivity to financial situations. For instance, a Client Staples ETF sometimes represents corporations producing important items, reminiscent of meals and family merchandise, which stay in demand even throughout financial slowdowns. In distinction, a Client Discretionary ETF covers sectors like retail, leisure, or luxurious items, the place demand is extra delicate to shopper spending and financial cycles. In durations of weakening demand or heightened threat, traders usually tend to cut back spending on discretionary objects, inflicting these property to underperform relative to staples. The change technique leverages this perception by allocating capital between defensive ETFs, reminiscent of Client Staples, and extra risky, growth-oriented ETFs, like Client Discretionary, in keeping with the sign from our indicator. When the indicator suggests robust demand or financial growth, the technique favors discretionary sectors; when the indicator indicators warning, it rotates towards staples to guard the portfolio. This method permits for a scientific response to shifts in financial situations whereas balancing threat and return.
SPY fashions as a naive benchark
Armed with these insights, we will now flip to empirical evaluation. To start out, we assemble a set of comparatively easy benchmark exams to judge the efficiency of our method. As an preliminary train, we apply the technique to the SPY ETF, which tracks the S&P 500 and serves as a broad-market reference. By testing on SPY first, we set up a baseline understanding of how the indicator-driven change technique behaves in follow, earlier than extending the evaluation to sector-specific ETFs or particular person shares. This preliminary benchmark helps spotlight each the potential advantages and the constraints of our methodology in a simple, managed setting.
SPY technique
1. Open place in SPY, when field predictor < 3M/6M/9M MA of field predictor.
2. Shut place in SPY, when field predictor reaches its shifting common.
3. If no place is opened, keep in money.
Desk 1: Efficiency metrics of SPY-based change methods.
PORTFOLIO
CAR p.a.
VOL p.a.
MAX DD
SHARPE
CALMAR
SPY portfolio
10.48%
14.84%
-50.76%
0.71
0.21
3M SPY portfolio
4.41%
10.68%
-27.62%
0.41
0.16
6M SPY portfolio
6.14%
9.94%
-19.44%
0.62
0.32
9M SPY portfolio
5.16%
8.25%
-19.44%
0.63
0.27
The outcomes clearly present that making use of this technique on to SPY will not be a promising method (when it comes to return). The explanation lies within the construction of the S&P 500 itself. As a broad market index, it accommodates corporations from nearly each sector, lots of that are solely marginally affected by fluctuations in consumer-sensitive industries. Though some portion of the index naturally displays modifications in shopper demand, the impact is diluted throughout a large and heterogeneous set of constituents. Consequently, the sign derived from packaging-related PPI knowledge doesn’t translate into sufficiently robust or well timed actions within the index. As well as, the technique stays inactive for prolonged durations, avoiding sure market downturns however on the similar time failing to seize sufficient of the index’s development phases. This mix of weak linkage to the underlying indicator and inadequate participation in rising markets results in total efficiency that’s removed from compelling when utilized to SPY.
Regardless that the general efficiency on SPY was not significantly robust, the outcomes reveal an necessary sample. The methods improved risk-adjusted metrics and subsequently are a attainable place to begin for our evaluation.
Change fashions for sector ETFs
Earlier outcomes confirmed that introducing a change mannequin can enhance risk-adjusted efficiency, however the method used thus far had two notable limitations. First, the defensive aspect of the change rule remained unchanged all through the evaluation, which restricted the mannequin’s skill to do no less than one thing in several market environments. Second, by making use of the technique to the broad S&P 500, we diluted the financial sign that originates from consumer-dependent exercise. This lowered the effectiveness of the indicator and restricted the technique’s potential.
A pure subsequent step is subsequently to shift the main focus from all the market to sector-specific ETFs. These provide extra direct publicity to the financial mechanisms our indicator captures. By inserting Client Discretionary (XLY) on the offensive aspect and Client Staples (XLP) on the defensive aspect, we align the technique with the elemental pattern we intention to take advantage of: discretionary spending expands strongly in favorable situations, whereas staples present resilience throughout downturns. Additionally it is cheap to check longer moving-average home windows to seize extra steady dynamics and doubtlessly produce extra sturdy outcomes. This mix permits us to higher match the indicator with the sectors most delicate to the underlying financial exercise and thereby improve the efficiency of the change mannequin.
XLY – XLP change technique
1. Open place in XLY, when field predictor < 3M/6M/9M/12M/24M/36M MA of field predictor.
2. Shut place in XLY, when field predictor reaches its shifting common. Open place in XLP.
3. Shut place in XLP, when field predictor falls under its 3M/6M/9M/12M/24M/36M MA.
Desk 2: Efficiency metrics of XLY – XLP change methods.
PORTFOLIO
CAR p.a.
VOL p.a.
MAX DD
SHARPE
CALMAR
XLY portfolio
9.70%
19.24%
-54.93%
0.5
0.18
XLP portfolio
6.73%
12.23%
-32.82%
0.55
0.21
3M XLY – XLP change portfolio
9.92%
15.98%
-33.95%
0.62
0.29
6M XLY – XLP change portfolio
11.12%
16.23%
-33.02%
0.69
0.34
9M XLY – XLP change portfolio
10.43%
15.84%
-28.05%
0.66
0.37
12M XLY – XLP change portfolio
9.54%
15.05%
-30.19%
0.63
0.32
24M XLY – XLP change portfolio
9.64%
14.57%
-28.05%
0.66
0.34
36M XLY – XLP change portfolio
7.33%
13.81%
-36.06%
0.53
0.20
The sector-level outcomes reveal that the change mannequin turns into a extremely efficient determination instrument when utilized to XLY and XLP. The enhancements are seen throughout all key metrics: total returns rise, the Sharpe ratio will increase, and the Calmar ratio strengthens as nicely. This implies that aligning the mannequin with consumer-driven financial cycles gives a a lot cleaner sign than working on the broad-market degree.
There’s, nonetheless, one necessary caveat. Such a robust enchancment raises the chance that the technique is partially overfitted to the particular sector pair or the chosen parameter home windows. To deal with this concern, it’s helpful to broaden the scope of the evaluation and discover ensemble-type approaches—fashions that mix a number of change methods directly. By aggregating a number of indicators, we could cut back sensitivity to any single sector, time window, or parameter alternative, and doubtlessly get hold of extra steady efficiency out-of-sample.
As a subsequent step, we assemble portfolios that evenly mix a number of moving-average home windows to scale back sensitivity to any single parameter alternative. Particularly, we create one set of portfolios that averages indicators throughout 3- to 12-month home windows, and one other set that spans 3- to 36-month home windows.
Desk 3: Efficiency metrics of mixing XLY – XLP change methods.
PORTFOLIO
CAR p.a.
VOL p.a.
MAX DD
SHARPE
CALMAR
3+6+9+12M XLY – XLP change portfolio
10.42%
14.76%
-26.57%
0.71
0.39
3+6+9+12+24+36M XLY – XLP change portfolio
9.90%
13.68%
-27.16%
0.72
0.36
Combining these a number of moving-average methods reduces each threat and drawdowns, making the method significantly efficient. By diversifying throughout completely different time home windows, the portfolio turns into much less delicate to remoted fluctuations and short-term noise, whereas nonetheless capturing significant market traits. This not solely smooths the fairness curve but additionally enhances the steadiness of risk-adjusted returns, reinforcing the sensible worth of the multi-window change technique.
Improve of defensive a part of portfolio
Having Client Staples (XLP) because the defensive element is definitely helpful, however it might not present enough diversification by itself. Each Utilities (XLU) and Well being Care (XLV) characterize sectors which might be traditionally resilient throughout financial downturns. Utilities have a tendency to supply regular money flows and are much less delicate to shopper spending cycles, whereas Well being Care advantages from constant demand for medical companies and merchandise, impartial of broader financial situations. Together with these sectors alongside XLP broadens the defensive publicity and reduces the danger of counting on a single sector for defense.
To implement this, we exchange the only XLP allocation with an evenly weighted defensive portfolio consisting of XLU, XLV, and XLP (marked as XLUVP in graphs and tables).
Desk 4: Efficiency metrics of XLY – XLUVP change methods.
PORTFOLIO
CAR p.a.
VOL p.a.
MAX DD
SHARPE
CALMAR
XLY portfolio
9.70%
19.24%
-54.93%
0.5
0.18
XLUVP portfolio
7.96%
11.38%
-33.45%
0.70
0.24
3M XLY – XLUVP change portfolio
10.83%
15.62%
-34.29%
0.69
0.32
6M XLY – XLUVP change portfolio
11.67%
15.77%
-36.79%
0.74
0.32
9M XLY – XLUVP change portfolio
11.70%
15.16%
-33.45%
0.77
0.35
12M XLY – XLUVP change portfolio
11.21%
14.27%
-35.14%
0.79
0.32
24M XLY – XLUVP change portfolio
10.34%
13.81%
-33.45%
0.75
0.31
36M XLY – XLUVP change portfolio
8.50%
13.06%
-33.45%
0.65
0.25
The introduction of a defensive combine composed of XLP, XLU, and XLV has noticeably improved the efficiency metrics. We attribute this enchancment to 2 components. First, the mix of those three sectors inherently displays stronger risk-adjusted traits in comparison with any single element. Second, combining a number of defensive property usually reduces total portfolio threat, smoothing returns and reducing drawdowns.
Constructing on this precept, it’s pure to increase the identical multi-window aggregation method that we beforehand utilized to XLP alone. By averaging indicators throughout completely different moving-average home windows for the mixed defensive portfolio, we intention to additional cut back volatility and improve stability. This enables the change technique to learn concurrently from each sector diversification and time-window diversification, doubtlessly producing much more sturdy risk-adjusted outcomes.
Desk 5: Efficiency metrics of mixing XLY – XLUVP change methods.
PORTFOLIO
CAR p.a.
VOL p.a.
MAX DD
SHARPE
CALMAR
3+6+9+12M XLY – XLUVP change portfolio
11.51%
14.23%
-31.83%
0.81
0.36
3+6+9+12+24+36M XLY – XLUVP change portfolio
10.93%
13.11%
-31.43%
0.83
035
As soon as once more, the outcomes verify that the defensive mixture of XLP, XLU, and XLV outperforms a portfolio utilizing solely XLP on the defensive aspect. The mix gives stronger risk-adjusted metrics, reduces drawdowns, and delivers a extra steady return profile, demonstrating the advantages of each sector diversification and multi-window sign aggregation throughout the change technique.
Does this method work for particular person shares as nicely?
A brand new query naturally arises from our sector-level outcomes. The change technique works exceptionally nicely with sector ETFs, which is encouraging, however its utility might be additional enhanced if it have been relevant to chose particular person shares. To discover this, we flip our consideration to Amazon (AMZN), one of many largest customers of packaging supplies tracked by our indicator. By making use of the identical multi-window, switch-based method, we intention to research whether or not the technique can seize comparable risk-adjusted enhancements and generate significant indicators on the single-stock degree.
Desk 6: Efficiency metrics of AMZN change methods.
PORTFOLIO
CAR p.a.
VOL p.a.
MAX DD
SHARPE
CALMAR
AMZN portfolio
20.72%
38.79%
-86.04%
0.53
0.24
3M AMZN change portfolio
21.67%
27.81%
-60.75%
0.78
0.36
6M AMZN change portfolio
18.18%
26.52%
-64.72%
0.69
0.28
9M AMZN change portfolio
20.14%
26.57%
-52.66%
0.76
0.38
12M AMZN change portfolio
17.61%
22.89%
-51.40%
0.77
0.34
24M AMZN change portfolio
17.94%
18.00%
-22.72%
1.00
0.79
36M AMZN change portfolio
9.01%
12.95%
-19.40%
0.70
0.46
We observe that as we lengthen the moving-average window within the MA mannequin, the uncooked returns have a tendency to say no. Whereas risk-adjusted metrics enhance barely, the discount in absolute efficiency is a priority. Furthermore, longer home windows usually end in extended durations of inactivity, throughout which the technique merely holds no place. To deal with this, we take into account integrating BIL, a short-term Treasury ETF, into the portfolio. By allocating idle money to BIL, the technique can generate no less than modest returns throughout inactive durations, enhancing total effectivity with out materially growing threat. This adjustment permits the mannequin to stay conservative when indicators are weak whereas nonetheless placing idle capital to productive use.
Altering the statement interval led to a discount in uncooked returns, nevertheless it additionally often improved sure risk-adjusted metrics.
Desk 7: Efficiency metrics of AMZN – BIL change methods.
PORTFOLIO
CAR p.a.
VOL p.a.
MAX DD
SHARPE
CALMAR
AMZN portfolio
25.64%
39.15%
-51.92%
0.65
0.49
3M AMZN – BIL change portfolio
19.68%
21.24%
-34.10%
0.93
0.58
6M AMZN – BIL change portfolio
17.80%
20.66%
-22.52%
0.86
0.79
9M AMZN – BIL change portfolio
15.73%
19.30%
-31.57%
0.81
0.50
12M AMZN – BIL change portfolio
14.46%
17.54%
-29.63%
0.82
0.49
24M AMZN – BIL change portfolio
11.60%
14.25%
-19.58%
0.81
0.59
36M AMZN – BIL change portfolio
6.57%
10.06%
-12.01%
0.65
0.55
We are able to additionally take into account whether or not combining a number of window lengths provides worth. To discover this, we revisit the mixed-horizon method used earlier and check combos of three–12 month and three–36 month MA-based methods.
Desk 8: Efficiency metrics of mixing AMZN – BIL change methods.
PORTFOLIO
CAR p.a.
VOL p.a.
MAX DD
SHARPE
CALMAR
3+6+9+12M AMZN – BIL change portfolio
17.23%
18.07%
-25.33%
0.95
0.68
3+6+9+12+24+36M AMZN – BIL change portfolio
14.69%
14.82%
-19.27%
0.99
0.76
As soon as once more, the outcomes present that combining a number of indicators produces a extra steady and balanced final result. The nominal return decreases, however the total conduct of the technique turns into extra disciplined, much less risky and extra resilient in tough market durations. Mixing completely different window lengths constantly improves risk-adjusted efficiency, which will be extra beneficial than purely maximizing uncooked returns.
Abstract and conclusion
Over the course of our evaluation, we experimented with a variety of approaches constructed across the thought of utilizing the corrugated-box PPI as a substitute market sign. We evaluated AR and MA fashions, completely different window lengths, single-indicator methods and blended multi-window programs. The clearest conclusion is that this system works most naturally with sector ETFs, the place the indicator gives a significant set off for rotating between offensive property reminiscent of XLY and defensive property reminiscent of XLP, XLU or XLV. Sector-level dynamics react extra visibly to modifications in underlying financial exercise, which makes the switching mechanism each intuitive and efficient.
When making use of the identical logic to particular person equities, the outcomes weren’t as clear. The indicator doesn’t translate as instantly into firm-level conduct, which implies that whereas nominal returns usually decline, we will nonetheless obtain enhancements in risk-adjusted metrics like Sharpe and Calmar. This comes at a noticeable price in uncooked efficiency, nevertheless it demonstrates that the core thought retains some worth even in a much less favorable setting.
Lastly, it’s price emphasizing that this PPI sequence is just one instance inside a wider universe of different indicators. Many unconventional macro or micro-level metrics could carry predictive construction that conventional price-based indicators fail to seize. Exploring these sources systematically can reveal new views on market conduct and doubtlessly uncover sturdy determination frameworks for lively allocation.
Writer:
David Belobrad, Junior Quant Analyst, Quantpedia
? Then utilise our Quantpedia’s Black Friday Deal and get extra quant insights within the yr 2026! You’ve final 24 hours! Solely yearly, Quantpedia’s Black Friday Deal presents you 33% extra of our companies for a similar value.
Purchase Quantpedia Prime now and get:3 + 1 months subscription for $29912 + 4 months subscription for $39936 + 12 months subscription for $699
Purchase Quantpedia Premium now and get:3 + 1 months subscription for $44912 + 4 months subscription for $59936 + 12 months subscription for $1199
Or purchase Quantpedia Professional and get:3 + 1 months subscription for $59912 + 4 months subscription for $79936 + 12 months subscription for $1599
Are you on the lookout for extra methods to examine? Join our e-newsletter or go to our Weblog or Screener.
Do you wish to be taught extra about Quantpedia Premium service? Test how Quantpedia works, our mission and Premium pricing provide.
Do you wish to be taught extra about Quantpedia Professional service? Test its description, watch movies, assessment reporting capabilities and go to our pricing provide.
Are you on the lookout for historic knowledge or backtesting platforms? Test our listing of Algo Buying and selling Reductions.
Would you want free entry to our companies? Then, open an account with Lightspeed and revel in one yr of Quantpedia Premium without charge.
Or comply with us on:
Fb Group, Fb Web page, Twitter, Linkedin, Medium or Youtube
Share onLinkedInTwitterFacebookCheck with a pal
