Dimitris Gatsos

Betting Account Performance Breakdown

Project Overview

This project was a routine part of my duties as a Data Analytics Assistant within a local Value Betting Syndicate (You can find the source file here). Tasks included:

• Collecting and converting PDF wagering histories to Excel.
• Verifing data accuracy and cleaning data.
• Processing data as needed (datetime segmentation, odds standardization, calculated fields).
• Calculated key metrics like ROI and P/L using pivot tables.
• Incorporating calculated metrics into dashboard.

---

Data Processing Steps

During the data processing phase of this project, several key steps were undertaken to refine and structure the dataset for in-depth analysis, including:

1. Seperated Datetime column into Date and Time for detailed analysis.

2. Refined column labels for better interpretability and clarity.

3. Extracted and standardized odds data for consistency.

4. Cleanup Selection Column to remove irrelevant information.

5. Standardized data via 'Find and Replace' to unify the bet_type and result columns.

6. Extracted and cleansed market information from the event column for data integrity.

7. Profit and Loss (P/L) calculation into a new column to measure financial performance.

8. Introduced Live Bet column to mark live bets based on the selection column data.

9. Calculated ommited data, where possible, to enhance dataset reliability and completeness.

10. Created Odds Buckets to categorize bets and analyze outcomes.

11. Classified Market Category to categorize bets and analyze market-specific trends.

The processed document went from this:

To this:

---

Analysis and Metrics Calculation

Next, I used pivot table functionalities to calculate key metrics to assess the strategic effectiveness and financial health of the betting account, including:

1. Return on Investment (ROI): Evaluates the profitability of betting activities.
2. Picks Volume: Reflects the frequency of betting activity
3. Hit Rate: Measures the precision of bet selections.
4. Profit/Loss (P/L): Financial summary indicating net outcomes.
5. Average Stake: Reveals the standard bet size, indicating the risk tolerance.
6. Average Odds: Offers insights into the risk-return profile.

Below we see each of these in comparison to their twelve month average for a quick performance comparison of the account.

---

Performance Evaluation

This specific account showed around average results on 'Hit Rate', 'Average Stake' and 'Average odds'. However, the amount of bets this account placed (47% above average) and a rare ROI performance (60% above average) drove its high profitability compared to a 12-months average P/L. The high ROI, coupled with the insights from the 'Monthly Growth' in the dashboard below might indicate that the account was in a pick in variance.

The dashboard below effectively captures all the critical information. It allows for advanced filtering, making it easy to interactively explore and analyze the account’s performance in detail.

For example, we can easily see above that 'Handicaps' were responsible for most of the P/L and almost 2/3 of the picks on this account.

While below we can easily see that the Odds bucket 1.5-1.99 in December returned 110% of the total P/L. That is how a great run looks like!

---

Project Impact

This project, which I managed end-to-end on my own, transformed from a bi-monthly task to several times within the same period. Primarily focused on reporting, my work facilitated smoother operations and informed decision-making across the organization.

---

Thanks for reading. Some information have been altered to ensure privacy. Feel free to reach out for any comment.

Close Window

Analyzing Value in the 'Star Wars Unlimited' Collection: A Data-Driven Approach

Project Overview

Welcome to our analytical journey through the galaxy of "Star Wars Unlimited" card game! After following my brother;s passion for card/ board games and his advise to purchase a "Spark of Rebellion" case full of collectibles, I wanted to:

• Better understand the economic value of our collection.
• Identify possible strategies for reselling part of our collection with a break even return on our investment in mind.

Therefore, I asked myself:

• What is the total estimated market value of our collection, and how does this breakdown by card rarity and type?
• What is the distribution of card values within each rarity category AND how does rarity affect the card's value?

To answer these I performed a detailed value analysis against the vibrant marketplace of cardmarket.com. To keep everything organized and make sense of it all, I used Excel. It helped me catalog every card and set up a dashboard. (You can find the source file here).

Some snapshot insights include:

• We have a collection predominantly composed of cards valued under 3 euros, yet boasting an overall approximate expected market value of 770€.
• A mere 9 cards, out of the 2304 purchased, belonging in the above 15€ bracket hold a treasure trove worth approximately 390€.
• The initial investment was 550€.

---

Card Indexing and Market Search

At the start of our project, we took every card in our "Star Wars Unlimited" collection and listed them in an Excel spreadsheet. For each card, we also found its current selling price on the "https://www.cardmarket.com/" website. This way, we could see both the name of each card and how much it might sell for, all in one place.

To make our spreadsheet even more useful, we used Excel's conditional formatting feature. This let us color-code the cards. With different colors, we could quickly tell which cards were available for certain game functions, which ones were rare or common, and how their market prices compared. This color-coding made it easy to spot patterns and differences at a glance, setting us up nicely to dive deeper into analyzing our collection. This first step of organizing and pricing was key to getting a clear picture of what our collection looked like in terms of variety and potential value.

This is how part of our Excel Sheet looks after we are done:

---

Analysis and Metrics Calculation

Moving forward with our analysis, we honed in on two critical questions:

First,what is the total worth of our "Star Wars Unlimited" collection, and how is this value distributed?

At the time of our market search, our total expected value amounted to 776.51€. Of course, this is only an approximate evaluation and does not take into account transportation, packaging, and other potential costs.

Now, on value distribution, we first checked total value by type.

And then, total value by rarity.

Total value, though, is only half the picture. In how many cards is this value distributed? Normal cards seem to hold much value, almost 47% of the total expected value, but would it be a viable strategy to try and harvest this value?

We needed a bit more information:

Okay, percentage-wise, normals do hold a decent amount of value, but €364 in 2039 cards could be a tricky play. With an average of 0.17€ per card, selling them in bulk seems like the only way. That's a good insight, but probably not our first strategy to pursue; we'll need to keep looking!

So then, what are some effective strategies for extracting value through reselling parts of our collection without compromising our competitive edge in the game?

Strategy number one comes directly from here:

Now, this is valuable information. The top 5 cards represent €323 or 41.6% of the total expected value. Interestingly, the Darth Vader card, which is allowed up to three times in play according to game rules, appears three times in our collection. Given Darth Vader's popularity, this could be a significant advantage. Though it means I'd need to stick to playing Rebel, it's a trade-off I'm willing to make. We might be onto a winning strategy here.

Then, we can either focus on selling a few other valuable cards:

Or, utilising some bulk selling.

---

Wrapping up

As we wrap up our look at the "Star Wars Unlimited" cards, we learned a lot about their worth and how to smartly sell some without spoiling the fun of the game. This project showed us how to balance making money with keeping the game exciting. I had a great time figuring out my collection's value, just as I love playing the game. These lessons on card value are something I'll use as I keep collecting and playing. This adventure into the world of card collecting was as fun as it was educational, and I can't wait to put what I've learned into action.

---

Thanks for reading. Feel free to reach out for any comment.

Close Window

Hubspot BGpatrasso

After helping in a critical migration error from Capsule to HubSpot CRM, a project found through Upwork, I was hired as a HubSpot Administration & Implementation support. This led me to learn more for the CRM through its Academy and compile an easy solution for a real business. This project is the sales pitch I prepared for solving the real life business problem.

---

This project's presentation is still on development. Further information will be published soon.

Close Window

Impact of Seasonal Fluctuations on Used Car Prices in Greece

Project Overview

In this project, the aim was to explore the data and see if car prices indeed rise in the winter months and drop in the summer. Along the way, I also hoped to improve my Python skills, especially when it came to web scraping and data analysis. Inspired by the thought of selling my car the next year and after hearing a similar advice from a merchand friend, I decided to gather some data to have a more solid understanding of the market trends and pricing. So, I decided to use python and Selenium to scrap data from Greece's largest used car website in 3 month intervals in 2024.

Here's a snapshot of the web scraping script I used to collect the data:

---

First steps processing and cleaning the data

Once the data was successfully scraped, the next step was to clean and process it for analysis.

First, I needed to consolidate the CSV files where the scraped data was saved. Since I scraped thousands of pages each time, I often hit the server's request limit and had to pause before resuming the process. To prevent losing all the data in case the server banned me or the script failed, I chose to break the data into several smaller files. This way, I could avoid starting over from scratch.

Next I moved on on loading and cleaning the data. This involved removing duplicates, handling missing values and outliers, and formatting the data for easier manipulation. Up until this point, I hadn't worked with Jupyter Notebooks, but from here on, I switched to using them to make data handling and manipulation more efficient. The interactive environment of Jupyter made it easier to test code snippets, visualize the data, and keep track of my progress as I cleaned and analyzed the dataset. Here's how I approached the data processing phase:

First, I installed and imported the necessary libraries:

Then, imported the data and took a first look:

Decided to drop 'date' and 'pagenumber' columns, since they were only helper columns for scrapping and not usefull anymore. Also, renamed a few columns for clarity

Next, I checked data types and shape:

Many unexpected objects types, so I checked for missing values before I attempt to .describe():

The scrapper did a good job of not leaving blank values. So, I formatted the data types appropriately:

And took a look at the data again:

Next, looked for duplicates and deleted them:

Out of the initial 305,595 rows, the data was reduced to 282,995 after deduplicating, meaning 22,600 entries were deleted—about 7.39% of the original dataset. This reduction is reasonable, given the high volume of pages scraped and the fact that the website mixed in a couple of ads with the listings using a dynamic object, making it impossible to filter them out while scraping frontend with Selenium.

Next, I looked for outliers in the price(€), mileage(km), and engine(cc) columns. I started by visualising the price(€) column for a first look:

Listings costing under 1500€ where excluded during scrapping. There are a few very expensive cars close to 1m euros, that are definatelly outliers, and many that according to the right whisker of this boxplot are branded outliers. I thought it would be helpfull to see the price of that right whisker:

Almost 32000€, which felt too low to exclude all cars above it, perhaps something around the 100k mark is more reasonable for the car market. Retrieved the number of listings above 100k euros, 2016 listings or the 0.71% of our initial data, seemed as a safer option for exlusion. Followed the same process for the 'mileage(km)' column, where I excluded the 1st and 99th percentile, and 'engine(cc)' column, where I excluded anything below the 0.4% and 99.9% of the data, and ended up with 274192 listings:

---

Exploratory Data Analysis (EDA)

By this point, I had already gotten a first look at the data and was ready to dive deeper now that it was cleaned:

First look in the clean data:

• The median price of €11,000 suggests that typical cars in the dataset are more affordable, while the higher mean price of €14,425 indicates that a few expensive cars are skewing the average upward.
• Mileage: The mileage data shows that most cars have significant usage, with a median mileage of 133,300 km. The high mileage is typical for a used car dataset.
• Engine Size: Most cars in the dataset fall within the typical range of 1,300 cc to 1,800 cc.

Visualised the price distribution, and saw that most car prices fall between €5,000 and €20,000. The distribution is right-skewed, indicating a small number of high-priced cars:

Followed, similar process for the engine sizes and found that the majority of cars have engines between 1,000 cc and 2,000 cc, which is typical for personal vehicles:

And the same for mileage, where most cars have mileage between 50,000 and 200,000 kilometers, with very few having extremely high or low mileage:

Then, I looked at my categorical data, starting with fuel types, where by far the most common fuel types are Petrol and Diesel, consisting almost the 91% of the total listings. Another noticable insight here, is that almost one in every twenty cars sold, use Gas/lpg as a fuel:

Lastly, I looked at the number of listings by season. They are fairly even across seasons, with a slight increase during the Spring:

---

Testing the Seasonal Pricing Hypothesis

After gaining a better understanding of the dataset through the EDA, it was time to focus on testing my initial hypothesis: whether car prices tend to rise in the winter and drop in the summer. For that, I calculated and visualized the mean and median car prices for each season

From this, I could see that:

• The mean prices across the three seasons—Winter (€14,290), Spring (€14,256), and Summer (€14,745)—are relatively close. The mean increase from winter to summer of 3.18%, and the median increase in the same period suggests that, overall, car prices are fairly stable throughout the year, with no drastic seasonal fluctuations.
• More, it contradicts the initial hypothesis, that car prices tend to rise in the winter. The data at hand show the exact opposite; car prices increase during the summer.

So, what if I tried to make similar comparisons in various segments of the data?

I intially segmented the data based on engine sizes, using the Greek vehicle tax categories as a framework:

And used a bar graph to quickly view them:

Then, grouped by 'Engine Category' and 'Season'; calculated the mean and median of 'price(€)', and visualised them:

Both graphs are consistent in indicating that car prices tend to increase from Winter to Summer. This suggests a seasonal trend where prices rise as we move toward the summer months, which is again contradicting my initial hypothesis. What is more, the flatter trend for smaller engine categories (A, B) in the second and third graph suggests that these categories don't contribute much to the overall price increases from the winter to summer.

To explore further segments of my dataI followed the same logic starting with mileage categories and season:

And then segmenting to price categories and combine them with season:

Staying consitent with the analysis so far, across both mileage and price categories, car prices tend to increase from Winter to Summer. This trend is visible whether we look at mean or median values. Although, not all sub-groups contribute the same to the total increase, some lines are flatter than others.

---

Concluding remarks

I truly enjoyed the entire process and learned a lot along the way. Interestingly, the initial hypothesis—that car prices tend to rise in the winter and fall during the summer—was not supported by the data I collected. In fact, the opposite seemed to be true: prices actually increased during the summer months.

Some key insights in bite sizes:

• The overal Mean price tends to increace by 3.18%, while the Median in the same period increases by 4.36%.
• Cars with larger engine sizes appear to drive the mean price increase from winter to summer.
• In general, the mean tends to increase more sharpy than the median, suggesting that a small amount of listings with extreme values in mileage, price, and engine sizes still have a noticeable impact.

There is plenty to do to expand this project, like including cars priced below 1,500€, which were excluded from this round, and explore motorbikes as well. Definatelly, I could try to collect more data over a longer period of time, perhaps scraping more frequently, like once a month, and gather data from different dealers. Additionally, I could dive deeper into segmenting the analysis—such as focusing on cheaper cars with high mileage throughout the year. I hope to have the time and data to further explore the topic.

---

Thanks for reading. Feel free to reach out for any comment.

---

Close Window

Evaluating a wagering strategy through the comparative operformance of nine accounts in the span of 6 months.

Project Overview

This is a project comparing wagering performance from different accounts. It spans from January 2022 to June 2022, although these are not the actual dates of the wagering (have been changed for privacy reasons), the data were agthered in the span of 6 months. Along with the actual dates, various other sensitive data have been altered for privacy reasons. (You can find the source file here).

This is a project steming from my previous employment as a Data Analytics Assistant. It was rather a rare way of analysing data, since usually each account would follow different strategies simultaneously. Therefore, this is actually the first step for gathering data and evaluating a given strategy. I was be responsible for collecting, processing and presenting the data to the stakeholders responsible for evaluating the performance compare to overall portfolio and current market situation. More specifically, I:

• Collected and consolidated various PDF files into an Excel sheet.
• Verified data accuracy and cleansed data.
• Processed data as needed (datetime segmentation, odds standardization, calculated fields).
• Calculated key metrics like ROI and P/L using pivot tables.
• Build a dashboard to present the necessary information.

---

Data Processing Steps

During the data processing phase of this project, several key steps were undertaken to refine and structure the dataset for in-depth analysis, including:

1. Seperated Datetime column into 'Date' and 'Time', and calculated 'Month', 'Time' and 'Day of the Week' columns.

2. Refined column labels for better interpretability and clarity.

3. Extracted and standardized odds data for consistency and calculated 'Odds Bucket' column.

4. Cleanup Selection Column to remove irrelevant information.

5. Categorised in 'Market Category' the events.

6. Calculated Profit and Loss (P/L).

7. Created 'Won' column out of 'Result' data column.

8. Introduced Live Bet column to mark live bets.

9. Calculated ommited data, where possible, to enhance dataset reliability and completeness.

Here is a snapshot of the processed data table:

---

Metrics Calculation and Dashboard

Next, I used pivot tables to calculate key metrics, including:

1. Return on Investment (ROI)

2. Hit Rate

3. Average Odds

4. Average Stake

5. Picks and P/L

6. Total growth over time

Which resulted to the dashboard seen below:

What is more, the data in the dashboard can be easily filtered by:

1. Account

2. Market

3. Odds Bucket

4. Month

5. Day of the Week

So that we could easily see, for example, how did Account 5 do on Handicaps?

Or, handicaps in general?

How did wagering in the 2.00 - 2.99 Odds Bucket perfromed in February?

Or, how was the strategy performing in Saturdays?

---

Project Impact

As mentioned in the beggining, this was the first step for evaluating a given waggering strategy across multiple accounts and overall. I was responsible for this project from start to finish, i.e from gathering and processing the data according to requirments to presenting the produced dashboard. I reported on the the process and all analysis done to contribute to the decision making process of the organisation.

---

Thanks for reading. Some information have been altered to ensure privacy. Feel free to reach out for any comment.

Close Window

Ireland's Rental Market

In this project, I focused on analyzing the current rental market in Ireland by scrapping real-time data from two major property websites, daft.ie and myhome.ie, in late September 2024. Using Tableau, I visualized the data to create an interactive dashboard, allowing for an in-depth look at trends in rental prices and property availability across different counties. This project helped me better understand the housing market dynamics and sharpened my skills in data scraping and visualization.

---

Here are some key insights from my analysis of Ireland's rental market.

• 96% of total listings fall in the €500 to €4700 range, with 66% of these listings found in and around Dublin.
• The rent-to-income ratio in the same areas is 35% for a couple, with a mean income of €6380 and mean rent of €2228.
• The counties around the four major cities - Dublin, Cork, Limerick, and Galway - have a mean rent of €2159, which is almost 28% higher than the rest of the counties where the mean rent is €1690.

---

For this project, I developed two Python-based scrapers using Selenium to extract rental data from daft.ie and myhome.ie. Both scrapers focus on gathering key details such as price, location, and property characteristics, but they differ in how they handle HTML variations between the websites. The daft.ie scraper collects raw data and applies cleaning and formatting functions (e.g., clean_price(), extract_info()) after the extraction.

In contrast, the myhome.ie scraper formats and cleans data during the extraction process, streamlining the data pipeline by integrating these steps directly into the scraping function.

---

Only around 100 listings have a price higher than €4700, Including some extreme ones. The rest 96% of total listings cost less than €4700, with 66% of these listings -or 1619 properties- found in and around county Dublin.

---

The rental landscape in Ireland reveals a significant divide between urban and rural areas, particularly around the major cities of Dublin, Cork, Limerick, and Galway. In these urban regions, the mean rent is €2159, making it nearly 28% higher than in the more rural counties, where the average rent is €1690. This disparity highlights the increased demand for housing in and around these major urban hubs, which drives prices higher.

---

Among the 2555 properties analyzed, 1992 have a recorded furnished status. Of these, the overwhelming majority —1841 furnished properties— command a mean rent of €2210. In contrast, the 151 unfurnished properties have a significantly lower mean rent, averaging €1355. This data reveals a clear rental premium associated with furnished properties, suggesting that renters are willing to pay more for the convenience of a fully equipped living space

---

Our analysis covers properties across ten different property types, yet only three of these—House, Apartment, and Flat—dominate the listings. These categories account for a total of 2407 out of 2555 listings, highlighting their prevalence in the rental market. This concentration suggests that these three property types are the primary options available for renters, with other types playing a much smaller role in the overall market.

---

Lastly, I added some really useful customization features to my Tableau dashboard, making it more interactive and easier to use. For instance, with filters for counties and property types, users can dive into the data based on their interests. These filters let people explore rental trends in specific regions or for particular types of properties, and the changes they make are applied across all charts and maps instantly. This feature really enhances the overall experience, turning the dashboard into a powerful tool for those who want to get detailed insights quickly and easily.

---

Thanks for reading. Some information have been altered to ensure privacy. Feel free to reach out for any comment.

---

Close Window

User Growth Trajectory Calculation

In this project, I've developed a straightforward model to calculate the monthly growth trajectory of our product's user base. The model consists of three main sections:

1. Parameters Section

This area allows for adjusting key variables that affect growth projections. The parameters include:

• Initial User Base: I assum an initial user base of 10 users, but this number can be adjusted.
• Conversion Behavior: Responses to the product are categorized into three groups:
  • Yes (Immediate Adoption): Users who convert within the specified timeframe.
  • No (Later Conversion): Users who convert after a delay.
  • Never: Users who never convert.
  These responses are modeled based on word-of-mouth marketing, assuming users influence others through product recommendations.
• Conversion Time Frame: Users are grouped into those who convert within the predefined conversion week and those who convert later, which is also adjustable.
• Influence Factor: This represents the number of additional users each initial user is expected to influence over a year. In the example, we assume each user influences 20 others.
• Market Cap: This defines the total potential market size. It helps track how close we are to saturating the available market.

2. Calculation Section

The model calculates how user conversions will unfold over time, considering the initial user base, the conversion behavior, and influence factor. For example, users who say "Yes" convert within the conversion week, while "No" users convert at a later stage. We also estimate how many people each user will influence and factor that into the growth projections.

3. Visualization Section

To help visualize the growth, we've created graphs that show user growth over a 12-month cycle, starting with the initial 10 users. These visualizations provide insights into when we'll reach the market cap and how quickly the user base expands, based on the given parameters.

This model offers flexibility in adjusting key factors that influence growth and provides a clear visual representation of how the user base can expand over time.

Close Window

Moving to Athens

New Image Description

This project's presentation is still on development. Further information will be published soon.

Close Window

Google Data Analytics Proj 1

New Image Description

This project's presentation is still on development. Further information will be published soon.

Close Window

Google Data analytics project 2

New Image Description

This project's presentation is still on development. Further information will be published soon.

Close Window