BOJ and the potential bigger switch in investment flows

The ultra-low rates and loose monetary policy of the Bank of Japan has had a tremendous impact on global markets. Money will seek higher returns around the world, so Japanese cash moved offshore to cash in on the gains.

Of course, the real story is more complex. Japanese investors will sell foreign assets when the yen depreciates, global yields rise, and the cost of hedge increase. The cost of hedging is high given the inverted US curve. Yields have risen over the last year, and the yen has seen a round of depreciation in 2023. Last year Japan was a net seller of foreign assets. Now, we have deal with BOJ adjustments that may add more selling pressure. Higher rates in Japan will inhibit foreign buying of assets.

The change in monetary policy last week changes the landscape albeit in the short run we will see more ambiguity. The central bank kept its formal target for the 10-year yield at close to 0, while saying that the .5% ceiling will become a reference point and not a rigid cap. The BOJ will offer to buy 10-year debt at 1% which sounds like the new cap that effectively doubles the range for 10-year bonds. 

Nevertheless, the BOJ intervened in the bond market today to stop bonds from moving higher. The range has been increased, but the central bank does not want the market to trade at the high end of the range. The central bank communication is ambiguous. Policy is changing, but the goal is not clear.

Inflation transitory story - it just took longer

The Fed's preferred measure of inflation, the PCE price deflator, continues to move lower with both the headline and ex food and energy indices showing a strong downtrend. The transitory story seems to make sense albeit it took longer than expected. The supply shock disruptions from the pandemic are done. The demand shock from the excessive stimulus also looks to be done. 

The decline in inflation was achieved to some degree in spite of the Fed increasing rates. It is not clear, at this time, that financial conditions have been tightened to offset demand and create a recession. Financial condition indices are not indicating overly tight conditions.

The question is whether inflation will get to the 2% target in the near-term and how much more does the Fed have to raise rates to help that process. The last mile is always the toughest. 

Will a recession occur? It is likely, but the story of a soft landing is gaining more followers and there is little evidence in the near-term of a recession. The recession story is being pushed into 2024. 

 

Bull markets are born on pessimism, grow on skepticism, mature on optimism and die on euphoria. 

- John Templeton

Great quote and everyone will likely agree with it, but what does it mean as a predictive tool. A quant is always looking for some indicator or number that can be used to describe the state of the market. How do we measure these terms: pessimism, skepticism, optimism, and euphoria?  We could use survey information, but it depends on the question. We are forced to find proxies and those are not easy to create.  Quotes are like candy - taste good but empty calories.

BlackRock return assumptions tells us about uncertainty

 

There are traders and investors. Traders take short-term risks and look for changes that will impact stocks from one day to at most a few weeks. Investors hold assets for months to years. Investors can change their allocations over short horizons, but their objective is to capture long-term premia and returns. They must think about the span of returns over years not weeks. 

The BlackRock asset assumption graph provides a good range of what type of uncertainty investors will face both for the upside and the downside. See BlackRock Investment Institute capital markets assumptions. Private equity and debt have a lot of return potential, but the risk is much greater than liquid equity and debt markets. You may get paid for the upside, but you may also lock-in downside returns. 

Don't just think about the mean return. Explore the range of returns and what will happen to your portfolio if you fall within the lower tail.  

Behavioral bias - They just will not go away

It has been 40 years since the great explosion of research on documenting behavioral biases in economics and finance. The list is extensive. There was no one single research event that pushed the change in thinking from efficient markets and super rationality to a more nuanced view concerning biases. It started as a crack in the existing paradigm and just kept getting bigger as an important area of research as more biases were documented. 

By this time, one should expect that all the biases should have bene stamped out. Nothing to see here, we got this under control. Yet, the biases still exist. We still make mistakes. The problem has not been solved.

So, what is it about human nature that makes it so easy for even very smart people to follow their biases. I don't know. How do we change behavior? 

Perhaps it just is not possible to make these changes. Biases are a part of being human. The only solution is to turnover your decisions to a set of rules that will eliminate the biases. I cannot help myself, so I am going to have to have an outside force bind me to good behavior. This is the foundation of quant investing - forcing good behavior on fallible humans.

Luxury watch prices declining

Economists are always looking for new data to tell a deeper story on what is happening to the economy. An interesting new series is an index for luxury watches formed by watchcharts.com. 

I cannot say how much stock to put into this index given it is less than 5 years old, but it does tell a story of high luxury spending peaking in the spring of 2022 with a strong decline over the last year. The index is on a down trend albeit still higher than January 2021. Nevertheless, it looks like flashy spending is declining and this is something to watch in the coming months.  

Thick versus thin modeling - An important choice

 

You want to get your prediction model right but that is not an easy task. The usual approach is to formulate a single model or specification and then use estimation techniques to generate forecasts. Nevertheless, we know that when we add or combine different forecasts, we will be able to get a better average will lower error or uncertainty. Simply put, we can form a number of models that may represent different views or hypothesis for specification and average to get a better result. This was developed years ago and given the name of thick modeling.  See "Thick Modeling" by the great econometric thinker Clive Granger. A single specification can be viewed as a thin model. A combination of models is a thick modeling approach. 

Thick model has been exploited in data science through several machine learning techniques. For example, a random forest approach can be viewed as a form of thick modeling.  Choosing a single thin specification may throw-out information from other specification which can be useful with forecasting. Instead of having a goal of forming a single specification, it can be helpful to use several specifications to obtain more information. Each specification may have slightly different features or x variables and thus will have slightly different parameter weights which may prove useful in certain regimes or environments. 

Use all information, take an ensemble approach, and form thick models. 

China - US relationship framework

 

This is a very good simple framework for thinking about US-China relationship. Overtime we may move between quadrants. While we are now in competitive confrontation, we may move to fortress America or switch to constructive competition or passive re-engagement. These changes will impact global growth as well as China and US growth. 

As we move to an election year, positions may harden to maintain the status quo. A switch in policy direction will be viewed as risky.

Saving rate is lower and this is a macro drag

 

The savings rate exploded to the upside during the pandemic. The first savings spike occurred when expenditures were cut because businesses were not open in 2020. The second spike was associated with further stimulus in 2021; however, the excess savings is ending. The savings rate is now below average and the amount of savings, while growing in 2023 is depleted to levels below 2018 and 2019. There is no extra money available for added spending, so future growth is likely to be lower. This is another reason for a soft landing even with a strong labor market. 

The big switch into risky assets? from bills to stocks

 

 Bank of America tracks the flows of client money and found that this week has shown a record flow out of T-bills. Forget about the safe asset. It seems like private clients are saying that it is time for holding risk-on assets. 

Inflation surprised to the downside. Recession fears have been reduced. The chances of further Fed hikes have also been diminished, so the word out; get into riskier assets. Flows suggest that investors are putting their money to work in equities, but it may not be in the high performing overvalued stocks. A switch from caution to risk-taking will disrupt the market behavior of the last month and can explain some of the recent short covering and movement to higher net long positions by active managers and hedge funds. 

Finding the optimal model - like Goldilocks

 “An optimal model is a ‘Goldilocks’ model. It is large enough that it can reliably detect potentially complex predictive relationships in the data, but not so flexible that it is dominated by overfit and suffers out-of-sample.” 

- S.B. Gu discussing machine learning and econometrics

A great way to think about a model is to use the Goldilocks analogy.  A model should not be too complex or too simple. Too complex with too many variables and we have an overfit problem. Too few variables and we are left with a problem that an unspecified factor will drive predictions. While many would like to believe that building models is a science, there is a lot art to finding the right model. A model will have the personality of the author. 

Concentration problem in key industries - This is to going to be easily solved

There has been growing talk about derisking and or decoupling trade with China as if this is something like a switch to either turn-on or turn-off. The global economy does not work that way. Supply chains are not easy to change. Capital has been committed across borders. Industries have developed around geographic locations for the simple reason that resources are in certain regions of the world. This is especially true with production for some key security technologies. 

The US and the world are especially sensitive to trade with China and the idea that we can decouple or derrick in the next year or two or five is just fantasy. These industries will be a major stress point for markets over the next half decade.

Transaction costs matter - So cost mitigation strategies should be employed

Transaction costs matter and are often not given enough focus when looking at strategy comparisons. A key reason for the decline in performance between backtested results and live returns is associated with not properly accounting for transaction costs. Of course, this is part of the larger issue that backtested numbers do not often match live performance. Nevertheless, we will focus on the issue of transaction costs.

To address this issue, investors need to focus on two questions: 1. what are at the true transaction costs of the strategy? and 2. what are the transaction cost mitigation strategies that can employed to reduce costs? Overall, transaction costs will lead to lower net Sharpe ratios, but an appropriate accounting of cost will lead to better live strategy performance.

There are three factors that impact performance from backtested results, model drag, structural issues, and transaction costs. Model drag is a function of the sample tested versus the live market regime. Structural issues include the ability to short and the cost of borrowing. Transaction costs are the bid-ask spread and liquidity concerns. 

Cost mitigation can be divided into four different strategies: rebalancing, banding, liquidity focus, and screening.

A good paper on how to deal with transactions is "Comparing Cost-Mitigation Techniques". This paper looks at each of the four major cost mitigation strategies and finds they have variable ability to boost returns. In general, using banding or holding position longer before action is the most appropriate cost mitigation strategy. However, the appropriate cost mitigation is a function of the type of strategy employed and the universe of stocks used for a given strategy. These cost mitigation strategies have less impact on large cap stocks while having a strong impact on micro caps. 

Using screening techniques will have a strong impact on returns by filtering which stocks to add to the portfolio and cut costs. For example, the size effect tied with momentum will have a strong impact on returns. Match a strategy with a filter will support better Sharpe ratio.

One of the key concerns when investing in new quant managers is the expected difference between backtested results and live returns. Addressing transaction costs is an important technique for reducing the difference between this potential drop in performance. How cost mitigation strategies are employed is one way to differentiate between managers.

Family offices plan to reduce cash

 

A recent survey from Goldman Sachs suggests that family offices plan to reduce their exposure to cash and cash equivalents over the next 12 months although nominal rates are now above 5%. Forget about the risk-free rate. Family offices are planning to increase exposures to public and private market equities as well as fixed income. Regardless of the tightening of monetary policy, families want to take on more risk and believe there will be better risk opportunities. There is always a lot to unpack with survey data, but this is not what one would expect if there is to be a slowdown in growth or a possible recession. 

How do we detect skill? An ongoing problem in finance

 

The age-old question for any investor - how do you detect skill in a trader or strategy? It is not easy, but you can go back to simple principles of gain and loss. Instead of focusing on the Sharpe ratio, the emphasis may be better placed on the actuarial concepts of frequency and magnitude. A good test of skill is answering two questions: 1. how often does the investment or strategy make money? and 2. how much does it gain (loss) when it is correct (wrong), that is, what is the probability of gain and what is the conditional expected gain or loss? See "Gains and Losses Revisited: Skill Detection and Similarity Assessment" from Sid Browne for a review and extension of these key concepts. 

The proportion of times a gain occurs can be viewed as the hit rate or the batting average for a trader. Of course, an investor wants to have trader working for him who has a high batting average. However, the second component can be considered the slugging percentage which is the ratio of the average gains to loss. There are now two skills that can be measured: 1. the timing skill which is your ability to get bets right and the sizing skill which is your ability to make more on your winning trades over your losing trades. This view of looking at the probability of gain and the gain versus loss is the foundation for key measures like the Kelly criterion. 

A key statistic for assessing trader is the Gain/Loss Ratio or GLR which is a product of two ratio: 1. the conditional expected gain versus conditional expected loss, and 2. the probability of a gain versus the probability of a loss. One ratio is the magnitude of gains versus loses while the other is the likelihood of gains versus loses.  To get a finer edge on the skill of traders, you can look more deeply than just the GLR but at the two components of the gain to loss ratio. 

Different strategies will have different hit rate and slugging percentages. For example, a carry strategy will likely have a higher hit rate but may have a lower profit to loss ratio because there may be small gains but outsized or concentrated losses. A trend strategy may have a lower hit rate but greater gains versus losses as profitable positions are held, and losses are cut.

The nice part of breaking trading skill into these two parts is that traders can be compared between backtested and life periods. Similarly, one trader can be compared to another to measure timing and magnitude skill. There are well formed test statistics for comparison. This approach will provide deeper insight on the type of skill any trader may possess. 

"Pockets of Predictability" - Markets are not always unpredictable

 

Markets are efficient. That is a good null hypothesis and a good basis for investing. Making money by trading is not easy. Investors will get a premium for taking on risk, but making any excess return is not easy. Most should just invest in passive indices. 

Yet, this general conclusion is not the end of the story.  There are opportunities in markets at specific times. What factors may be less unpredictable or marginal on average may actually have high predictability and make money during focused times. This idea changes how you think about trading and the importance of different factors. As recently put by in a Journal of Finance research paper, there are "Pockets of Predictability".  I love this term.

A given feature or predictor may not make significant money over a long period. The models may have limited significance and r-squared, but there may be selected out of sample periods when there is strong predictability and the opportunity to make money. Predictability may be associated with local or short periods of time. The researchers generate a simple way of testing predictability through time to find these pockets. These pockets are associated with a sticky expectations model where beliefs are slowly updated which allow some factors to make improved expectations.

A key issue is not identifying these pockets but being able to predict these pockets so capital can be husbanded for the right time. These pockets can be found in both daily and monthly data and are associated with simple predictors like the lagged divided price ratio, the yield on 3-month Treasury bills, the Treasury term spread between 10-year bonds and 3-month bills, and realized volatility.

Looking for or finding pockets of predictability is easy and not the secret to riches, but it changes our perspective on how we should look at models and opportunities. When expectations are sticky, a model may provide better insights at a specific or local time. Using a combination of factors may identify opportunities that can provide a slight edge. Predictive models can provide an alternative way of assessing the market environment that offer support over localized periods. This is worth exploring in more detail.