Qaasid News

Download Our App

Category: 3. Business

  • Why is the Japanese Yen falling?

    Why is the Japanese Yen falling?

    Back when I worked at Goldman, I remember one particular Fed meeting when the central bank had hiked but – to my complete consternation – the Dollar fell. I asked the head of currency trading at the time how this could have possibly happened. He looked me straight in the eye and said: “there were more sellers than buyers.”

    That’s pretty much what’s going on with the Yen right now. The Bank of Japan (BoJ) is hiking, but the Yen is down to its lowest level in over 20 years, tumbling below the low it made mid-2024 when Japanese interest rates were much lower. It might sound trite to attribute this to there being “more sellers than buyers,” but there’s a lot more wisdom in this comment than you might think. This post explains what’s going on.

    The black line in the chart above shows the trade-weighted Yen against the majors, where I use the same weights as the BoJ to average up bilateral currency pairs. The blue line shows the analogous interest rate differential based on 30-year government bond yields. As I’ve noted in many previous posts, longer-term Japanese yields have risen very sharply this year, which has moved the 30-year differential sharply in favor of the Yen. That should make it more attractive for global capital markets to invest in Japan and should therefore cause the Yen to appreciate. That isn’t happening, which might seem like a puzzle but it really isn’t.

    The vertical axis in the chart above shows the 30-year government bond yields that go into the rate differential in the first chart. The horizontal axis plots gross government debt in percent of GDP. While it’s true that the interest differential has moved a lot in favor of the Yen, it’s also true that – given Japan’s monstrous level of government debt – longer-term yields are still much too low relative to where they would be if the BoJ weren’t still a massive buyer of government debt. This bond buying is keeping yields artificially low, which should really be much higher due to risk premia. Because these risk premia aren’t allowed to show up in the bond market, they show up in the Yen instead, which is the reason it keeps falling.

    Here’s the uncomfortable truth: Japan’s longer-term yields have been rising, but – on a risk-adjusted basis – that rise isn’t nearly enough to stabilize the Yen. Another way to say this: markets think risk of a debt crisis is rising sharply. Yen depreciation won’t stop until yields are allowed to rise far more, forcing the government to pursue fiscal consolidation and bring down debt. Japan needs to stop being in denial.

    Continue Reading

  • A Powerball player in Arkansas has won a $1.817 billion lottery jackpot : NPR

    A Powerball player in Arkansas has won a $1.817 billion lottery jackpot : NPR

    The jackpot for the Powerball lottery game is displayed outside Ted’s State Line Mobil station, Wednesday, Dec. 24, 2025, in Methuen, Mass.

    Charles Krupa/AP


    hide caption

    toggle caption

    Charles Krupa/AP

    A Powerball player in Arkansas won a $1.817 billion jackpot in Wednesday’s Christmas Eve drawing, ending the lottery game’s three-month stretch without a top-prize winner.

    The winning numbers were 04, 25, 31, 52 and 59, with the Powerball number being 19.

    Final ticket sales pushed the jackpot higher than previous expected, making it the second-largest in U.S. history and the largest Powerball prize of 2025, according to www.powerball.com. The jackpot had a lump sum cash payment option of $834.9 million.

    “Congratulations to the newest Powerball jackpot winner! This is truly an extraordinary, life-changing prize,” Matt Strawn, Powerball Product Group Chair and Iowa Lottery CEO, was quoted as saying by the website. “We also want to thank all the players who joined in this jackpot streak — every ticket purchased helps support public programs and services across the country.”

    The prize followed 46 consecutive drawings in which no one matched all six numbers.

    The last drawing with a jackpot winner was Sept. 6, when players in Missouri and Texas won $1.787 billion.

    Organizers said it is the second time the Powerball jackpot has been won by a ticket sold in Arkansas. It first happened in 2010.

    The last time someone won a Powerball jackpot on Christmas Eve was in 2011, Powerball said. The company added that the sweepstakes also has been won on Christmas Day four times, most recently in 2013.

    Powerball’s odds of 1 in 292.2 million are designed to generate big jackpots, with prizes growing as they roll over when no one wins. Lottery officials note that the odds are far better for the game’s many smaller prizes.

    “With the prize so high, I just bought one kind of impulsively. Why not?” Indianapolis glass artist Chris Winters said Wednesday.

    Tickets cost $2, and the game is offered in 45 states plus Washington, D.C., Puerto Rico and the U.S. Virgin Islands.

    Continue Reading

  • Dalian iron ore extends gains on easier home buying in Beijing – Business Recorder

    1. Dalian iron ore extends gains on easier home buying in Beijing  Business Recorder
    2. MMi Daily Iron Ore Report (December 24)  Shanghai Metals Market
    3. Iron Ore Holds Rebound from 5-Month Low  TradingView — Track All Markets
    4. Dalian iron ore extends gains on tight BHP supply, firmer hot metal production  Mining.com
    5. Iron ore futures slip  Business Recorder

    Continue Reading

  • Inverse design of cellular structures with the targeted nonlinear mechanical response

  • Bezek, L. B. et al. Effect of part size, displacement rate, and aging on compressive properties of elastomeric parts of different unit cell topologies formed by vat photopolymerization additive manufacturing. Polymers 16, 3166 (2024).

    Google Scholar 

  • Yang, L. et al. Additive manufacturing of metal cellular structures: design and fabrication. Jom 67, 608–615 (2015).

    Google Scholar 

  • Lin, H. et al. 3d printing of porous ceramics for enhanced thermal insulation properties. Adv. Sci. 12, 2412554 (2025).

    Google Scholar 

  • Schaedler, T. A. et al. Designing metallic microlattices for energy absorber applications. Adv. Eng. Mater. 16, 276–283 (2014).

    Google Scholar 

  • Schaedler, T. A. & Carter, W. B. Architected cellular materials. Annual Rev. Mater. Res. 46, 187–210 (2016).

    Google Scholar 

  • Boursier Niutta, C., Ciardiello, R. & Tridello, A. Experimental and numerical investigation of a lattice structure for energy absorption: application to the design of an automotive crash absorber. Polymers 14, 1116 (2022).

    Google Scholar 

  • Mohsenizadeh, M., Gasbarri, F., Munther, M., Beheshti, A. & Davami, K. Additively-manufactured lightweight metamaterials for energy absorption. Mater. Des. 139, 521–530 (2018).

    Google Scholar 

  • Uribe-Lam, E., Treviño-Quintanilla, C. D., Cuan-Urquizo, E. & Olvera-Silva, O. Use of additive manufacturing for the fabrication of cellular and lattice materials: a review. Mater. Manuf. Process. 36, 257–280 (2021).

    Google Scholar 

  • Mueller, J., Raney, J. R., Shea, K. & Lewis, J. A. Architected lattices with high stiffness and toughness via multicore-shell 3d printing. Adv.Mater. 30, 1705001 (2018).

    Google Scholar 

  • Lei, H. et al. Evaluation of compressive properties of slm-fabricated multi-layer lattice structures by experimental test and \(\mu\)-ct-based finite element analysis. Materi. Des. 169, 107685 (2019).

    Google Scholar 

  • Kumar, A., Collini, L., Daurel, A. & Jeng, J.-Y. Design and additive manufacturing of closed cells from supportless lattice structure. Additive Manuf. 33, 101168 (2020).

    Google Scholar 

  • Nakarmi, S. et al. The role of unit cell topology in modulating the compaction response of additively manufactured cellular materials using simulations and validation experiments. Model. Simul. Mater. Sci. Eng. 32, 055029 (2024).

    Google Scholar 

  • Nakarmi, S. et al. Mesoscale simulations and validation experiments of polymer foam compaction-volume density effects. Mater. Lett. 382, 137864 (2025).

    Google Scholar 

  • Xia, L. & Breitkopf, P. Design of materials using topology optimization and energy-based homogenization approach in matlab. Struct. Multidisciplinary Optim. 52, 1229–1241 (2015).

    Google Scholar 

  • Radman, A., Huang, X. & Xie, Y. Topology optimization of functionally graded cellular materials. J. Mater. Sci. 48, 1503–1510 (2013).

    Google Scholar 

  • Bauer, J., Hengsbach, S., Tesari, I., Schwaiger, R. & Kraft, O. High-strength cellular ceramic composites with 3d microarchitecture. Procd. National Acad. Sci. 111, 2453–2458 (2014).

    Google Scholar 

  • Nguyen, J., Park, S.-I. & Rosen, D. Heuristic optimization method for cellular structure design of light weight components. Int. J. Precision Eng. Manuf. 14, 1071–1078 (2013).

    Google Scholar 

  • Meier, T. et al. Obtaining auxetic and isotropic metamaterials in counterintuitive design spaces: an automated optimization approach and experimental characterization. npj Comput. Mater. 10, 3 (2024).

    Google Scholar 

  • Vangelatos, Z. et al. Strength through defects: A novel bayesian approach for the optimization of architected materials. Sci. Adv. 7, eabk2218 (2021).

    Google Scholar 

  • Ramesh, A. et al. Zero-shot text-to-image generation. In International conference on machine learning, 8821–8831 (Pmlr, 2021).

  • Ramesh, A., Dhariwal, P., Nichol, A., Chu, C. & Chen, M. Hierarchical text-conditional image generation with clip latents. arXiv preprint arXiv:2204.061251, 3 (2022).

  • Yao, Z. et al. Inverse design of nanoporous crystalline reticular materials with deep generative models. Nat. Mach. Intell. 3, 76–86 (2021).

    Google Scholar 

  • Sanchez-Lengeling, B. & Aspuru-Guzik, A. Inverse molecular design using machine learning: Generative models for matter engineering. Science 361, 360–365 (2018).

    Google Scholar 

  • Zhavoronkov, A. et al. Deep learning enables rapid identification of potent ddr1 kinase inhibitors. Nat. Biotechnol. 37, 1038–1040 (2019).

    Google Scholar 

  • Liao, W., Lu, X., Fei, Y., Gu, Y. & Huang, Y. Generative ai design for building structures. Autom. Construct. 157, 105187 (2024).

    Google Scholar 

  • Kingma, D. P., Welling, M. et al. Auto-encoding variational bayes (2013).

  • Goodfellow, I. J. et al. Generative adversarial nets. Adv. Neural Inf. Process. Syst. 27 (2014).

  • Ho, J., Jain, A. & Abbeel, P. Denoising diffusion probabilistic models. Adv. Neural Inf. Process. Syst. 33, 6840–6851 (2020).

    Google Scholar 

  • Sohn, K., Lee, H. & Yan, X. Learning structured output representation using deep conditional generative models. Adv. Neural Inf. Process. Syst. 28 (2015).

  • Mirza, M. & Osindero, S. Conditional generative adversarial nets. arXiv preprint arXiv:1411.1784 (2014).

  • Dhariwal, P. & Nichol, A. Diffusion models beat gans on image synthesis. Adv. Neural Inf. Process. Syst. 34, 8780–8794 (2021).

    Google Scholar 

  • Lee, D., Chen, W., Wang, L., Chan, Y.-C. & Chen, W. Data-driven design for metamaterials and multiscale systems: a review. Adv. Mater. 36, 2305254 (2024).

    Google Scholar 

  • Zheng, X., Zhang, X., Chen, T.-T. & Watanabe, I. Deep learning in mechanical metamaterials: from prediction and generation to inverse design. Adv. Mater. 35, 2302530 (2023).

    Google Scholar 

  • Wang, L. et al. Deep generative modeling for mechanistic-based learning and design of metamaterial systems. Comput. Methods App. Mech. Eng. 372, 113377 (2020).

    Google Scholar 

  • Zheng, L., Karapiperis, K., Kumar, S. & Kochmann, D. M. Unifying the design space and optimizing linear and nonlinear truss metamaterials by generative modeling. Nat. Commun. 14, 7563 (2023).

    Google Scholar 

  • Tian, J., Tang, K., Chen, X. & Wang, X. Machine learning-based prediction and inverse design of 2d metamaterial structures with tunable deformation-dependent poisson’s ratio. Nanoscale 14, 12677–12691 (2022).

    Google Scholar 

  • Zheng, X., Chen, T.-T., Guo, X., Samitsu, S. & Watanabe, I. Controllable inverse design of auxetic metamaterials using deep learning. Mater. Des. 211, 110178 (2021).

    Google Scholar 

  • Challapalli, A., Patel, D. & Li, G. Inverse machine learning framework for optimizing lightweight metamaterials. Materi. Des. 208, 109937 (2021).

    Google Scholar 

  • Vlassis, N. N. & Sun, W. Denoising diffusion algorithm for inverse design of microstructures with fine-tuned nonlinear material properties. Comput. Methods Appl. Mech. Eng. 413, 116126 (2023).

    Google Scholar 

  • Bastek, J.-H. & Kochmann, D. M. Inverse design of nonlinear mechanical metamaterials via video denoising diffusion models. Nat. Mach. Intell. 5, 1466–1475 (2023).

    Google Scholar 

  • Meier, T. et al. Scalable phononic metamaterials: Tunable bandgap design and multi-scale experimental validation. Mater. Des. 252, 113778 (2025).

    Google Scholar 

  • Kumar, S., Tan, S., Zheng, L. & Kochmann, D. M. Inverse-designed spinodoid metamaterials. npj Comput. Mater. 6, 73 (2020).

    Google Scholar 

  • Nakarmi, S., Leiding, J. A., Lee, K.-S. & Daphalapurkar, N. P. Predicting non-linear stress-strain response of mesostructured cellular materials using supervised autoencoder. Comput. Methods Appl. Mech. Eng. 432, 117372 (2024).

    Google Scholar 

  • McNeel, R. et al. Grasshopper-algorithmic modeling for rhino. http://www.grasshopper3d.com (2013).

  • Dassault Systèmes. Abaqus Analysis User’s Manual, Version 2020 (2020).

  • Mooney, M. A theory of large elastic deformation. J. Appl. Phys. 11, 582–592 (1940).

    Google Scholar 

  • Rivlin, R. Large elastic deformations of isotropic materials. i. fundamental concepts. Philosophical Trans. Royal Soc. London. Series A, Math. Phys. Sci. 240, 459–490 (1948).

    Google Scholar 

  • Abdi, H. & Williams, L. J. Principal component analysis. Wiley Interdisciplinary Rev. Comput. Statist. 2, 433–459 (2010).

    Google Scholar 

  • Yang, C., Kim, Y., Ryu, S. & Gu, G. X. Prediction of composite microstructure stress-strain curves using convolutional neural networks. Mater. Des. 189, 108509 (2020).

    Google Scholar 

  • Ioffe, S. & Szegedy, C. Batch normalization: Accelerating deep network training by reducing internal covariate shift. In Int. Conference Mach. Learn., 448–456 (pmlr, 2015).

  • Li, X., Chen, S., Hu, X. & Yang, J. Understanding the disharmony between dropout and batch normalization by variance shift. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2682–2690 (2019).

  • Kullback, S. & Leibler, R. A. On information and sufficiency. Annals Math. Statist. 22, 79–86 (1951).

    Google Scholar 

  • Higgins, I. et al. Early visual concept learning with unsupervised deep learning. arXiv preprint arXiv:1606.05579 (2016).

  • Fu, H. et al. Cyclical annealing schedule: A simple approach to mitigating kl vanishing. arXiv preprint arXiv:1903.10145 (2019).

  • Smith, S. L., Kindermans, P.-J., Ying, C. & Le, Q. V. Don’t decay the learning rate, increase the batch size. arXiv preprint arXiv:1711.00489 (2017).

  • Liu, Y., Neophytou, A., Sengupta, S. & Sommerlade, E. Relighting images in the wild with a self-supervised siamese auto-encoder. In Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision, 32–40 (2021).

  • Wang, Z., Bovik, A. C., Sheikh, H. R. & Simoncelli, E. P. Image quality assessment: from error visibility to structural similarity. IEEE Trans. Image Process. 13, 600–612 (2004).

    Google Scholar 

  • Dice, L. R. Measures of the amount of ecologic association between species. Ecology 26, 297–302 (1945).

    Google Scholar 

  • Zhao, F., Huang, Q. & Gao, W. Image matching by normalized cross-correlation. In 2006 IEEE International Conference on Acoustics Speech and Signal Processing Proceedings, vol. 2, II–II (IEEE, 2006).

Continue Reading

  • Systematic hyperparameter analysis of GRU and LSTM across demand pattern types: a demand-characteristic-driven meta-learning framework for rapid optimization

  • Soto-Ferrari, M., Bhattacharyya, K., Schikora, P. & POST-BaLSTM: A bagged LSTM forecasting ensemble embedded with a postponement framework to target the semiconductor shortage in the automotive industry. Comput. Ind. Eng. 185, 109602 (2023).

    Google Scholar 

  • Lee, K. H., Abdollahian, M., Schreider, S. & Taheri, S. Supply chain demand forecasting and price optimisation models with substitution effect. Mathematics 11, 1–28 (2023).

    Google Scholar 

  • Wang, C. H. & Chen, T. Y. Combining biased regression with machine learning to conduct supply chain forecasting and analytics for printing circuit board. Int. J. Syst. Sci. Oper. Logist. 9, 143–154 (2022).

    Google Scholar 

  • Huber, J. & Stuckenschmidt, H. Daily retail demand forecasting using machine learning with emphasis on calendric special days. Int. J. Forecast. 36, 1420–1438 (2020).

    Google Scholar 

  • Weng, T., Liu, W. & Xiao, J. Supply chain sales forecasting based on LightGBM and LSTM combination model. Ind. Manag Data Syst. 120, 265–279 (2020).

    Google Scholar 

  • Omar, H., Klibi, W., Babai, M. Z. & Ducq, Y. Basket data-driven approach for omnichannel demand forecasting. Int. J. Prod. Econ. 257, 108748 (2023).

    Google Scholar 

  • Panda, S. K. & Mohanty, S. N. Time series forecasting and modeling of food demand supply chain based on regressors analysis. IEEE Access. 11, 42679–42700 (2023).

    Google Scholar 

  • Noh, J., Park, H. J., Kim, J. S. & Hwang, S. J. Gated recurrent unit with genetic algorithm for product demand forecasting in supply chain management. Mathematics 8, (2020).

  • Li, K. et al. Capacity and output power Estimation approach of individual behind-the-meter distributed photovoltaic system for demand response baseline Estimation. Appl. Energy. 253, 113595 (2019).

    Google Scholar 

  • Li, G. & Zhong, X. Parking demand forecasting based on improved complete ensemble empirical mode decomposition and GRU model. Eng. Appl. Artif. Intell. 119, 105717 (2023).

    Google Scholar 

  • Kim, Y. & Park, K. Outlier-Aware demand prediction using recurrent neural Network-Based models and statistical approach. IEEE Access. 11, 129285–129299 (2023).

    Google Scholar 

  • Chung, J., Gulcehre, C., Cho, K. & Bengio, Y. Empirical Evaluation of Gated Recurrent Neural Networks on Sequence Modeling. 1–9. At http://arxiv.org/abs/1412.3555 (2014)

  • Saeed, N., Nguyen, S., Cullinane, K., Gekara, V. & Chhetri, P. Forecasting container freight rates using the prophet forecasting method. Transp. Policy. 133, 86–107 (2023).

    Google Scholar 

  • Bommidi, B. S., Teeparthi, K. & Kosana, V. Hybrid wind speed forecasting using ICEEMDAN and transformer model with novel loss function. Energy 265, 126383 (2023).

    Google Scholar 

  • Tian, Z., Liu, W., Jiang, W. & Wu, C. CNNs-Transformer based day-ahead probabilistic load forecasting for weekends with limited data availability. Energy 293, 127019 (2024).

    Google Scholar 

  • Zhou, H. et al. Beyond Efficient Transformer for Long Sequence Time-Series Forecasting. (2021).

  • Zhou, T., Wen, Q., Wang, X., Sun, L. & Jin, R. FEDformer: Frequency Enhanced Decomposed Transformer for Long-term Series Forecasting. (2022).

  • Fan, H. Enhancing Long-Term time series forecasting via hybrid DLinear-PatchTST ensemble framework. Appl. Comput. Eng. 0, 145–150 (2025).

    Google Scholar 

  • Lin, P. et al. Multi-timescale short-term urban water demand forecasting based on an improved PatchTST model. J. Hydrol. 651, 132599 (2025).

    Google Scholar 

  • Ghimire, S., Deo, R. C. & Casillas-pérez, D. Salcedo-sanz, S. Electricity demand uncertainty modeling with Temporal Convolution neural network models. Renew. Sustain. Energy Rev. 209, 115097 (2025).

    Google Scholar 

  • Sun, Y., Ding, J., Liu, Z. & Wang, J. Combined forecasting tool for renewable energy management in sustainable supply chains. Comput. Ind. Eng. 179, 109237 (2023).

    Google Scholar 

  • Bischl, B. et al. Hyperparameter optimization: Foundations, algorithms, best practices, and open challenges. Wiley Interdiscip Rev. Data Min. Knowl. Discov. 13, 1–43 (2023).

    Google Scholar 

  • Wojciuk, M., Swiderska-Chadaj, Z., Siwek, K. & Gertych, A. Improving classification accuracy of fine-tuned CNN models: impact of hyperparameter optimization. Heliyon 10, e26586 (2024).

    Google Scholar 

  • Valencia, C. H., Vellasco, M. M. B. R. & Figueiredo, K. Echo state networks: novel reservoir selection and hyperparameter optimization model for time series forecasting. Neurocomputing 545, 126317 (2023).

    Google Scholar 

  • Dhake, H., Kashyap, Y. & Kosmopoulos, P. Algorithms for hyperparameter tuning of LSTMs for time series forecasting. Remote Sens. 15, 1–17 (2023).

    Google Scholar 

  • Wu, X. et al. AutoCTS+: Joint Neural Architecture and Hyperparameter Search for Correlated Time Series Forecasting. Proc. ACM Manag. Data 1Association for Computing Machinery, (2023).

  • Pranolo, A., Mao, Y., Wibawa, A. P., Utama, A. B. P. & Dwiyanto, F. A. Robust LSTM with Tuned-PSO and Bifold-Attention mechanism for analyzing multivariate Time-Series. IEEE Access. 10, 78423–78434 (2022).

    Google Scholar 

  • Chen, X. et al. A novel loss function of deep learning in wind speed forecasting. Energy 238, 121808 (2022).

    Google Scholar 

  • Nussipova, F., Rysbekov, S., Abdiakhmetova, Z. & Kartbayev, A. Optimizing loss functions for improved energy demand prediction in smart power grids. Int. J. Electr. Comput. Eng. 14, 3415–3426 (2024).

    Google Scholar 

  • Kenaka, S. P., Cakravastia, A., Ma’ruf, A. & Cahyono, R. T. Enhancing intermittent spare part demand forecasting: A novel ensemble approach with focal loss and SMOTE. Logistics 9, 1–25 (2025).

    Google Scholar 

  • Syntetos, A. A., Boylan, J. E. & Croston, J. D. On the categorization of demand patterns. J. Oper. Res. Soc. 56, 495–503 (2005).

    Google Scholar 

  • Rožanec, J. M., Kaži, B., Škrjanc, M., Fortuna, B. Automotive-OEM-demand-forecasting-A-comparative-study-of-forecasting-algorithms-and-strategiesApplied-Sciences-Switzerland (1).pdf (2021).

  • Szilagyi, E. et al. Cost-effective energy management of an islanded microgrid. Energy Rep. 10, 4516–4537 (2023).

    Google Scholar 

  • Huskova, K. & Dyntar, J. Increasing efficiency in inventory control of products with sporadic demand using simulation. Acta Inf. Pragensia. 11, 254–264 (2022).

    Google Scholar 

  • Hasan, N., Ahmed, N. & Ali, S. M. Improving sporadic demand forecasting using a modified k-nearest neighbor framework. Eng. Appl. Artif. Intell. 129, 107900 (2024).

    Google Scholar 

  • Zhang, Q. & Zhou, X. Assessing peak-to-mean ratios of odour intensity in the atmosphere near swine operations. Atmosphere (Basel). 11, 1102 (2020).

    Google Scholar 

  • Barry, P. J. A note on peak-to-mean concentration ratios. Boundary-Layer Meteorol. 2, 122–126 (1971).

    Google Scholar 

  • Wunderlich, A. & Sanders, A. The expected Peak-to-Average power ratio of white Gaussian noise in sampled I/Q data. IEEE Trans. Instrum. Meas. 74, 1–8 (2025).

    Google Scholar 

  • Ahmad, T. & Chen, H. Deep learning for multi-scale smart energy forecasting. Energy 175, 98–112 (2019).

    Google Scholar 

  • Shen, Q. et al. Short-Term load forecasting based on Multi-Scale ensemble deep learning neural network. IEEE Access. 11, 111963–111975 (2023).

    Google Scholar 

  • Fang, X. & Yuan, Z. Performance enhancing techniques for deep learning models in time series forecasting. Eng. Appl. Artif. Intell. 85, 533–542 (2019).

    Google Scholar 

  • Ham, Y. G., Kim, J. H. & Luo, J. J. Deep learning for multi-year ENSO forecasts. Nature 573, 568–572 (2019).

    Google Scholar 

  • Cheng, M., Fang, F., Kinouchi, T., Navon, I. M. & Pain, C. C. Long lead-time daily and monthly streamflow forecasting using machine learning methods. J. Hydrol. 590, 125376 (2020).

    Google Scholar 

  • Papacharalampous, G. A., Tyralis, H. & Koutsoyiannis, D. Comparison of stochastic and machine learning methods for multi-step ahead forecasting of hydrological processes introduction 1. 1 time series forecasting in hydrology and beyond. Eur. Water. 59, 161–168 (2017).

    Google Scholar 

  • Niu, T., Wang, J., Lu, H., Yang, W. & Du, P. Developing a deep learning framework with two-stage feature selection for multivariate financial time series forecasting. Expert Syst. Appl. 148, 113237 (2020).

    Google Scholar 

  • Zhang, X. et al. Multi-period learning for financial time series forecasting. 2848–2859 (2025). https://doi.org/10.1145/3690624.3709422

  • Livieris, I. E., Stavroyiannis, S., Pintelas, E. & Pintelas, P. A novel validation framework to enhance deep learning models in time-series forecasting. Neural Comput. Appl. 32, 17149–17167 (2020).

    Google Scholar 

  • Fang, J. et al. An attention-based deep learning model for multi-horizon time series forecasting by considering periodic characteristic. Comput. Ind. Eng. 185, 109667 (2023).

    Google Scholar 

  • Kolassa, S. & Schütz, W. Advantages of the MAD/Mean ratio over the MAPE. Foresight Int. J. Appl. Forecast. 6, 40–43 (2007).

    Google Scholar 

  • Makridakis, S., Spiliotis, E. & Assimakopoulos, V. The M4 competition: 100,000 time series and 61 forecasting methods. Int. J. Forecast. 36, 54–74 (2020).

    Google Scholar 

  • Makridakis, S., Spiliotis, E. & Assimakopoulos, V. The M5 competition: Background, organization, and implementation. Int. J. Forecast. 38, 1325–1336 (2022).

    Google Scholar 

    • Continue Reading

  • China’s offshore yuan breaks major benchmark as appreciation momentum builds

    China’s offshore yuan breaks major benchmark as appreciation momentum builds

    China’s offshore yuan strengthened further on Thursday, dipping below the benchmark rate of seven against the US dollar – another sign of the currency’s continued appreciation after a brief breach of the same threshold on Wednesday evening.

    The recent fluctuations mark the currency’s first appreciation past the major psychological marker in 15 months, suggesting a change in market sentiment and providing more support for the global investors and economists who have argued the currency has been undervalued.

    The offshore yuan’s exchange rate reached a high of 6.9960 on Thursday morning, according to figures from Chinese financial data provider Wind, after briefly moving to 6.9999 on Wednesday. The onshore yuan, meanwhile, hit 7.01 against the US dollar on Thursday. Both rates were reached for the first time since September 2024.

    Market optimism for the Chinese currency continues to rise, with analysts predicting the yuan will appreciate further, though small and medium-sized exporters may have concerns over the pressures a stronger yuan could exert on their operations.

    The yuan’s recent strengthening reflects both a weaker US dollar and shifts in the supply and demand of foreign exchange, analysts said.

    Sustained trade surpluses and concentrated settlement of foreign exchange among companies have provided a temporary boost in demand for China’s currency, compounded by investor concerns related to the sustainability of US government debt.

    Continue Reading

  • Essex man invites strangers to pubs at Christmas

    Essex man invites strangers to pubs at Christmas

    Mr Perryman said he was usually “super allergic” to social media, but had been sharing videos as part of his campaign to his 40,000 followers.

    He is familiar with getting strangers to meet up; for his day job, he organises events for singletons.

    Mr Perryman, who now lives in Stratford in east London, said he had a “real mix” and a “lovely bunch” of people coming to his scheduled meet-ups. He hopes to keep in touch with his new friends.

    “I don’t want to put myself out there and then disappear after people have had the courage to come out and see me on my own in the pub; I’m not going to leave them behind.

    “Sometimes a four-hour conversation like that is a deeper conversation than you might have with a friend that you only see once every four months or whatever, and it’s really nice.”

    Continue Reading

  • Essex man invites strangers to pubs at Christmas

    Essex man invites strangers to pubs at Christmas

    Supplied A long table in a dimly lit pub or restaurant, with six people sat down for a selfie while they smile at the camera. There are some alcoholic drinks on tableSupplied

    Dean Perryman (bottom right) usually wears bright orange so people can spot him easily

    A man has been inviting strangers to sit with him in pubs every day in December in his effort to tackle loneliness.

    Dean Perryman, from Hockley in Essex, has been reserving empty tables and advertising his whereabouts online.

    The 29-year-old says more than 64 people have joined him and he has booked a table at a pub in Rayleigh on Christmas morning.

    Mr Perryman felt compelled to run his “empty chair” campaign after his best friend recently took his own life, and he said the response had been “really overwhelming” and “really positive”.

    “It’s very new for me and I’ve loved every second of it.”

    Mr Perryman said he was usually “super allergic” to social media, but had been sharing videos as part of his campaign to his 40,000 followers.

    He is familiar with getting strangers to meet up; for his day job, he organises events for singletons.

    Mr Perryman, who now lives in Stratford in east London, said he had a “real mix” and a “lovely bunch” of people coming to his scheduled meet-ups. He hopes to keep in touch with his new friends.

    “I don’t want to put myself out there and then disappear after people have had the courage to come out and see me on my own in the pub; I’m not going to leave them behind.

    “Sometimes a four-hour conversation like that is a deeper conversation than you might have with a friend that you only see once every four months or whatever, and it’s really nice.”

    Supplied About 10 people gathered round a couple pub tables. There are menus and glasses on the tables. It is dimly lit. A woman in the foreground is taking the selfie photo.Supplied

    Mr Perryman has been booking tables in Essex and London

    Mr Perryman said he had no idea how depressed his good friend was.

    “There are so many people out there who live their life seeing multiple empty chairs because they feel lonely,” he said.

    “I’m not great at a lot of things but I can sit down in the pub and chat so I thought I’d start there.”

    Mr Perryman arranged a group walk, meeting at the Serpentine Lido in Hyde Park on Christmas Eve morning, followed by a gathering at a pub in Southend-on-Sea in the afternoon.

    He has booked tables at pubs in Chelmsford, Hockley and Leigh-on-Sea in the days following Christmas.

    • If you have been affected by this story or would like support then you can find organisations which offer help and information at the BBC Action Line.
    You may also be interested in

    Continue Reading

  • How the AI market could splinter in 2026

    How the AI market could splinter in 2026

    The AI market is tipped to splinter in 2026.

    The last three months of 2025 were a rollercoaster of tech sell-offs and rallies, as circular deals, debt issuances, and high valuations fueled concerns over an AI bubble.

    Such volatility may be an early sign of how AI investment is set to evolve as investors pay closer attention to who is spending money and who is making it, according to Stephen Yiu, chief investment officer at Blue Whale Growth Fund.

    Investors, especially retail investors who are exposed to AI through ETFs, typically have not differentiated between companies with a product but no business model, those burning cash to fund AI infrastructure, or those on the receiving end of AI spending, Yiu told CNBC.

    So far, “every company seems to be winning,” but AI is in its early innings, he said. “It’s very important to differentiate” between different types of companies, which is “what the market might start to do,” Yiu added.

    This illustration taken on April 20, 2018, in Paris shows apps for Google, Amazon, Facebook and Apple, plus the reflection of a binary code displayed on a tablet screen.

    Lionel Bonaventure | Afp | Getty Images

    He sees three camps: private companies or startups, listed AI spenders and AI infrastructure firms. 

    The first group, which includes OpenAI and Anthropic, lured $176.5 billion in venture capital in the first three quarters of 2025, per PitchBook data. Meanwhile, Big Tech names such as Amazon, Microsoft and Meta are the ones cutting checks to AI infrastructure providers such as Nvidia and Broadcom.  

    Blue Whale Growth Fund measures a company’s free cash flow yield, which is the amount of money a company generates after capital expenditure, against its stock price, to figure out whether valuations are justified.  

    Most companies within the Magnificent 7 are “trading a significant premium” since they started heavily investing in AI, Yiu said.

    “When I’m looking at valuations in AI, I would not want to position — even though I believe in how AI is going to change the world — into the AI spenders,” he added, adding that his firm would rather be “on the receiving end” as AI spending is set to further impact company finances.  

    The AI “froth” is “concentrated in specific segments rather than across the broader market,” Julien Lafargue, chief market strategist at Barclays Private Bank and Wealth Management, told CNBC. 

    The bigger risk lies with companies that are securing investment from the AI bull run but are yet to generate earnings — “for example, some quantum computing-related companies,” Lafargue said. 

    “In these cases, investor positioning seems driven more by optimism than by tangible results,” he added, saying that “differentiation is key.”

    The need for differentiation also reflects an evolution of Big Tech business models. Once asset-light firms are increasingly asset-heavy as they gobble up technology, power and land needed for their bullish AI strategies.

    Companies like Meta and Google have morphed into hyperscalers that invest heavily in GPUs, data centers, and AI-driven products, which changes their risk profile and business model.

    Dorian Carrell, Schroders’ head of multi-asset income, said valuing these companies like software and capex-light plays may no longer make sense — especially as companies are still figuring out how to fund their AI plans.

    “We’re not saying it’s not going to work, we’re not saying it’s not going to come through in the next few years, but we are saying, should you pay such a high multiple with such high growth expectations baked in,” Carrell told CNBC’s “Squawk Box Europe” on Dec. 1.

    Tech turned to the debt markets to fund AI infrastructure this year, though investors were cautious about a reliance on debt. While Meta and Amazon have raised funds this way, “they’re still net cash positioned,” Quilter Cheviot’s global head of technology research and investment strategist Ben Barringer told CNBC’s “Europe Early Edition” on Nov. 20 — an important distinction from companies whose balance sheets may be tighter.

    The private debt markets “will be very interesting next year,” Carrell added. 

    If incremental AI revenues don’t outpace those expenses, margins will compress and investors will question their return on investment, Yiu said. 

    In addition, the performance gaps between companies could widen further as hardware and infrastructure depreciate. AI spenders will need to factor into their investments, Yiu added. “It’s not part of the P&L yet. Next year onwards, gradually, it will confound the numbers.” 

    “So, there’s going to be more and more differentiation.” 

    Continue Reading

  • TikTok shopping habits go from budget to Burberry

    TikTok shopping habits go from budget to Burberry

    Unlock the Editor’s Digest for free

    What is the best place to buy a luxury watch? In China, more and more high-end shoppers are turning to social media. That is an opportunity for luxury brands seeking younger customers. For social networks themselves — including TikTok, currently the subject of a complex US-China carve-up — it could be a big driver of value.

    The first to benefit from this trend has been Douyin, TikTok’s sister platform in China. There, shoppers buy luxury items the same way they buy lipstick or air fryers: by tapping into a livestream. A chatty host demos products and makes jokes, while viewers can buy anything they see on the stream with a single tap. It has become the default way millions of young consumers shop in the country.

    Douyin sold 46 per cent more merchandise by value in the year to July than it did in the same period of 2024, according to the company. It became the third-largest online shopping platform in China last year. European brands have taken notice. Versace has hosted livestreams and opened an official Douyin flagship store. Burberry joined the platform’s “Super Brand Day” and collaborated with Douyin to dress virtual avatars.

    TikTok is trying to replicate Douyin’s success globally by rolling out many of the same features in markets such as the US, including in-app checkout systems, product search tabs and shoppable videos. Its brand makes its push into luxury a tough sell: outside China, TikTok’s commerce business is more like a virtual dollar store.

    Its efforts may nonetheless bear fruit. After all, TikTok and its ilk have a strong following among the young. More than half of Gen Z discovers products on Instagram and TikTok, according to Emarketer. Gen Z’s share of global luxury spending is expected to rise from 4 per cent today to 25 per cent by 2030, according to BCG.

    This suggests that luxury groups — which have historically been cautious about selling on social media due to brand dilution concerns — may reconsider. Staying away from TikTok would mean missing out on their largest group of future customers.

    Even if the luxury companies themselves prove hard to sway, they are not the only ones selling handbags. The market for second-hand fashion and luxury is growing rapidly. The global resale market size is projected to reach up to $360bn in the next five years, according to BCG. An average annual growth rate of 10 per cent means resale is growing nearly three times faster than the primary luxury market. Resale platforms like The RealReal have seen rising shopper demand and growing investor interest.

    With younger consumers increasingly important and sellers proliferating, the next phase of the luxury market overlaps neatly with social media platforms. Future luxury buyers are now forming their brand loyalties, and TikTok is becoming the place where they do it.

    june.yoon@ft.com

    Continue Reading