<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[Stories by Ujjwal Gulecha on Medium]]></title>
        <description><![CDATA[Stories by Ujjwal Gulecha on Medium]]></description>
        <link>https://medium.com/@ujjwal.gulecha?source=rss-e4476d34824a------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*Vj4twUYCYIR3-0dIxw-3Ow.jpeg</url>
            <title>Stories by Ujjwal Gulecha on Medium</title>
            <link>https://medium.com/@ujjwal.gulecha?source=rss-e4476d34824a------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Sun, 22 Nov 2026 10:12:52 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@ujjwal.gulecha/feed" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[Improve your page load latency]]></title>
            <link>https://medium.com/@ujjwal.gulecha/improve-your-page-load-latency-7af870826e22?source=rss-e4476d34824a------2</link>
            <guid isPermaLink="false">https://medium.com/p/7af870826e22</guid>
            <category><![CDATA[software-engineering]]></category>
            <category><![CDATA[performance-optimization]]></category>
            <category><![CDATA[scalability]]></category>
            <category><![CDATA[distributed-systems]]></category>
            <category><![CDATA[application-development]]></category>
            <dc:creator><![CDATA[Ujjwal Gulecha]]></dc:creator>
            <pubDate>Tue, 23 Sep 2025 22:00:12 GMT</pubDate>
            <atom:updated>2025-09-30T00:14:51.114Z</atom:updated>
            <content:encoded><![CDATA[<p>At scale, every millisecond matters. When your application serves millions of users across diverse network conditions and devices, traditional performance advice falls short. After optimizing page load performance for systems handling <strong>50M+ monthly requests</strong>, I’ve learned that sustainable performance improvements require a fundamentally different approach than what most guides suggest.</p><p>This isn’t about basic database indexing or adding Redis. We’ll take a systematic look at the architecture decisions, measurement strategies, and mobile-specific patterns that separate high-performing backend services from the rest.</p><h3>The Real Performance Problem</h3><p>Most performance discussions focus on individual techniques-caching, compression, database tuning. But at scale, the challenge isn’t knowing these techniques; it’s orchestrating them into a coherent system that remains fast under real-world conditions.</p><p>Consider this: reducing your P50 latency from 200ms to 150ms sounds great, but if your P99 jumps from 500ms to 2s because of cache stampedes, you’ve made the user experience worse for the users who matter most-those on the edge of your performance envelope.</p><p><strong>Real-world example</strong>: Discord scaled individual servers from tens of thousands of concurrent users to approaching two million concurrent users in the past few years. Their biggest challenge wasn’t individual slow queries-it was managing the quadratic scaling problem where fanout work grows exponentially with server size. With 100,000 people online, a single message becomes 10 billion notifications to deliver <a href="https://discord.com/blog/maxjourney-pushing-discords-limits-with-a-million-plus-online-users-in-a-single-server">[1]</a>.</p><h3>Step 1: Build Performance-First Observability</h3><h3>What to Measure That Actually Matters</h3><p><em>“You can’t improve what you can’t measure.”</em> — Often attributed to Peter Drucker. This is extremely important when you are solving latency issues. If you don’t know what the breakdown of page load latency is, you will not be able to solve the problem effectively.</p><p>Always start with the user and how the user experiences things. In your case, it could be an actual person or it could be another microservice calling your endpoint.</p><h3>Add Observability at All Stages</h3><ol><li><strong>End-user to service</strong> — Record the time from firing a request to getting a response back. Capture P50, P90, P99 and P999 latencies. Users complain about the tail (P99/P999), not the average. Google’s “The Tail at Scale” paper is a must-read <a href="https://research.google/pubs/the-tail-at-scale/">[2]</a>.</li><li>Measure not just latency, but also error budgets (SLOs from SRE practices <a href="https://sre.google/sre-book/service-level-objectives/">[3]</a>). If 99.9% of requests need to be &lt;200ms, observability should highlight where that budget is getting burned.</li><li><strong>Client-side latency</strong> — Measure how much time it took to decode data, deserialize it, and render the UI. Should also segment performance by device type, network conditions, app version, and geographic region. A 200ms response time on WiFi is excellent; on 3G in rural areas, it might cause timeouts.</li><li><strong>Within your service</strong> — Break down time taken for each component. If you have a DAG workflow, measure time per node, and identify the critical path. Distributed tracing (via OpenTelemetry or Jaeger) is key for microservices. For example, if one slow DB call in a DAG delays the entire critical path, tracing highlights it instantly.</li><li><strong>System-level metrics</strong> — CPU utilization, memory usage, garbage collection, threads blocked, threads queued, disk I/O, and network bandwidth.</li></ol><p>Tools like <strong>OpenTelemetry</strong>, <strong>Datadog</strong>, or <strong>New Relic</strong> can help set up structured observability pipelines. With this you can get an idea where the problem is — user device? Network? Database? server?</p><p><strong>Pro tip</strong>: Set up error budgets alongside latency monitoring. If 99.9% of requests need to be under 200ms, your observability should highlight exactly where that budget is getting burned.</p><h3>Step 2: Systematic Optimization Strategies</h3><h3>1. Compression — Smaller Payloads, Faster Loads</h3><p>Enable <strong>Brotli compression</strong> (preferred over Gzip on modern devices) to reduce payload size. But always experiment: compression and decompression cost CPU cycles, and for very small payloads it might not help.</p><ul><li><strong>Dynamic vs static compression</strong>: Static for assets (JS, CSS, images), dynamic for JSON/HTML.</li><li><strong>Skip already compressed formats</strong>: JPEG, PNG, and MP4 don’t benefit from recompression.</li></ul><p><strong>Case study</strong>: LinkedIn’s adoption of Brotli reduced JSON payloads by ~14% on average <a href="https://www.linkedin.com/blog/engineering/optimization/boosting-site-speed-using-brotli-compression">[4]</a>.</p><h3>2. Smart Pagination and Progressive Loading — Don’t load everything at once</h3><p>The first screen users see is critical. If it’s slow, they bounce. Focus on above-the-fold prioritization-load only what’s visible, fetch the rest later.</p><ul><li><strong>Cursor-based pagination</strong> (more reliable than offset-based when data changes)</li><li><strong>Skeleton UIs and shimmer loaders</strong> (Facebook and YouTube use these to make waiting feel faster)</li><li><strong>Progressive image loading</strong> with placeholder blur effects</li></ul><p>Remember: Pagination is as much about perception of speed as actual latency.</p><h3>3. Caching — Fast but Requires Strategy</h3><p>A great caching strategy can cut load times by orders of magnitude. A poor one can cause stale data, stampedes, or outages.</p><p><strong>Multi-layer approach</strong></p><ul><li><strong>L1 caching</strong>: In-memory (fast but local)</li><li><strong>L2 caching</strong>: Distributed Redis/Memcached</li><li><strong>CDN caching</strong>: Geographic distribution with smart invalidation</li></ul><p><strong>Advanced patterns</strong></p><ul><li><strong>Stale-while-revalidate</strong>: Serve cached data immediately, refresh in background</li><li><strong>Jittered TTLs</strong>: Prevent cache stampedes when many keys expire simultaneously</li><li><strong>Read-through vs write-through</strong> strategies based on your consistency needs</li></ul><p><strong>Case study</strong>: How Facebook Scaled and Optimized for Massive Request Volumes <a href="https://blog.bytebytego.com/p/how-facebook-served-billions-of-requests">[5]</a></p><p>For a deep dive, see the <strong>Caching Best Practices</strong> guide from AWS <a href="https://aws.amazon.com/caching/best-practices/">[6]</a></p><h3>4. Code Optimization — Hunt Down Inefficiencies</h3><p>You might be surprised/not surprised how much inefficiency lies in production code.</p><ul><li><strong>Profile hot paths</strong> using tools like perf, py-spy, jvisualvm.</li><li><strong>Optimize parsing/serialization</strong> (e.g., switch to simdjson for faster JSON parsing).</li><li><strong>Avoid quadratic loops</strong> and unnecessary allocations.</li><li><strong>Use object pooling</strong> where it helps with high churn objects.</li></ul><p>Netflix, for example, optimized its JSON-to-UI rendering pipeline and <strong>shaved 200ms off median latency</strong> <a href="https://netflixtechblog.com/seamlessly-swapping-the-api-backend-of-the-netflix-android-app-3d4317155187">[7]</a>.</p><p><strong>Real-world example from production</strong>: At DoorDash, we encountered a counterintuitive optimization case. A Kotlin service was processing multiple lists of objects (list1, list2, list3, list4, list5) that needed to be cross-referenced by identifiers to produce a final aggregated list.</p><pre>// Initial &quot;obvious&quot; optimization - but turned out to be slower<br>fun processOrdersOptimized(<br>    restaurants: List&lt;Restaurant&gt;,<br>    orders: List&lt;Order&gt;,<br>    drivers: List&lt;Driver&gt;,<br>    customers: List&lt;Customer&gt;,<br>    locations: List&lt;Location&gt;<br>): List&lt;DeliveryInfo&gt; {<br>    val restaurantMap = restaurants.associateBy { it.id }  // Hash computation bottleneck<br>    val driverMap = drivers.associateBy { it.id }          // Hash computation bottleneck<br>    val customerMap = customers.associateBy { it.id }      // Hash computation bottleneck<br>    val locationMap = locations.associateBy { it.id }     // Hash computation bottleneck<br><br>    return orders.map { order -&gt;<br>        DeliveryInfo(<br>            restaurant = restaurantMap[order.restaurantId]!!,  // O(1) lookup<br>            driver = driverMap[order.driverId]!!,              // O(1) lookup<br>            customer = customerMap[order.customerId]!!,        // O(1) lookup<br>            location = locationMap[order.locationId]!!         // O(1) lookup<br>        )<br>    }<br>}<br><br>// What actually performed better - linear search on small lists<br>fun processOrdersFaster(<br>    restaurants: List&lt;Restaurant&gt;,<br>    orders: List&lt;Order&gt;,<br>    drivers: List&lt;Driver&gt;,<br>    customers: List&lt;Customer&gt;,<br>    locations: List&lt;Location&gt;<br>): List&lt;DeliveryInfo&gt; {<br>    return orders.map { order -&gt;<br>        DeliveryInfo(<br>            restaurant = restaurants.first { it.id == order.restaurantId }, // O(n) but faster<br>            driver = drivers.first { it.id == order.driverId },             // O(n) but faster<br>            customer = customers.first { it.id == order.customerId },       // O(n) but faster<br>            location = locations.first { it.id == order.locationId }       // O(n) but faster<br>        )<br>    }<br>}</pre><p>The “obvious” optimization was converting lists to HashMaps for O(1) lookups instead of O(n) linear searches:</p><p>The issue? <strong>Hash computation overhead</strong>. These complex domain objects had expensive hashCode() implementations, and the CPU cost of computing hashes for map creation exceeded the benefit of O(1) lookups. With small list sizes (typically &lt;50 items), linear search was significantly faster overall.</p><p><strong>Key takeaway</strong>: Big O notation assumes operations have constant cost, but in reality, the “constant” matters enormously. Always profile real workloads-theoretical optimizations can become practical pessimizations.</p><h3>5. Smart Parallelization</h3><p>Parallelizing workloads can dramatically reduce latency when handling multiple independent dependencies. The key is doing it safely.</p><p><strong>What works:</strong></p><ul><li>Async/await patterns for I/O-bound operations</li><li>Parallel processing of independent API calls</li><li>Background job processing for non-critical tasks</li></ul><p><strong>What to avoid</strong></p><ul><li>Over-parallelization that saturates database connection pools</li><li>Missing backpressure controls that can overwhelm downstream services</li></ul><h3>6. Database Query Optimization — The Usual Suspect</h3><p>Databases are frequently the bottleneck, but the solutions go beyond adding indexes.</p><ul><li><strong>Optimize for reads</strong> — Design your schema and queries for read performance, even if writes become more complex.</li><li><strong>Avoid N+1 queries</strong> — Batch database requests whenever possible</li><li><strong>Materialized views</strong> — Precompute expensive query results</li><li><strong>Sharding strategies</strong> — By user ID (balanced load) or geography (latency reduction).</li></ul><p><strong>Case study</strong>: Instagram scaled feeds to billions of users by sharding Postgres across IDs <a href="https://www.google.com/search?q=Instagram+Engineering+-+Scaling+Feeds+with+Postgres+Sharding&amp;oq=Instagram+Engineering+-+Scaling+Feeds+with+Postgres+Sharding&amp;gs_lcrp=EgZjaHJvbWUyBggAEEUYOTIHCAEQIRigATIHCAIQIRiPAtIBBzE5M2owajSoAgCwAgA&amp;sourceid=chrome&amp;ie=UTF-8">[8]</a>.</p><p><strong>References</strong>: Use the Index, Luke! <a href="https://use-the-index-luke.com/">[9]</a>, [Postgres Performance Tips] <a href="https://www.postgresql.org/docs/current/performance-tips.html">[10]</a>.</p><h3>Wrapping Up</h3><h3>The Systematic Approach</h3><ol><li><strong>Measure comprehensively</strong> — Set up observability across your entire stack</li><li><strong>Identify the biggest bottleneck</strong> — Use data, not assumptions</li><li><strong>Apply targeted optimizations</strong> — Focus on highest-impact changes first</li><li><strong>Measure the results</strong> — Validate that your changes actually improved user experience</li><li><strong>Repeat</strong> — Performance optimization is an ongoing process</li></ol><h3>Success Metrics That Matter</h3><p>Track both technical and business metrics:</p><ul><li><strong>Technical</strong>: P99 latency reduction, error rate improvements, throughput increases</li><li><strong>Business</strong>: Bounce rate reduction, conversion rate improvements, user engagement increases</li></ul><p><strong>Key insight</strong>: A 100ms improvement in page load time can increase conversion rates by 1–2% for e-commerce sites. At scale, this translates to millions in revenue impact.</p><p>Performance optimization at scale is about <strong>systems thinking</strong>, not just individual optimizations. Start with comprehensive observability, then systematically address bottlenecks based on real user impact data.</p><p>As Jeff Dean from Google puts it: <em>“If you want your system to be fast, first make it correct. Then profile, measure, and optimize the hot spots.”</em></p><p>Whether you’re building mobile apps, APIs, or distributed systems, the principles remain the same: <strong>measure, experiment, optimize, and repeat</strong>. The difference at scale is that every optimization needs to work harmoniously with all the others-and that’s where the real engineering challenge lies.</p><p><strong>References:</strong></p><p>[1] <a href="https://discord.com/blog/maxjourney-pushing-discords-limits-with-a-million-plus-online-users-in-a-single-server">Discord Engineering Blog — Scaling Message Fanout</a> <br> [2] <a href="https://research.google/pubs/the-tail-at-scale/">Google Research — The Tail at Scale</a> <br> [3] <a href="https://sre.google/sre-book/service-level-objectives/">Google SRE Book — Service Level Objectives</a> <br> [4] <a href="https://www.linkedin.com/blog/engineering/optimization/boosting-site-speed-using-brotli-compression">LinkedIn Engineering Blog — Adopting Brotli Compression</a> <br> [5] <a href="https://blog.bytebytego.com/p/how-facebook-served-billions-of-requests">Facebook Engineering — Scaling and Optimizing for Request Volumes</a> <br> [6] <a href="https://aws.amazon.com/caching/best-practices/">AWS Best Practices — Caching Strategies</a> <br> [7] <a href="https://netflixtechblog.com/seamlessly-swapping-the-api-backend-of-the-netflix-android-app-3d4317155187">Netflix Tech Blog — JSON-to-UI Pipeline Optimization</a> <br> [8] <a href="https://www.google.com/search?q=Instagram+Engineering+-+Scaling+Feeds+with+Postgres+Sharding&amp;oq=Instagram+Engineering+-+Scaling+Feeds+with+Postgres+Sharding&amp;gs_lcrp=EgZjaHJvbWUyBggAEEUYOTIHCAEQIRigATIHCAIQIRiPAtIBBzE5M2owajSoAgCwAgA&amp;sourceid=chrome&amp;ie=UTF-8">Instagram Engineering — Scaling Feeds with Postgres Sharding</a> <br> [9] <a href="https://use-the-index-luke.com/">Use the Index, Luke! — SQL Performance Explained</a> <br> [10] <a href="https://www.postgresql.org/docs/current/performance-tips.html">PostgreSQL Performance Tips — Official Documentation</a></p><p><em>Originally published at </em><a href="https://www.ujjwalgulecha.com/performance/systems/engineering/2025/09/23/improve-your-page-load-latency.html"><em>https://www.ujjwalgulecha.com</em></a><em> on September 23, 2025.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=7af870826e22" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Academy Awards biased towards some genres?]]></title>
            <link>https://medium.com/movie-time/academy-awards-biased-towards-some-genres-bd8eec9c82b7?source=rss-e4476d34824a------2</link>
            <guid isPermaLink="false">https://medium.com/p/bd8eec9c82b7</guid>
            <category><![CDATA[oscars]]></category>
            <category><![CDATA[bias]]></category>
            <category><![CDATA[movies]]></category>
            <category><![CDATA[academy-awards]]></category>
            <category><![CDATA[data-science]]></category>
            <dc:creator><![CDATA[Ujjwal Gulecha]]></dc:creator>
            <pubDate>Sun, 09 Jul 2017 06:05:50 GMT</pubDate>
            <atom:updated>2018-06-19T03:52:59.780Z</atom:updated>
            <content:encoded><![CDATA[<p>My educated guess is that the Academy Awards are not biased towards specific genres. The Academy of Motion Picture Arts and Sciences (AMPAS), a professional honorary organization, maintains a voting membership of 5,783 as of 2012 [1]. This organization votes for the movie to be chosen as the best movie. I think that since it is not decided by just a panel of very limited members, but a group of size 6000 members, the possibility of bias reduces substantially.</p><h3>Introduction</h3><p>I think this question is of great interest as the Academy Awards are considered the most prestigious awards in the movie industry and it is quintessential for it to be bias-free. There is no official organization or group that keeps a check if there exists bias in the awards, thus it is essential to find out whether or not that is the case.</p><p>There have been various reports indicating that the Academy Awards are biased to- wards specific genres like romantic historical epics, biographical dramas, romantic dramedies, and family melodramas[2]. This motivated me to find out if this is true or is it a speculation. There have been studies that have tried to answer this question, for example IbTimes found that Sci Fi, Foreign Films And Horror Movies never get Best Picture Nominations. However, these studies establish that the awards are biased without considering public opinion of genres of movies. I thought that in order to truly determine if the organization is biased towards some genre, it would be crucial to check if the general movie-watching population thinks the same about those genres or not.</p><h3>Methodology</h3><p>These are the steps I went through for answering my question:<br> 1.) Extracted the genres of the movies which have received academy awards as well as nominations for the best movie since the inception of the awards, i.e 1929. I did this using using IMDBpy. It is a python library, to which when we feed in a list of movies, we can extract various information about the movie like language, cast, director, genre and so on.</p><p>2.) Extracted the genres of the movies which are rated the highest by users on IMDB. IMDB maintains a list of top 250 highest rated movies by users. I used this data to com- pare with the data I have extracted from academy awards list. Then using the multiple charts visualization, I compared the two set of data. The different properties are the dif- ferent genres like drama, comedy, historical and so on. The x-axis in each of the graph represent the data collected from Academy Awards and the data collected from IMDB. The y-axis will indicate the percentage of movies that are of a particular genre. If there is a big discrepancy between the two for multiple categories, then we can establish the fact that there exists a bias. The reason why this comparison is important is that if suppose like the reports suggest that drama, romantic, historical based movies win more awards, and if the majority of the public (IMDB Top 250) also thinks the same, then it can be said that its not that the organization is biased, it is just that those specific genres are appreci- ated more by everyone. However, if there is a big difference between them, it can indicate a bias.</p><h3>Results</h3><p>Here are the results which I found out.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*8GopbNFUYiAgPgEY2JrXdA.png" /><figcaption>IMDB Top 250 and Academy Awards</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*MUk9u2Ard4BUvMb-lqB8Ig.png" /><figcaption>IMDB Top 250 and Academy Awards</figcaption></figure><p>As we can see from the results that Academy Awards seem to be biased against some genres. Few points to notice from this result:<br> 1. Sci-Fi and Horror Movies have not won an award nor have gotten any nomination.<br> 2. Academy Awards seem more biased towards the Biography, Drama, History, Musical and Romance Genres.<br> 3. Academy Awards seem to be biased against the Action, Adventure, Crime, Fantasy, Mystery and Thriller Genres.<br> 4. The genres which seem to match with general user’s ratings are Comedy, Family, Film- Noir, Music, Sport, War and Western Categories.</p><h3>Potential Pitfalls</h3><p>Potential pitfalls I think I could have made during this analysis are:<br> 1. The assumption that I have made is that IMDB Top 250 Movies list reflects the choice of the majority of the population. This may not be the case. A lot of people would not prefer IMDB and not vote on it. This might make the results not accurate. To counter this, we can gather as much as data as possible to eliminate bias. We can gather data from other websites like rotten tomatoes, meta critic and so on. To gather more data, we could also make an interactive Facebook quiz wherein people would answer what their top 20 favorite movies were and use that data to add to our repository of data.</p><p>2.) It might be hard to decide the boundary between declaring the awards are biased or not biased. For example, if the difference of percentages in the drama category is 10%, but for other categories, the difference is ¡5%, should we declare that the awards are bi- ased or not? To workaround this issue, we could use various techniques in statistics like cosine similarity to check how different two sets of data are and then with this help we could make a decision more accurately.</p><h3>Conclusion</h3><p>My initial educated guess turned out to be false according to the results I have obtained. There seems to be a bias with regard to some genres. Specially Sci-Fi and Horror movies which have not even got a nomination. I cannot think of a reason why this would have happened. As for the other categories, some of the genres like Drama and Romance have got a higher number of nominations and wins than other genres. I think this can be attributed to the fact that people in general favor these categories. I would want to do more research on this if possible.</p><h3><strong>References</strong></h3><p>[1] <a href="http://www.ibtimes.com/oscar-genre-bias-why-sci-fi-foreign-films-horror-movies-never-get-best-picture-1824032">Web Archive</a></p><p>[2] Wikipedia</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=bd8eec9c82b7" width="1" height="1" alt=""><hr><p><a href="https://medium.com/movie-time/academy-awards-biased-towards-some-genres-bd8eec9c82b7">Academy Awards biased towards some genres?</a> was originally published in <a href="https://medium.com/movie-time">Movie Time Guru</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Zodiac signs of the richest people on earth]]></title>
            <link>https://medium.com/@ujjwal.gulecha/zodiac-signs-of-the-richest-people-on-earth-bb85664dcdee?source=rss-e4476d34824a------2</link>
            <guid isPermaLink="false">https://medium.com/p/bb85664dcdee</guid>
            <category><![CDATA[forbes]]></category>
            <category><![CDATA[zodiac]]></category>
            <category><![CDATA[billionaires]]></category>
            <category><![CDATA[data-science]]></category>
            <category><![CDATA[astrology]]></category>
            <dc:creator><![CDATA[Ujjwal Gulecha]]></dc:creator>
            <pubDate>Sun, 09 Jul 2017 05:35:01 GMT</pubDate>
            <atom:updated>2017-07-09T06:09:54.696Z</atom:updated>
            <content:encoded><![CDATA[<p>I was getting a little bored on the weekend and thought of analyzing the zodiac signs of the richest people on earth.</p><p>I obtained the top 100 richest people list from <a href="https://www.forbes.com/billionaires/list/">Forbes</a>. This data was taken on July 8 2017. After processing this data and converting it to useful information; I found some surprising results from this small research that I did.</p><p>The most common zodiac signs were <strong>Leo</strong> and<strong> Libra</strong>. These two signs together accounted for around a <strong>quarter</strong> of the top 100 billionaires! If the spread of billionaires were distributed equally amongst all zodiac signs, each sign would account for roughly <strong>8.33%</strong> of the entire list. However, these two signs were dominant in the top 100. The least frequently occurring zodiac sign was <strong>Sagittarius</strong>. Below is a bar chart depicting the percentage of top 100 billionaires in the respective zodiac sign.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*m9oobhM5QVeCB_uObevplw.png" /><figcaption>Zodiac sign distribution</figcaption></figure><p>There was another notable observation as well: The first Cancer billionaire to appear on the list was <strong>rank 69</strong> and the first Sagittarius billionaire was at <strong>rank 75</strong>. Co-incidentally these two signs were also the least common zodiac signs amongst the billionaires.</p><p>Few notes about the data:</p><ol><li>There were totally 100 data points. However there were some which could not be included as their date of births were not available online. These people are: <em>Karl Albrecht Jr. and Beate Heister, Maria Franca Fissolo, Joseph Safra, Theo Albrecht, Jr., James Simons, Wang Wei, Hinduja family, David and Simon Reuben Thomas and Raymond Kwok, Pallonji Mistry, Wang Wenyin, Thomas Peterffy and Iris Fontbona.</em></li><li>Maybe if the data for the above people were available, there could be a more equally distribution or it could get even more skewed. There is no answer for that unless we have that data.</li><li>Please do not use this data in any way to predict your or any other person’s chances of being a billionaire. This article was written just for entertainment purposes :)</li></ol><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=bb85664dcdee" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>