Ex-JPMorgan Metals Trader ‘Disco' Is Joining Trafigura
Amsbury, known in the tight-knit world of the LME as ' Disco,' is the latest veteran of JPMorgan's metals team to take a high-profile role in the physical trading industry.
Try Our AI Features
Explore what Daily8 AI can do for you:
Comments
No comments yet...
Related Articles
Yahoo
5 minutes ago
- Yahoo
Appeal to create 'peaceful haven' for youngsters
A charity which works with neurodivergent young people is appealing for help to transform an empty storeroom into a "peaceful haven". MAIN, based in Middlesbrough, is seeking £25,000 to create a quiet room - a sensory-controlled space for youngsters to use when feeling overwhelmed. The "calm, safe and low-stimulation environment" would allow them to manage emotions, reduce stress and "help prevent crisis situations", the charity said Some businesses have offered their expertise for free, but funds are needed for materials to complete the project. MAIN supports children and adults who are autistic, neurodivergent, have learning disabilities or complex needs. Services include one-to-one support, opportunities to experience new activities, a parent forum group, along with advice and guidance. MAIN CEO Heather Whyman, said: "Many of the children we support experience sensory overload that can lead to high levels of distress, sometimes resulting in aggression towards themselves or others. "By having access to a dedicated space where they can de-escalate safely and in their own time, we can help prevent crisis situations and better support their continued participation." The quiet room would be a "vital part" of creating a "stable, inclusive and therapeutic environment", she said. "We have such great support from the local community and we know that times are hard for lots of people, but we are a very small charity which relies heavily on donations. "Everyone who is helping us – from the architect to the building works, public relations to marketing support - is doing it for free simply because they strongly believe it is the right thing to do." Follow BBC Tees on X, Facebook, Nextdoor and Instagram. Related internet links MAIN More on this topic 'Going to A&E is unbearable for my autistic son' Councillor's autism 'benefits politics' Second cafe staffed by disabled workers to open
Forbes
8 minutes ago
- Forbes
How To Use Claude AI For Crypto Management: A Practical Guide For 2025
With increasing trading volumes and volatility, keeping up with crypto training is nearly impossible ... More as a human. Can an AI like Claude help? The crypto universe has reached truly planetary scale. Total market capitalization now hovers around $3.4 trillion, spread across 17,581 actively-tracked coins. Crypto never sleeps, as it trades 24/7, so every hour, those assets move through time zones, generating an endless feed of prices, blockchain metrics and gossip. Social-intelligence platform LunarCrush says it collects more than two million crypto-related posts every single day, ingesting tweets, Reddit threads, YouTube comments and even Discord chat logs, before ranking each for sentiment and engagement. For portfolio managers, that torrent is both opportunity and overload. A mid-level analyst in New York costs roughly $99,600 a year to sift the noise, while Anthropic charges just $3 for a million input tokens on its 200 k-context Claude 3 Sonnet model. The math is clear: 2025 is the first year it is cheaper, and faster, to let AI watch the market full-time and escalate only what matters. Can AI Help with Crypto Management? Crypto's risk profile dwarfs that of traditional assets. A 2023 academic study comparing daily returns found that 80% of S&P 500 moves sit inside a ±1% band, whereas just 40% of bitcoin's do; in other words, BTC's intraday swings are roughly five times larger than blue-chip equities. Add 24/7 trading and thousands of micro-cap tokens, and the surveillance burden mushrooms exponentially. Meanwhile, pressure on professional fees is rising. Institutional investors now demand minute-level oversight, but balk at paying for night-shift quants. Plugging Claude into a data pipeline costs pennies per hour and delivers machine-speed pattern recognition with a legally auditable paper-trail. The result is a structurally cheaper and demonstrably calmer portfolio operation. What Is Claude AI? Anthropic's Claude 3 family launched in March 2024 and has since moved to Sonnet 3.7 and beta 3.5 releases, each keeping the 200 k-token context window, but adding function-calling, deterministic JSON output and enterprise 'Trust Center' options. At $3/MTok input and $15/MTok output, Sonnet offers the best cost-to-comprehension ratio in the model line-up. Two design characteristics of Claude matter for regulated finance. First, constitutional AI: teams can embed hard rules ('never suggest deterministic price targets,' 'flag any address on an OFAC list') that Claude must follow. Second, Anthropic provides full audit logs, satisfying SOC 2, MAS TRM and ESMA retention requirements without duct-taped loggers. Key Data Feeds Claude Can Ingest Claude is modality-agnostic: if you can get the data in text or a machine-readable format like JSON, you can embed it. Typical crypto desks stream four pillars into Claude: A cron script bundles each feed into hourly JSON objects and posts them to Claude's /v1/messages endpoint. Latency? Typically under two seconds per 100k tokens. How Claude AI Can Help Analyze Crypto Market Data Sample prompt: You are a crypto news analyst. Collapse the 25 headlines below into a three-sentence risk brief ranked by market impact. Output JSON with fields {riskLevel, who, what, why}. A well-structured prompt turns 25 raw headlines into a three-sentence brief ordered by potential price impact and frees up analysts to sanity-check rather than skim headlines. In addition, as the output arrives in strict JSON, dashboard code can colour-code 'high-impact' stories red without human touch. Give Claude the past 30 days of bitcoin price data and ask it to calculate some popular technical indicators such as the 14-day Relative Strength Index (RSI), the 20-day simple moving average, or Bollinger Bands, which help identify price volatility. The model can handle these calculations quickly, returning clean outputs that can be directly plugged into dashboards or trade alerts. Beyond the basics, with good prompting and input data, Claude can calculate z-scores for funding rates, analyze open interest trends or identify volatility skews in the options market using export files from trading platforms. It's a versatile way to structure raw market data into insight. The real value of Claude emerges when it's used to connect social sentiment with on‑chain activity. Take Dogwifhat (WIF), the Solana‑based memecoin that ripped ≈40 % in a single session after Coinbase's listings chief teased an imminent roadmap addition—a tweet that sent #WIF mentions on X soaring and LunarCrush social‑volume scores up triple‑digits. Claude, ingesting that LunarCrush feed alongside Glassnode's spike in large‑holder transfers, flagged the anomaly, drafted a concise risk brief and suggested checking Solana‑perp funding‑rate and options open‑interest screens to confirm real capital was piling in. When Coinbase's formal listing post dropped and WIF printed $4.21, desks running the Claude pipeline were already long while most traders were still digesting the news. Automating Portfolio Rebalancing And Trade Alerts Claude doesn't execute trades or hold private keys, but it plays a critical role in monitoring portfolio rules and drafting trade suggestions the moment thresholds are hit. For instance, if bitcoin's weight in a portfolio drifts more than three percentage points from its target allocation, Claude can immediately flag the deviation and generate a suggested rebalance. This kind of rules-based automation helps desks stay disciplined without constant monitoring. Claude can translate technical signals, like a moving average crossover or volatility spike, into structured, machine-readable trade tickets or webhook triggers. It's not replacing decision-makers, but giving them a faster way to move from insight to action. Risk Management And Sentiment Monitoring Glassnode's market metrics recently showed that 94% of bitcoin supply was sitting in profit, with NUPL indicators edging into euphoric territory. When this kind of data is streamed into Claude alongside a social sentiment heatmap, the AI can quickly distill the signals into a clear and actionable message, something like: 'profitability and hype at simultaneous highs; consider trimming risk.' That output is then routed to Slack or an internal dashboard, where visual alerts change from cautionary orange to deep red. From raw data to a risk signal in under five minutes, Claude helps traders stay ahead of sentiment-driven reversals. Limitations And Best Practices Claude is powerful, but it's not a trading engine. It cannot, and should not, place trades directly. Instead, it is an intelligent assistant, surfacing signals, drafting trade suggestions and flagging risks. Execution should go through a secure, rules-based layer, whether that's a broker API, human-in-the-loop approval, or both. Claude's output is only as good as the input, which means validation is important. Every signal or summary should be reviewed before action is taken. Best practice is to treat Claude as a co-pilot, not a decision-maker. Use structured prompts, apply strict formatting (like JSON schemas) and cross-check against multiple data sources. If Claude surfaces a market-moving insight, it should be confirmed through at least one independent feed—whether that's a newswire, on-chain metric or sentiment provider. By building guardrails into your process, you get the best of both worlds: speed, scale and a human layer of judgment. Bottom Line Crypto's data firehose is only getting wider—more tokens, more trades, more noise. Human teams alone can't keep up. Claude helps by distilling hours of charts, tweets and on-chain activity into structured, readable insights in seconds. Most importantly, it is transparent: Prompts and responses can be logged, audited and improved over time, providing teams with both speed and control. In a market where just 40% of bitcoin's daily price movements stay within a ±1% band, compared to 80% for the S&P 500, volatility is the norm, not the exception. Success depends on pairing AI-powered pattern recognition with human-level judgment. Claude won't give you perfect answers, but it will help you find the right questions faster. The edge goes to those who act first. Frequently Asked Questions (FAQs) Can Claude Connect To Exchanges Directly? No. Claude cannot execute trades or hold API keys. Instead, Claude generates structured outputs that can be passed to a broker system or routed through a human approval layer. Is Claude Accurate With Technical Analysis? Yes—when given clean input data, Claude can calculate common indicators like RSI or moving averages with precision comparable to major charting platforms. Does Claude Provide Price Predictions? Not directly. While it can analyse trends and offer scenario-based insights, it doesn't provide fixed price targets. Its role is to support analysis, not to forecast prices deterministically. Can Claude Manage Multiple Portfolios? Yes. With proper prompt structuring, Claude can track and analyse multiple portfolios at once. Its large context window allows for clear separation and scalability.
Forbes
38 minutes ago
- Forbes
Will Oil Demand Hit 123 Million Barrels Per Day By 2050 As OPEC Says?
(Photo: Bob Riha, Jr.) Earlier this month, the Organization of Petroleum Exporting Countries made yet another attempt to counter energy market chatter about oil demand peaking over the medium- to long-term. At its biennial international seminar in Vienna, Austria from July 9 to 10, the oil producers' group said there was ample evidence in the market for the long-term need for black gold. This need has now reached a point of definitiveness that no one is talking about peak oil demand any more if OPEC's World Oil Outlook 2050 - a fresh assessment report on emerging energy market trends - is to be believed. In the foreword of the report, published just before the conclusion of the seminar, OPEC Secretary General Haitham Al Ghais said: "There is no peak oil demand on the horizon," given that "oil underpins the global economy and is central to our daily lives." That said the producers' group actually cut its global oil demand forecasts for the next four years faced with lower growth in China, a spread of electric vehicles in key markets, and an uncertain macroeconomic climate in the OECD countries. However, it lifted its longer-term view. Crude Projections On Oil Demand That view is based on an assumption that oil will remain 'indispensable' in supporting the economic progress of developing countries, and ensuring the steady output of mission critical hard-to-abate sectors like heavy industry, aviation and haulage. As for the projection figures, the OPEC report forecast that oil demand will average 105 million barrels per day this year. It then expects demand to grow to average 106.3 million bpd in 2026, and then rise to 111.6 million bpd in 2029, and continue rising to as high as 123 million bpd by 2050. 'Despite a marginal decline in its share, oil is set to maintain the largest share in the energy mix in 2050, at just below 30%. The combined share of oil and gas is expected to stay above 50% between 2024 and 2050. At the same time, the share of other renewables in the energy mix increases to 13.5% in 2050, up by 10 percentage points from 2024,' the report noted further. India, along with other Asian nations, the Middle East and Africa, are set to be the 'primary sources of long-term oil demand growth.' Combined demand in these four regions is set to increase by 22.4 million bpd between 2024 and 2050, the OPEC report noted, with India alone adding 8.2 million bpd. 'China's oil demand is projected to increase by less than 2 million bpd over the same time horizon. Moreover, a large part of China's increase is expected to occur over the medium-term, with fewer demand changes expected for the rest of the forecast period.' Not Quite Some Say However, OPEC's long-term oil demand projections don't quite align with what a number of other commentators think. The International Energy Agency expects global oil demand to peak at 105.6 million bpd in 2029 before marginally declining as the end of the current decade approaches. Some in the industry also believe peak demand for oil is imminent this decade. Energy major BP said last year that it even may happen sooner than most people expect, including as early as this year under a specific set of circumstances, and given the rapid growth of renewable energy. Meanwhile, the Energy Institute's recent Statistical Review of World Energy 2025, a global report that was once compiled by BP until very recently, did not directly predict a peak, but noted that some regions and nations - especially China - are seeing a slowdown or plateau in oil demand. Given a 2050 horizon is more than two decades and a half away from now, predicting a peaking of oil demand or otherwise may have a direct corelation with the make-up of the global economy and varying regional productivity levels, boosted by digital tools, each year. Maxime Darmet, Senior Economist for U.S., France and the UK at Allianz Trade said: 'In the coming decades, global productivity will be shaped by the adoption of digital technologies such as artificial intelligence and the enhancements and efficiencies they bring. This will likely drive up energy consumption, but also the efficient usage of energy. Countries that take the lead here would steal a march on others.' The Allianz Trade economist believes given the time horizon when it comes to predicting energy consumption and which source would dominate 25 to 30 years from now will likely be a tricky guesstimate at best. 'Look at the turmoil the global economy is currently facing in the wake of U.S. tariffs slapped by President Donald Trump. Such developments can change the trajectory of demand (and supply) of most commodities however long or short that impact is. Oil is no exception.' Furthermore, if the future of productivity, and indeed the global economy, is digital and AI driven requiring hyperscale data centers, then many including the likes of the IEA, energy majors Shell and Chevron, believe natural gas will likely be the near- to longer-term energy source that benefits. Be that as it may, few dispute that hydrocarbons will be part of the global economy and its energy mix for a while yet. Just that oil - OPEC's preferred one - might not be as dominant a source as the producers' group expects and hopes it would be. Therefore, the debate over peak oil demand and when it will occur won't be settled just yet.
