Preparing Feedloom

Gathering your research workspace

Your research feeds

Feed

UX Research · Baymard Institute
Food Delivery & Takeout UX Benchmark: 3,100+ Performance Scores and 1,900+ Best Practice Examples
Published Aug 13, 2026
(Note: Unfortunately, e-mail and RSS don’t support advanced layouts and features. If the graphics in this article look strange, you may want to read the article in your web browser.) At Baymard, we’ve just released a new UX benchmark with 8 “Food Delivery & Takeout” UX case studies. This follows our large-scale user testing on Food Delivery & Takeout UX and adds to our existing ecommerce UX benchmark . In this article, we give you a snapshot of the overall UX performance. 8 Food Delivery & Takeout UX Case Studies and the Overall Performance Just Eat mediocre Food Delivery & Takeout 49 page designs: mobile, app KFC mediocre Food Delivery & Takeout 48 page designs: mobile, app Uber Eats mediocre Food Delivery & Takeout 54 page designs: mobile, app Domino’s Pizza poor Food Delivery & Takeout 34 page designs: mobile, app DoorDash poor Food Delivery & Takeout 48 page designs: mobile, app McDonald’s poor Food Delivery & Takeout 39 page designs: mobile, app Deliveroo poor Food Delivery & Takeout 53 page designs: mobile, app Burger King broken Food Delivery & Takeout 41 page designs: mobile, app YourSite.com? Want to know how your site performs? Get Premium access to review your own site or have it audited by Baymard researchers. These are the 8 in-depth Food Delivery & Takeout UX case studies. The 8 mobile sites and apps have been manually assessed across 390+ research-based UX parameters relevant to Food Delivery & Takeout , resulting in 3,100+ weighted UX performance scores and 1,900+ best practice examples from these mobile sites and apps. Each of the 3,100+ UX performance scores from the 8 case studies is summarized in the interactive scatterplot below — showing you how they perform collectively and individually: {{ scatterplot-graph: size=big + habitat=public + base-sites=collection:online-food-delivery + view-structure-id=gemini-st_l9jw5tjp }} The overall UX performance of the Food Delivery & Takeout sites and a
Score 5/100 · Below threshold
Why this score
Description is available (+5)
Open original source (opens in a new tab)

TravelTravel Research
ActiveUpdated Aug 16, 2026
Crypto ResearchCrypto Research
ActiveUpdated Aug 16, 2026
UX ResearchUX Research
ActiveUpdated Aug 15, 2026

Crypto Research · arXiv — Crypto Trading & Market Microstructure

TradingMoE: Routing the Right Experts in Evolving Markets

Published Aug 12, 2026

Large language models (LLMs) have shown strong potential for financial analysis and trading, but direct trading remains challenging because the predictive capabilities required can vary across assets, decision fields, and market conditions. Existing LLM-based trading systems either coordinate human-defined external experts or adopt conventional internal Mixture-of-Experts (MoE) routers that do not directly evaluate how individual experts contribute to trading decisions. Moreover, these routers receive no direct signal indicating when an inactive expert has become more suitable as market conditions change. We find that native router scores poorly reflect how much individual experts improve trading decisions, frequently leaving better alternatives unselected. We further reveal that token-specific expert usefulness exhibits a compact low-dimensional structure. Based on these findings, we propose TradingMoE, a trading-oriented sparse MoE that augments a frozen dense LLM with lightweight residual experts. We introduce a Query-Key router that represents the expertise required by each token under the current market context as a low-dimensional query and matches it with learnable expert keys. We further propose a sparse expert selection update mechanism that samples a few inactive experts during training and estimates whether they should replace the weakest expert in the current Top-k route. This mechanism enables the router to update expert selection as market conditions change while preserving sparse computation. Experiments against 22 baselines on stock and cryptocurrency markets show that TradingMoE improves cumulative return over the best-performing baselines by 30.89% and 30.7%, respectively. Rolling paper-trading experiments further demonstrate that its advantage persists under forward-only deployment.

cs.LG

Score 5/100 · Below threshold

Why this score

Description is available (+5)

Open original source (opens in a new tab)

Crypto Research · arXiv — Crypto Trading & Market Microstructure

Anomaly detection in European cryptocurrency exchange-traded products

Published Aug 10, 2026

Cryptocurrency exchange-traded products (ETPs) listed on European exchanges provide a regulated environment for studying intraday market anomalies. We study four Bitcoin and Ethereum ETPs traded on Xetra and Nasdaq Stockholm over the period January 2024 - December 2025 using one-minute bars. As a benchmark, we adopt an extreme value theory approach in which anomalous bars are defined as returns falling below a threshold estimated by fitting a generalised Pareto distribution to left-tail exceedances. We then propose three new binary anomaly indicators. The first, a cross-venue divergence anomaly, identifies venue-specific price divergence between the two exchanges. The second is a no-recovery anomaly that identifies extreme price drops followed by little or no recovery over the next ten active bars. The third is a momentum-reversal anomaly that identifies extreme price drops following positive short-term momentum. Although each anomaly type represents fewer than 1% of one-minute bars, statistical analysis using Mann-Whitney U tests shows that anomaly observations exhibit significantly higher effective spreads, higher values of liquidity-related ratios, and more pronounced order-flow imbalances than non-anomalous bars. Furthermore, employing an out-of-sample prediction methodology with four classifiers - random forest, logistic regression, extreme gradient boosting, and light gradient boosting machine - shows that all four anomaly types are predictable one bar ahead, with AUC-ROC values of up to 0.82. Permutation importance indicates that short-term volatility and drawdown measures are generally more useful for prediction than microstructure variables.

q-fin.MF

Score 75/100 · Meets threshold

Why this score

Matches an include keyword (+50)
Description is available (+5)
Custom rule: Out-of-sample evidence (+20)

Open original source (opens in a new tab)

UX Research · Baymard Institute

Ecommerce Quantitative UX: 3 High-Level Trends & Takeaways from 20+ Charts

Published Aug 4, 2026

(Note: Unfortunately, e-mail and RSS don’t support advanced layouts and features. If the graphics in this article look strange, you may want to read the article in your web browser.) Key Stats & Takeaways 20+ new insights on general ecommerce shopper habits and preferences 1,083 US online shoppers surveyed in this quantitative UX study General ecommerce shopping habits remain consistent overall, but a few noteworthy shifts stand out this year, from declining return frequencies to changing newsletter preferences At Baymard, we’ve released updated Quantitative Insights into the habits and preferences of online shoppers across key ecommerce stages. These insights are survey-based data visualizations that complement and deepen our large-scale UX research findings . The 20+ insights cover the online shopping experience across a wide range of topics: product discovery, cross-sell preferences, sale-seeking behaviors, checkout friction, return behaviors, account feature expectations, and newsletter management. Ecommerce shoppers make decisions with deeply ingrained expectations regarding transparency, flexibility, and convenience, standards that continually evolve as the online shopping landscape evolves. While the vast majority of our foundational data remains highly stable year after year, tracking this data over time reveals a few subtle, critical shifts in reported consumer habits. Over the last 3 years, consumer habits show clear directional changes in how frequently shoppers make online returns for their past purchases. Further down the funnel, they frequently leverage accounts to manage post-purchase utility, carefully balancing their participation in brand incentive structures with their tolerance for direct retailer outreach. Each of these evolving dynamics plays out at a different stage of the user experience, highlighting exactly where optimization priorities should be re-evaluated. In this article, we’ll highlight 3 high-level findings that reflect the most

Score 55/100 · Below threshold

Why this score

Matches an include keyword (+50)
Description is available (+5)

Open original source (opens in a new tab)

UX Research · Baymard Institute

Internet Service Providers UX Benchmark: 1,700+ Performance Scores and 1,100+ Best Practice Examples

Published Jul 28, 2026

(Note: Unfortunately, e-mail and RSS don’t support advanced layouts and features. If the graphics in this article look strange, you may want to read the article in your web browser.) At Baymard, we’ve just released a new UX benchmark with 7 “Internet Service Providers (ISP)” UX case studies. This follows from our large-scale user testing and adds to our existing ecommerce UX benchmark . In this article, we give you a snapshot of the overall UX performance. 7 Internet Service Providers UX Case Studies and the Overall Performance BT mediocre Internet Service Providers 26 page designs: desktop, mobile Telekom mediocre Internet Service Providers 35 page designs: desktop, mobile Frontier mediocre Internet Service Providers 25 page designs: desktop, mobile COX mediocre Internet Service Providers 24 page designs: desktop, mobile CenturyLink mediocre Internet Service Providers 25 page designs: desktop, mobile Viasat mediocre Internet Service Providers 32 page designs: desktop, mobile Spectrum mediocre Internet Service Providers 15 page designs: desktop, mobile YourSite.com? Want to know how your site performs? Get Premium access to review your own site or have it audited by Baymard researchers. These are the 7 in-depth ISP UX case studies. The 7 ISP sites have been manually assessed across 280+ research-based UX parameters relevant to Internet Service Providers , resulting in 1,700+ weighted UX performance scores and 1,100+ best practice examples from these desktop and mobile sites. Each of the 1,700+ UX performance scores from the 7 ISP case studies is summarized in the interactive scatterplot below — showing you how they perform collectively and individually: {{ scatterplot-graph: size=big + habitat=public + base-sites=collection:internet-service-providers + view-structure-id=gemini-st_9en1b7zd }} The overall UX performance of the ISP sites is “mediocre,” with 1 site just on the border of “poor”. ISP sites typically offer a narrow set of services, but these services come

Score 5/100 · Below threshold

Why this score

Description is available (+5)

Open original source (opens in a new tab)

Crypto Research · arXiv — Crypto Trading & Market Microstructure

Are cryptocurrencies real financial bubbles? Evidence from quantitative analyses

Published Jul 23, 2026

The growth of peer-to-peer exchanges and the blockchain technology has led to a proliferation of cryptocurrencies and to a massive increase in the number of investors who actually negotiate digital money. Cryptocurrencies trade at prices mainly driven by investor sentiment, becoming a potential source of financial bubbles and instabilities. In this work, we apply quantitative models to the study of Bitcoin and Ether, two of the most famous cryptocurrencies. Our bubble detection methodology combines the Log Periodic Power Law (LPPL) model, originally created by Johansen, Ledoit and Sornette (JLS), and the statistical model developed by Phillips, Shi, and Yu (PSY). In particular, we employ three different versions of JLS model, i.e. Ordinary Least Square (OLS), Generalised Least Squares (GLS) and Maximum Likelihood Estimation (MLE), and two PSY statistical tests (BSADF and BSADF*). We find that, during the sample period 1st December 2016 - 16th January 2018, Bitcoin shows typical hallmarks of a bubble phase in mid December 2017 and in the first half of January 2018, anticipating the large crashes observed thereafter. Also the Ether price dynamics reveals bubble evidence in mid June 2017, anticipating the crash observed on 12th June, and a weaker signal around 12th January 2018, anticipating the crash observed in the same days. This paper confirms the high risk of speculative bubbles associated with cryptocurrencies, related to investor exuberance pumping market prices far away from their fundamental values, thus creating critical situations subject to possible crashes. Our methodology is general and can be applied to virtually any financial time series, and may support investing and risk management strategies.

q-fin.RMecon.GNq-fin.CPq-fin.GNq-fin.ST

Score 55/100 · Below threshold

Why this score

Matches an include keyword (+50)
Description is available (+5)

Open original source (opens in a new tab)

Crypto Research · arXiv — Crypto Trading & Market Microstructure

Predictive Extrema, Unprofitable Policies: An AI-Assisted Audit of Candle-Based Binance Spot Timing Models

Published Jul 21, 2026

We audit whether candle-based machine-learning models can turn predictions of cryptocurrency extrema or short-horizon outcomes into positive Binance Spot paper policies after assumed costs. Numerical results come from scripted fixed-seed model runs and deterministic simulators; human-supervised AI agents supported the July 20 evidence-integrity revision through literature retrieval, separately tasked critique, artifact reconciliation, documentation, and source packaging, not trading decisions. The strongest later-period evidence, conditional on extensive predecessor search, is negative: an unchanged ten-pair mandatory-daily selector lost 6.72\% over 19 July cycles at an assumed 31-bps completed-cycle cost, with 3 wins and 16 losses. In short model-specific July evaluations, the validation-selected local-minimum policy returned -1.79\%, while the local-maximum sell-to-cash/re-entry policy underperformed continuous holding by 2.80\%; their gross mean advantages of 11.11 and 12.21 bps were below even the 21-bps stress. A Gurgul-inspired, OHLCV-only daily adaptation attained minimum/maximum ROC AUC of 0.874/0.896 but average precision of only 0.134/0.116 and lost 44.30\% over seven cycles, versus -41.20\% for buy-and-hold. A forensic audit also downgraded an earlier One4All "30-day holdout": its dates had influenced prior architecture work, its four-hour outcome horizon was not purged at split boundaries, it used same-close entry, and its raw result directories were absent. Across the tested, mostly exploratory protocols, event-ranking performance did not establish positive executable policy value. Every operational decision remains NO\_TRADE.

cs.LGcs.AIq-fin.STq-fin.TR

Score 5/100 · Below threshold

Why this score

Description is available (+5)

Open original source (opens in a new tab)

UX Research · Baymard Institute

Time is Money: How Pre-Validated UX Research Protects Your Resources & Budget

Published Jun 30, 2026

(Note: Unfortunately, e-mail and RSS don’t support advanced layouts and features. If the graphics in this article look strange, you may want to read the article in your web browser.) When you try to prove the value of your design decisions, you often hit a wall. To back up your designs confidently, you need robust usability research . But conducting that research yourself for every minor UI change or UX decision is expensive and time-consuming . Baymard’s pre-validated research gives teams like yours access to established UX best practices in an instant, which helps you ship designs faster, more confidently, and protects your costs by reducing the research overheads. To demonstrate that value, we’ve created a calculator that shows how a Baymard subscription helps you move faster and keep your UX research costs down. Note: This calculator demonstrates initial cost savings based on consuming Baymard’s insights versus running new user research yourself. Baymard users typically see further value realizations from implementing those insights and improving website user experience and conversion rates. The Hidden Cost of Your In-House User Research It’s easy to fall into the trap of thinking: “We can just replicate this research in-house for less cost.” But when you try to build everything from scratch without access to pre-validated UX insights , your team quickly runs into three major roadblocks: You move slower: Your product teams lose momentum and your project timelines stretch out because you are forced to conduct research from scratch You risk harming conversions: When timelines get tight, you risk making critical UX decisions based on gut feelings, which can ultimately harm your conversion rates You spend more money: To keep up with research demands, you end up spending vastly more on external agency services or expanding your internal headcount just to conduct baseline testing What It Actually Costs You to Produce a UX Insight To protect your resources, you need to

Score 40/100 · Below threshold

Why this score

Matches an include keyword (+50)
Description is available (+5)
Custom rule: Recycled Advice (-15)

Open original source (opens in a new tab)

UX Research · Baymard Institute

Flight Booking & Airlines Quantitative UX: 3 High-Level Takeaways from 30+ Charts

Published Jun 24, 2026

(Note: Unfortunately, e-mail and RSS don’t support advanced layouts and features. If the graphics in this article look strange, you may want to read the article in your web browser.) Key Stats & Takeaways 30+ new insights on Flight Booking & Airline shopper habits and preferences 3,125 US online shoppers surveyed in this quantitative UX study Flight Booking & Airline shoppers use a variety of sources when looking for flights, are highly likely to participate in loyalty programs, and commonly bundle other travel-related bookings at the same time, particularly when travelling for business. At Baymard, we’ve just released new Quantitative Insights into people who shop on “ Flight Booking & Airlines ” sites , expanding our understanding of the habits and preferences of online shoppers in this category. These insights are visualizations based on survey data that supplement and support our large-scale UX research findings and benchmarking of the Flight Booking & Airlines industry. The 30+ insights address the Flight Booking & Airlines online shopping experience , spanning a breadth of topics: sources for flight searching, types of add-ons purchased, usage of apps and external tools, importance of cancellations, sales and scarcity incentives, and motivations for bundling and loyalty programs. Some of the charts also explore sub-segments of the broader audience such as business travelers and whether they are more likely to purchase add-ons , or how device usage between desktop and mobile changes as flight price becomes more expensive. These Quantitative Insights empower you to align stakeholders through objective, survey-backed data, streamline A/B testing with high-potential hypotheses, and discover and address industry-specific UX challenges. In this article, we’ll highlight 3 high-level findings that reflect how shoppers evaluate and purchase flights online: Flight shoppers use a variety of sources when looking for flights Three quarters of flight

Score 55/100 · Below threshold

Why this score

Matches an include keyword (+50)
Description is available (+5)

Open original source (opens in a new tab)

Crypto Research · arXiv — Crypto Trading & Market Microstructure

Do Prediction Markets Match Option Prices? Bitcoin Threshold Evidence from Binance and Polymarket

Published Jun 17, 2026

The digitization of financial markets has produced two classes of platforms that price, in principle, the same state - contingent payoffs: centralized crypto-option exchanges and blockchain-based prediction markets. This paper provides the first option-implied benchmark test of prediction-market pricing for cryptocurrency threshold contracts. For each hour in a matched sample, we compare the Polymarket Yes price with the discounted risk-neutral binary value implied by a listed Binance call option on the same underlying, strike, and maturity, and study the gap between them. In the main September 2023 Bitcoin contract, the mean pricing gap equals 5.6 percentage points across 214 hourly observations (t = 6.46, p < 10^{-9}). Pooling three Binance-compatible Bitcoin threshold markets yields a mean gap of 6.3 percentage points across 287 observations, robust to HAC and block-bootstrap inference. The gap is persistent - with an AR(1) half-life of roughly four hours - yet mean-reverting, consistent with slow information transmission between segmented venues rather than mechanical noise. Cross-sectional regressions reveal that the wedge is largest at low option-implied probabilities and long maturities, a pattern consistent with speculative demand for prediction-market contracts rather than measurement error. A delta-hedged arbitrage proxy remains profitable after conservative transaction costs, though with marginal statistical precision. A Deribit extension on the same three Bitcoin contracts produces a larger pooled gap of 11 percentage points, while a smaller Ethereum exercise yields mixed evidence. The results demonstrate that digital fragmentation of financial markets generates systematic, persistent pricing wedges even for economically identical payoffs.

q-fin.TR

Score 70/100 · Meets threshold

Why this score

Matches an include keyword (+50)
Description is available (+5)
Custom rule: Realistic transaction costs (+15)

Open original source (opens in a new tab)

UX Research · Baymard Institute

Apparel & Accessories Quantitative UX: 3 High-Level Takeaways from 40+ Charts

Published Jun 12, 2026

(Note: Unfortunately, e-mail and RSS don’t support advanced layouts and features. If the graphics in this article look strange, you may want to read the article in your web browser.) Key Stats & Takeaways 40+ new insights on Apparel & Accessories shopper habits and preferences 1,922 US online shoppers surveyed in this quantitative UX study Apparel and accessories shoppers face uncertainty at every stage of their journey, from how their self-identification corresponds to retailer categories to how they evaluate fit on the product page We’ve released new Quantitative Insights into people who shop on “ Apparel & Accessories ” sites , adding to our growing body of data on the habits and preferences of online shoppers across key ecommerce categories. These insights are survey-based data visualizations that complement and deepen our large-scale UX research findings and benchmarking of the Apparel & Accessories industry. The 40+ insights cover the Apparel & Accessories online shopping experience across a wide range of topics: online trip drivers, size and fit confidence, product page evaluation, reviews usage, returns, loyalty programs, and delivery expectations. Apparel and accessories shoppers make decisions with incomplete information at every stage of their online journey. They can’t feel a fabric, try out a fit, or know whether a size label will translate to their body. Beyond that, they may not know whether the items they buy will ultimately feel “worth it” to them, or whether they’ll shop frequently enough to make a rewards program worth joining. Each of these uncertainties plays out at a different stage of the journey, and each has distinct implications for how Apparel & Accessories sites should be designed. In this article, we’ll highlight 3 high-level findings that reflect how apparel and accessories shoppers navigate that uncertainty: Shoppers’ self-reported size categories often diverge from industry definitions Infrequent shopping, not

Score 55/100 · Below threshold

Why this score

Matches an include keyword (+50)
Description is available (+5)

Open original source (opens in a new tab)

Crypto Research · arXiv — Crypto Trading & Market Microstructure

Dynamic Multi-Pair Trading Strategy in Cryptocurrency Markets with Deep Reinforcement Learning

Published Jun 3, 2026

This study aims to determine whether the application of Deep Reinforcement Learning (DRL) as a specialized execution overlay can enhance pair trading in highly volatile cryptocurrency markets. Although classical implementations of the strategy have proven successful in traditional equities, they frequently exhibit rigidity and suffer from severe divergence risks when applied to high-variance environments. To address this need, this research introduces novel concepts. To construct a robust system, we developed a hierarchical "Filter-then-Rank" pair selection methodology and a proprietary "Fixed Risk, Adaptive Mean" execution model. The system employs a Proximal Policy Optimization (PPO) agent with a Long Short-Term Memory (LSTM) layer to govern execution decisions within strict deterministic risk management boundaries. Evaluated on 1-hour interval data from the Binance USD-M Futures market, the optimized RL policy achieved an out-of-sample performance that substantially outperformed the heuristic baseline. A stationary circular block bootstrap robustness check confirms that the agent's risk-adjusted outperformance is statistically significant at the 10 percent level. Although falling marginally short of the stricter 5 percent threshold, this result highlights the extreme idiosyncratic variance characteristic of digital assets. Ultimately, this thesis contributes to the quantitative finance literature by introducing a hybrid architecture that combines statistical arbitrage with DRL execution policies. Furthermore, it delivers a novel framework for safe reinforcement learning via deterministic shielding, proving that anchoring a neural policy to statistically robust boundaries successfully mitigates severe divergence risks.

cs.LGcs.NEq-fin.STq-fin.TRstat.ML

Score 75/100 · Meets threshold

Why this score

Matches an include keyword (+50)
Description is available (+5)
Custom rule: Out-of-sample evidence (+20)

Open original source (opens in a new tab)

Gathering your research workspace

Feed

Food Delivery & Takeout UX Benchmark: 3,100+ Performance Scores and 1,900+ Best Practice Examples

Topics you define

When you should (or shouldn’t) use the Amex Gold Card

American Express Platinum: Valuable benefits and an offer as high as 175,000 bonus points

TradingMoE: Routing the Right Experts in Evolving Markets

Anomaly detection in European cryptocurrency exchange-traded products

F$^2$Agent: Financial Fusion of Agentic Intelligence for Multimodal Trading

Ecommerce Quantitative UX: 3 High-Level Trends & Takeaways from 20+ Charts

CoLAS: Multimodal Corroboration of Latent Asset Signals for Financial Trading

Internet Service Providers UX Benchmark: 1,700+ Performance Scores and 1,100+ Best Practice Examples

Are cryptocurrencies real financial bubbles? Evidence from quantitative analyses

Predictive Extrema, Unprofitable Policies: An AI-Assisted Audit of Candle-Based Binance Spot Timing Models

Detecting unusual trading patterns on cryptocurrency exchanges by means of complexity measures

ManiScope: LLM-Assisted Visual Analytics of Cryptocurrency Manipulation Risk

A Truckload of Satoshis: Detecting and Measuring One-Way Arbitrage in the Wild

The Quarter-Hour Effect: Periodic Algorithmic Trading and Return Predictability in Cryptocurrency Futures

Time is Money: How Pre-Validated UX Research Protects Your Resources & Budget

Flight Booking & Airlines Quantitative UX: 3 High-Level Takeaways from 30+ Charts

Time-dependent weighted directed networks of cryptocurrency interaction from high-frequency returns

Do Prediction Markets Match Option Prices? Bitcoin Threshold Evidence from Binance and Polymarket

Trading in the Sunshine or in the Shade: Market Impact and Adverse Selection on Hyperliquid

Apparel & Accessories Quantitative UX: 3 High-Level Takeaways from 40+ Charts

Building Social World Models with Large Language Models

Dynamic Multi-Pair Trading Strategy in Cryptocurrency Markets with Deep Reinforcement Learning

When market boundaries weaken: Network reconfiguration and regime-dependent cross-asset spillovers

Machine Learning-Based Bitcoin Trading Under Transaction Costs: Evidence From Walk-Forward Forecasting