Attilio Meucci

QUICK ANALYSIS: What next for the #Mag7 given the macro shifts we have seen ❓ 👉 We noted a few weeks back in our Linkedin post that the Mag7 12 month Forward PE was 2 standard deviation *expensive* to the overall macro information set (RED CIRCLE) 🚨 Now it has swung to 2 standard deviations **cheap** to macro (GREEN CIRCLE) One can model the PE as a function of economic growth, inflation, nominal & real rates, credit spreads, energy prices, the Dollar and various other measures of investor risk aversion. The chart shows: 1️⃣ The Actual Fwd PE (white line, left scale) 2️⃣ The GAP between the Actual PE and the Macro Warranted PE in standard deviation terms Indeed, you can see this for 12,000 stocks, indices and sectors globally. This has been one enormous swing. The Fwd PE has gone from pricing in strong growth (no landing at all) to a crash landing #mag7 #apple #tesla #meta #alphabet #amazon #netflix #nvidia

22

8 Comments

Majority of allocators (LPs) trust their capital in Quant Investment Strategies that are based on back tested results What kind of bias does such allocators have ? Do they think that Quant Investment strategies are free from bias ? Are back tested results better than live portfolio simulation results ? The difference between back-tested results and a live portfolio simulation are : Back-test results are useful for evaluating the theoretical performance of a strategy based on historical data, but they are inherently limited by the assumptions made in the test, and they often fail to account for real-world constraints such as slippage, transaction costs, and changing market conditions. Live portfolio simulations provide a more accurate, real-time assessment of how a strategy performs in the actual market, including real-world frictions and challenges. However, like backtests, live simulations can also be imperfect and may diverge from real live performance due to unforeseen market factors.

2

Just posted what (I hope) is the final revision for An Analytic Solution for Asset Allocation with a Multivariate Laplace Distribution on SSRN here https://lnkd.in/eVQNNyuF This version includes a large-portfolio limit in which, very surprisingly to me, the Golden Ratio shows up. #portfolioselection #quantitativefinance #markets #economics #assetallocation #tradingstrategy

32

3 Comments

Since vol is back again - one chart and quite a bit of data visualized if this helped. First chart shows the SX5E. Second one are its 3m atm implied vols. Third chart is 3m 25d Skew. Fourth one is a 3m vs. 12m atm termstructure comparison. The bottom chart shows 10d, 30d and 60d realized vol. Additionally all indices with Bollinger Bands for context.

39

1 Comment

The last episode of the "Theory to Practice" series with Benjamin George, "Historical vs. Implied Volatility" is available now on YouTube. Tune in to learn the differences between historical volatility and implied volatility, and see a practical example in QuantConnect: https://lnkd.in/eSYBV6Nc #finance #trading #quantconnect #volatility

1

On the same topic of AD, to get a more general view, you may read the book Antoine Savine and Jesper Andreasen: https://lnkd.in/ekapAsA9

20

1 Comment

In this article, we’ll explore how to harness Numba for common Monte Carlo tasks like basic random sampling, parallel simulation, and Markov Chain Monte Carlo (MCMC) implementations.  #statology

3

Understanding Inverse Problems in Investment Analysis: From Theory to Practice #InverseProblem #QuantFinance #RiskAnalysis #FactorInvesting #DataScience #FinancialModeling #MarketAnalysis #AlgoTrading #PortfolioOptimization #ImpliedVolatility

1

Decent paper from the BIS describing how standard optimization for long-horizon portfolios commonly overstates returns or understates risk by relying on square root of time volatility adjustment. I've never seen an ALM that properly addresses this and this gets worse as you add more private asset classes. "It is a widespread practice to use daily or monthly data to design portfolios with investment horizons equal or greater than a year. The computation of the annualized mean return is carried out via traditional interest rate compounding – an assumption free procedure –, while scaling volatility is commonly fulfilled by relying on the serial independence of returns’ assumption, which results in the celebrated square-root-of-time rule. While it is a well-recognized fact that the serial independence assumption for asset returns is unrealistic at best, the convenience and robustness of the computation of the annual volatility for portfolio optimization based on the square-root-of-time rule remains largely uncontested. As expected, the greater the departure from the serial independence assumption, the larger the error resulting from this volatility scaling procedure. Based on a global set of risk factors, we compare a standard mean-variance portfolio optimization (eg square-root-of-time rule reliant) with an enhanced mean-variance method for avoiding the serial independence assumption. Differences between the resulting efficient frontiers are remarkable, and seem to increase as the investment horizon widens" https://lnkd.in/e3Ruq_4Q

22

5 Comments

Most cited paper "Equity Portfolio Diversification" by William N. Goetzmann and Alok Kumar in Review of Finance, Volume 12, Issue 3, 2008, Pages 433–463, https://lnkd.in/dD64A-s Published: 28 March 2008 Abstract This study shows that U.S. individual investors hold under-diversified portfolios, where the level of under-diversification is greater among younger, low-income, less-educated, and less-sophisticated investors. The level of under-diversification is also correlated with investment choices that are consistent with over-confidence, trend-following behavior, and local bias. Furthermore, investors who over-weight stocks with higher volatility and higher skewness are less diversified. In contrast, there is little evidence that portfolio size or transaction costs constrains diversification. Under-diversification is costly to most investors, but a small subset of investors under-diversify because of superior information. JEL G11 - Portfolio Choice; Investment DecisionsG12 - Asset Pricing; Trading volume; Bond Interest Rates Oxford University Press OUP Academic Yale School of Management Texas McCombs School of Business European Finance Association

57

1 Comment

"Gaussian Processes for Implied Volatility Estimation": "A Bayesian nonparametric method generates posterior distributions of implied volatility functions, producing confidence intervals and quantifiable measures of model uncertainty." https://lnkd.in/eNiK5VAv

88

1 Comment

There is no doubt the XVA topics are among the hardest ones in Quantitative Finance, it blends all its micro domains of application (Equity, FI, FX, Commodities) with the macro ones (Economic Capital, analysis at the macro scale of books, the bank, etc.) while being highly non linear in nature due to important codependances implied by wrong way risks. The development of its study is one of the many consequence of the 2008 crisis. On top of that, you also need to consider the model risk induced by this new modelling. The document below quoted by Gaby Frangieh is a good little journey in this domain. (I discover this piece thanks to Nicholas Burgess who is also a good one to follow :-))

105

4 Comments

In the world of quantitative finance, the Vasicek model is a landmark for modeling the dynamics of short-term interest rates. Created based on a mean-reversion process, it is widely used to predict rate behavior over time, assuming they tend to revert to a long-term average value, even when experiencing sharp short-term fluctuations. Despite its limitations, the Vasicek model remains an essential tool, especially for fixed-income instruments and derivative pricing, where a structured projection of short-term rates is crucial. #financialmodeling #investments #quant #financialmarkets

9

With the number of LLMs steadily increasing nowadays, how much time do we spend on preparing and formatting the input chat data? Each model has very different input formats for the same chat. Huggingface has provided a nice API (apply_chat_template) to simplify the preparation of model-specific input format from a "consensus" chat format. But, still, we are spending too much time remapping/reformatting data among all different formats. We are always nervous about the performance of finetuning/training when imperfect formats are accidentally used. To simplify this, all data should prepare a copy in consensus format, otherwise, it would be nice to have a reverse API to map model-specific formats to a consensus format, and then transform to other target formats. This could be a nice LLM input format adapter project. #LLM #datapreparation #format

29

Detailed thread on Understanding and Trading Vanna and Charm effect for trading🧵 Vanna and charm effects are important concepts in options trading and are closely related to the Greek metrics used to measure and manage risk https://lnkd.in/eTBNPNjt

13

Weekend post. For those who don't know: Peter Jaeckel found an analytical formula for implied normal volatility and wrote a paper about it.   https://lnkd.in/eMxsy9_Q   We use this in #OpenSourceRiskEngine in different contexts for quite some time. However, in some ways #QuantLib seems to be the better home for the little gem, so we moved it over and it is available in the latest QuantLib release 36. It replaces the Choi approximation as the default way to retrieve implied normal vols: https://lnkd.in/eBzuhkDf   It improves accuracy by some orders.

168

3 Comments