Safety Stock Calculation Methods

Can we use AI agents for stock market prediction? 😮 Recently, LLM-based agents have demonstrated remarkable advancements in handling multi-modal data, enabling them to execute complex, multi-step decision-making tasks. This research introduces a multi-modal multi-agent system designed specifically for financial trading tasks. The framework employs a team of specialized LLM-based agents, each adept at processing and interpreting various forms of financial data, such as textual news reports, candlestick charts, and trading signal charts. The framework comprises four primary components: the Summarize Module, the Technical Analyst Module, the Prediction Module, and the Reflection Module. The Summarize Module condenses large volumes of textual news data into concise summaries that highlight factual information influencing stock trading decisions. The Technical Analyst Agent leverages the visual reasoning capabilities of LLMs to analyze candlestick charts with technical indicators, providing interpretations for next-day trading strategies. The Reflection Module consists of two parts: one assesses the short-term and medium-term performance of previous trades, while the other plots past trading signals, generates charts, and offers insights into the effectiveness of trades. The Prediction Agent integrates information from these components to forecast trading actions, determine position size as a percentage of the portfolio, and provide a detailed explanation of the decision. Based on the Prediction Agent’s output, the Reward Agent executes trades and calculates performance metrics. These metrics are then used by the Reflection and Prediction Agents in the subsequent iterations. The detailed flow of our framework is illustrated in the Figure. Know more about the framework in this practical research paper: https://lnkd.in/gxvEUGAA Here is my simple video explaining how AI agents work: https://lnkd.in/d_V9DqbH This is my practical hands-on guide on building multi-agent AI system: https://lnkd.in/gdaA5s3Z

If you are a US firm that is importing by container from India, chances are your imports come in through New York, Savannah, or Norfolk based on Census Bureau data. This raises the question: from a safety stock standpoint, are you better off using a carrier that is routing 100% of vessels around the Red Sea via the Cape of Good Hope (and adding ~8 days of transit time to what would usually be a ~26 day trip) (https://lnkd.in/egGZQjQR) versus a carrier that is rolling the dice about going through the Red Sea and may have some vessels either pause or have to divert to go around Africa (resulting in an even longer trip than with a planned Cape of Good Hope route)? Looking at some rough calculations using the most complete conventional formula for setting safety stock, the answer is direct routing around Africa is the better scenario for fast-moving ‘A’ items. One table below. Thoughts: •This table shows inputs for three scenarios: a baseline (left), a Cape of Good Hope planned reroute (center) and the Red Sea (current). With the Cape of Good Hope routing, I’ve added 8 days of to the lead time but not changed the variance of lead time. For the Red Sea scenario, I’ve added 2 days to the average lead time (to account for delays and some unplanned around Africa reroutes) but increased the variance of lead time by 0.25 days (to account for added uncertainty). Demand is assumed to be 100 units a day with a standard deviation of 20 units, resulting in a coefficient of variation of 0.2, which is commonly found for fast moving ‘A’ items. •As can be seen, safety stock is largest under the Red Sea scenario. The reason is due to the mathematics for how the standard deviation of demand during lead time (5th row) is calculated: [(Average Lead Time * Stdev Demand^2) + (Average Demand^2 * Stdev Lead Time^2)]^(1/2) What matters in this equation is that the variance of lead time is multiplied by average demand SQUARED. Thus, for any item where the coefficient of variation of demand is well below 1, an increase in the standard deviation of lead time has a much greater impact on the standard deviation of demand during lead time than an increase in average lead time. Implication: More variable lead times associated with Red Sea disruptions result in more safety stock than longer but less volatile lead times for shipping directly around the Cape of Good Hope. #supplychain #supplychainmanagement #shipsandshipping #logistics #freight

A few years ago, I interviewed a seasoned supply planner from a global FMCG giant. I asked him, "How do you ensure uninterrupted service when forecasts are often wrong?" He smiled and replied, "I don’t trust forecasts blindly. I trust buffers." That stuck with me. We often talk about safety stock like it’s just another calculation - based on service levels, variability, and lead time. But what we often miss is that safety stock is not a backup plan - it’s a confidence plan. When I worked with a food company in North India, we faced wild swings in demand during festive seasons. Despite best efforts, our forecast error remained in the 25–30% range. Initially, we adjusted demand. Then we tried pushing supply. Nothing worked consistently. Until we recalibrated safety stock - not as a static percentage, but as a dynamic lever. We used historical MAPE to segment SKUs: ↳ High forecast error items had higher safety stock, but only if they were fast-movers ↳ For low runners, we capped safety stock and focused on lead time reduction This single change lifted our service levels from 87% to 95% - without inflating inventory across the board. Here’s what I learned: Safety stock isn’t about covering up forecasting failures. It’s about strategically absorbing volatility where it matters most. It’s not "extra" inventory—it’s "essential" inventory. We often praise forecast accuracy, but sometimes, it’s the silent buffers - well-planned, SKU-specific safety stocks - that save the day. Would love to hear - how do you approach safety stock? Static formula or dynamic levers?

Inflation isn't just about rising prices; it's a catalyst for changing consumer behaviors. As purchasing power shifts, businesses must adapt swiftly to meet evolving demands. Hindustan Unilever Limited (HUL), a leader in the FMCG sector, showcases how embracing AI can turn these challenges into opportunities. 📌 The Challenge #HUL observed significant fluctuations in demand across its diverse product portfolio during inflationary periods. Premium products experienced slower sales, leading to overstock situations, while budget-friendly items frequently faced stockouts. Traditional forecasting methods, relying heavily on historical sales data, struggled to keep pace with these rapid changes in consumer preferences. 📊 The Solution: AI-Driven Demand Forecasting To address this, HUL integrated AI-powered analytics into its demand forecasting processes. This advanced system enabled the company to: Analyze Real-Time Consumer Behavior: By examining current purchasing patterns and consumer sentiment, HUL could detect emerging trends and shifts in preferences. Incorporate External Economic Indicators: The AI model factored in various economic indicators, such as inflation rates and consumer confidence indices, to predict their impact on product demand. Optimize Inventory Management: With precise demand forecasts, HUL adjusted its inventory levels accordingly, ensuring optimal stock across all product categories. 🔹 Key Insight: The AI-driven approach revealed that demand for budget-friendly products was increasing at a rate three times higher than traditional models had predicted, while premium product sales were declining in specific regions. 📈 The Impact 20% Reduction in Unsold Premium Stock: By aligning inventory with actual demand, HUL minimized excess stock of premium items. 35% Improvement in Stock Availability for Budget-Friendly Products: Ensuring that high-demand, cost-effective products were readily available led to increased customer satisfaction. Enhanced Revenue and Profit Margins: Optimized inventory management reduced holding costs and prevented lost sales, positively impacting the bottom line. 💡 The Lesson In times of economic uncertainty, relying solely on historical data can be a pitfall. HUL's proactive adoption of AI-driven demand forecasting exemplifies how leveraging advanced analytics allows businesses to stay agile and responsive to market dynamics, ensuring they meet consumer needs effectively How is your organization utilizing data analytics to navigate market fluctuations? #datadrivendecisionmaking #businessstrategies #dataanalytics #demandforecasting

🚨 Uncertainty is near an all-time high 🚨 Since 1985, the U.S. Federal Reserve has tracked an uncertainty index—and it's now skyrocketing, fast approaching its pandemic-era peak. But you can THRIVE in these conditions. Here are 8 ways to do it: 🔹 1. Uncertainty Matrix – Map out what’s certainly known, certainly unknown, unevenly recognized in your organization, and critical blind spots. 🔹 2. Scenarios – Develop a few truly distinct scenarios (not just based on your company’s outcomes, but on market shifts). What actions can you take today to thrive in each future scenario? 🔹 3. Portfolio Plan – Assess the risk level, risk type, and maturity of your investments. Think of it as a diversified portfolio—how will it hold up in different market conditions? 🔹 4. Platforms vs. Products – Shift from rigid products to flexible platforms. Netflix, for example, is a platform that can evolve with the market—traditional broadcast networks do not. 🔹 5. Capture New Markets – Disruptive events create major opportunities. Fintech boomed after the financial crisis—where’s your industry’s next opening? Consider all dimensions: goods companies can grow into non-tariffed services, you can expand geographically, and more. 🔹 6. Agile Planning – Static, annual strategic plans don’t work during high uncertainty. Instead, focus on dynamic strategies that separate fixed priorities from adaptable tactics. 🔹 7. Reduce Inter-Dependencies – Create modular, flexible value propositions that can have both more agility and lower costs. 🔹 8. Put Customers First – Your customers’ Jobs to be Done remain constant—use them as your North Star for strategy, cost reduction, and option development. 📚 Want to go deeper? Our materials on FutureCasting and the book Rogue Waves address approaches 1 – 4, our book Capturing New Markets tackles point 5, our book The Innovative Leader focuses on point 6, and our books Costovation and Jobs to be Done concentrate on points 7 and 8. Dig into them or get in touch for a discussion. Uncertainty = Opportunity. Seize it!! 🚀 #Leadership #Strategy #Innovation #JobsToBeDone #Growth #Agility

Because inventory bleeds cash, profit, and sanity... Here are 12 inventory parameters that planners cannot get wrong: ✅1️⃣ Safety Stock 👉 Concept: buffer to protect against variability in demand and lead time 🧮 Calculation: Z×σd×√LT ❓ Where: Z = Z-score (based on desired service level), σd = Standard deviation of demand, LT = Lead Time ✅2️⃣ Cycle Stock 👉 Concept: inventory held to meet demand between replenishment cycles 🧮 Calculation: Cycle Stock = Order Quantity / 2 ✅3️⃣ Service Level 👉 Concept: percentage of customer demand that can be met without stockouts 🧮 Calculation: calculated based on target service level factors from normal distribution ✅4️⃣ Backorder Level 👉 Concept: quantity of customer orders waiting to be fulfilled due to stock shortages 🧮 Calculation: difference between orders received and orders fulfilled ✅5️⃣ Economic Order Quantity (EOQ) 👉 Concept: ideal size order that meets demand while minimizing ordering and holding costs 🧮 Calculation: Q = √ (2DS / H) ❓ Where: D = Annual Demand in Units of a Product, S = Ordering Cost per Order, H = Holding Cost per Unit of Product ✅6️⃣ Lot Size 👉 Concept: quantity of items ordered or produced in a single batch 🧮 Calculation: EOQ or other operational considerations ✅7️⃣ Min-Max 👉 Concept: Min level triggers a reorder; Max level prevents overstocking 🧮 Calculation: Min Level = Reorder Point, Max Level = Reorder Point + EOQ or another value for the max inventory level ✅8️⃣ Lead Time 👉 Concept: Time between order with supplier and receipt of goods 🧮 Calculation: Delivery Date – Order Date ✅9️⃣ Days of Supply 👉 Concept: shows how many days of sales we are keeping in inventory 🧮 Calculation: Days of Supply = Inventory on Hand / Average Daily Usage ✅1️⃣0️⃣ Inventory Turnover 👉 Concept: number of times inventory is sold and replaced in a period 🧮 Calculation: Cost of Goods Sold (COGS)/ Average Inventory ✅1️⃣1️⃣ Reorder Point 👉 Concept: point to order before start using safety stock 🧮 Calculation: Average Lead Time X Average Daily Demand + Safety Stock ✅1️⃣2️⃣ ABC Classification 👉 Concept: segmenting inventory based on importance (A for high value, B for medium, C for low) 🧮 Calculation: based on their percentage contribution to total sales or inventory value Any others to add?

Existing stock prediction solutions mainly fall into regression, classification, and stock recommendation methods. Regression methods formulate stock prediction as a pure time-series forecasting problem, predicting future stock prices or returns by learning from historical stock time-series data. Classification methods, however, treat stock prediction as a binary up/down classification problem, developing accurate classifiers to predict stock movement. Nevertheless, general regression and classification methods for stock recommendation have a significant drawback: they are not directly optimized for the target of investment (i.e., profit maximization). Most of these methods focus on developing powerful learning models to improve investment profit while ignoring effective risk modeling. This can lead to abnormal results, such as accurate prediction models earning less profit than inaccurate models. For example, research has shown that accurate stock prediction methods (either regression or classification) like LSTM and CNN may earn less profit than inaccurate models like ARIMA and MLP. This deficiency may limit the SOTA (state-of-the-art) models' effectiveness in practical stock investing and cause significant paper losses, which may be intolerable to some investors and force them to stop investing prematurely to prevent bankruptcy. This motivates researchers to develop models that can mitigate stock selection and recommendation risks. To address the challenges highlighted above, the authors of [1] propose a novel framework called SVAT (Split Variational Adversarial Training), which leverages adversarial perturbations to reduce stock recommendation risks (i.e., a risk-aware stock recommendation). SVAT encourages the stock model to be sensitive to adversarial perturbations of risky stock examples and enhances the model’s risk awareness by learning from these perturbations. They used STHAN-SR (Spatiotemporal Hypergraph Attention Network for Stock Ranking) as the backbone recommendation model and combine it with SVAT to achieve the state-of-the-art (SOTA) results. #QuantFinance Experiments on several real-world stock market datasets demonstrate the superiority of the proposed SVAT method. By lowering the volatility of the stock recommendation model, SVAT effectively reduces investment risks and outperforms SOTA baselines (ARIMA, LSTM, GCN, RSR-E, RSR-I, ANN-SVM and STHAN-SR) by more than 30% in terms of risk-adjusted profits. The links to their paper [1] and #Python code [2] are posted in the comments.

Hospitals are making less money because of these mistakes! In healthcare, managing inventory to align with real demand is a constant challenge. With items billed to in-patients, out-patients, or not billed at all, the risk of overstock or stockouts can be high. Consider the impact of one hospital’s approach: This issue affects cost, resource allocation, and patient care. But what if healthcare facilities could analyze consumption patterns and align supply with actual demand? Here’s how leading hospitals are using data-driven strategies to reduce waste, ensure fulfillment, and cut costs. Many hospitals stock up to avoid shortages. The first step? Analyzing usage across the board. Track demand through metrics like bed days, duration of stay, department, and care provider, hospitals gain a complete view of supply needs, item by item. With this data, they can build statistical models that accurately forecast inventory levels, applying correction factors based on operational changes. Here’s how this data-driven model is transforming inventory management: 1) Demand-driven forecasting: Tracking metrics such as patient stay duration and care provider needs enables precise demand planning. 2) Item-level alignment: Each department and provider receives supplies matched to actual usage, reducing waste and unnecessary stock. 3) Correction factors: By adjusting for seasonal or operational changes, hospitals avoid costly overstocks and stockouts. 4) Financial impact: Reduced inventory costs mean more resources for direct patient care. The outcome? A supply chain where inventory is optimized, every item accounted for, and every dollar maximized. In this way hospitals save time and money to work effectively across all the channels.

Safety Stock Matrix   An easy way to bucketize Safety stock based on ABC and XYZ matrix.   Start with an ABC or pareto analysis for the SKUs, below are steps for ABC analysis. 1.   Sort cumulative sales of last 12 months in descending order of sales value. 2.   Allocate top 80% SKU as A, Next 15% as B and remaining 5% as C class. Second step is to calculate COV and categorize SKUs in XYZ categorization. 1.   Take measure COV for same set of sales data for a period of 24 months. 2.   Categorize X=30%, Y=31%-75%, Z=>75%, This range can be customized as per industry. COV is Coefficient of variance is a statistical measure of dispersion of data points in a data series around the mean. The main idea of the ABC XYZ analysis is to combine ABC and XYZ categories across two dimensions: we end up with a matrix of 9 categories. Then, we can classify items around 4 extremes: AX: High sales volumes, stable AZ: High sales volumes, very volatile CX: Low sales volumes, stable CZ: Low sales volumes, very volatile To use effectively the ABC XYZ matrix, we need to define an Inventory Management Policy: setting service level and safety stock targets. Roughly speaking, if you want a better service level, you need higher safety stock. For example, we can choose to hold more inventory for the A category, as A items are the major drivers of your business, and we want to maximize the service level for those products. For AX items, we can afford to hold less safety stock than AZ items, as we have better visibility over the demand. The same logic applies to B items. Regarding C codes, you can decide to hold low inventory for both CX and CZ categories, for two reasons: CX items have a low impact on the business, so we can afford to set a lower service level. Also, they are stable, so we require even less safety stock. CZ items are very volatile. We can think we need therefore a bit more safety stock, but we know by experience that most of the time It is not worth it: because they are both low-selling items and very unpredictable, CZ items are often a source of high stock levels and unnecessary headaches. I noticed that CZ items represent a big part of the total sleeping inventory of most companies. It might be wise to set lower service levels for this category.   This is just an example. There is no golden rule, it all depends on your own supply chain challenges. #inventorymanagement#supply chain

𝗜𝗻𝘃𝗲𝗻𝘁𝗼𝗿𝘆 𝗰𝗼𝗻𝘁𝗿𝗼𝗹 𝗶𝘀 𝗻𝗼𝘁 𝗮𝗯𝗼𝘂𝘁 𝗰𝗼𝘂𝗻𝘁𝗶𝗻𝗴 𝘀𝘁𝗼𝗰𝗸.  𝗜𝘁’𝘀 𝗮𝗯𝗼𝘂𝘁 𝗰𝗼𝗻𝘁𝗿𝗼𝗹𝗹𝗶𝗻𝗴 𝗰𝗮𝘀𝗵 𝗳𝗹𝗼𝘄, 𝗰𝘂𝘀𝘁𝗼𝗺𝗲𝗿 𝘀𝗲𝗿𝘃𝗶𝗰𝗲, 𝗮𝗻𝗱 𝗰𝗵𝗮𝗼𝘀. If you're not applying structured inventory techniques, you're inviting stockouts, overstocking, or worse—cash trapped in the wrong places. Here are 6 high-impact inventory control techniques used by top-performing supply chains: (1). ABC Analysis Categorizes items by value contribution: • A = High-value, tight control • B = Moderate-value, periodic review • C = Low-value, simple checks Focus where it financially matters most. (2). XYZ Classification Uses Coefficient of Variation (CV) to classify demand variability: • X = Stable • Y = Moderate • Z = Erratic Drives how much buffer or planning flexibility you need. (3). EOQ (Economic Order Quantity) Finds the optimal order size that minimizes total holding + ordering cost. Formula: EOQ = √(2DS/H) (4). ROP (Reorder Point) Calculates when to place the next order so you never run dry. Formula: ROP = Daily Demand × Lead Time (5). Safety Stock Holds extra inventory to cover demand or supply shocks. Formula: SS = Z × σ × √LT Z = service level, σ = demand variability (6). VED Classification Ranks inventory by criticality: • Vital – no stockout allowed • Essential – important, but manageable • Desirable – lowest priority Crucial in healthcare, aerospace, and military supply chains. 🧠 I use this exact framework when training supply chain teams or auditing stock strategies. Which technique do you use most? #InventoryManagement #SupplyChain #DemandPlanning