News & Blog

Special Event Planning: A Data-Driven Approach


Planning for special events can be challenging and time-consuming for Table Games Operators. There are multiple factors to consider such as open hours, game-mix, minimum bets and staffing. The difference between an optimal and sub-optimal plan could cost the operations 10% or more in missed profit opportunities, either from missed incremental revenue or excessive labor costs.

Events such as public holidays, concerts, sporting events or tournaments are a generally busy time for casinos, attracting crowds that go beyond the regular weekly trend. With the increase in demand, creating the right environment by aligning the table supply with the player demand can help capture incremental profit.

Traditionally operators either use weekend spreads or simply open to capacity during peak hours during special events. While better than a static spread, this may not accurately align with the demand that actually attends the upcoming event.

 

Relative Demand Approach

We start with two sources of data; the first is demand from similar events from the past to model the upcoming event, and the second is macro changes in demand from the previous year. This allows operators to plan for any upcoming special events, using a combination of representative past special events as well as recent demand trends.

We first calculate the difference in player demand from the historical data i.e. what is the % uplift from previous representative special events, relative to the same day of the week from the weeks prior to the event. This gives an indication of how the special event influenced demand in the past.


Then, the average demand uplift percentage is applied to the recent demand trends.

The benefit of this approach is that (1) it automatically takes in macro economic factors and (2) any changes to the casino floor are accounted for since we are projecting the same relative changes in demand to the current demand trends.


With more granularity (e.g. hourly and by different sections of the casino), the uplift forecasts can be more representative of the anticipated trend in demand. This also comes with more noise in the data. From our experiments, segmentations of demand by section of the casino, at an hourly granularity, yielded the best results.

 

Detailed Example: Vegas Superbowl

The Superbowl is a massive event in Las Vegas. The increased demand is not only on the day of the big game, but covers the entire weekend. Most operators will experience a Saturday that far exceeds their average Saturday with all of the visitors converging on the city.

The Sunday itself provides an interesting blend of high and low demand with increased volumes on either side of the game, and decreased volumes during the game.

In a previous blog post, we advocated for the use of representative events in planning for times like The Superbowl. In this edition we take this one step further by also leveraging year over year changes in business volumes.

As we begin to assess the weekend demand we need to determine The Superbowl Effect. This is the proportional change in demand vs a regular weekend leading up to that event.

In the example below, we assess The Superbowl 2018 weekend (Saturday, February 3, 2018 and Sunday, February 4, 2018) to use as a representative event to forecast The Superbowl 2019 weekend. The hourly player volumes for The Superbowl 2018 weekend are represented in the graph below with the dotted line indicating an average weekend, and the bold line indicating our representative weekend.

 

For the two days combined we determine the effect is an increase of 29% in overall demand. Saturday is 27% higher than average and Sunday is 32% higher.

Taking a closer look at Sunday, February 4, 2018, when the Superbowl game took place, it is important to note that our overall uplift of +32% on Sunday is not consistent for the entire day. Our factor can range from +115% (more than double our typical demand) to -33% during the game itself, and anywhere between.

 

 

Now that we’ve determined our hourly uplifts we can apply this against a more recent average weekend. This allows us to take macro changes in business volumes into consideration when forecasting the demand.

The chart below shows that our recent business volumes, in 2019 (grey dotted line) are higher than historical volumes, in 2018 (red dotted line).

 

 

Next, by applying our hourly uplifts to our more recent demand trends we determine a forecasted demand (green bold line) that is higher overall than our similar event in the past (red bold line).

 

Finally, putting it all together, it is evident that forecasting demand at an hourly level is important in order to successfully capture opportunities throughout the weekend.

 

 

Method Comparison

So what is the overall impact of this approach vs a simpler approach of using last years’ demand?

 

Metric

Representative Period Only

Uplift Method

Forecasted Patron Hours

8.3K

9.8K

Optimized Table Hours

3.2K

3.5K

Planned Average Table Min

$63

$66

 

The uplift method results in 20% higher forecasted patron hours, indicating the need for 10% more table hours, and a 5% higher average table minimum.

In fact, if we apply our enhanced demand forecast against the spread that would have been planned based on the original forecast, it results in 9.8% less operational profit. By using the uplift method, and more accurately estimating our demand we are able to capture that nearly 10% in incremental profit.

The high-level metrics do not show the breakdown by casino area, game, and price point, but indicate the significant impact that alternate planning approaches can have.

The failure to take into account the macro changes in demand, and micro changes in casino game mix, and pricing strategy would have resulted in significant lost revenue due to under-supply, and under-priced offerings for The Big Game.

 


Authors:

IMG_0092Paul Stephens
As Senior Manager of Analytics and Optimization, Paul coaches partners to leverage the full potential of the TYM Software to enhance their operations. He has 10 years of experience in data analytics, client success, and table optimization.

 

 

Fernando Valdes

Fernando is the Senior Product Manager at Tangam Systems. He utilizes a data-driven approach to solve real-market problems and ensures that technical developments in the TYM product roadmap are aligned with the business targets of Tangam’s customers. Graduating from one of the top engineering schools, Fernando has a Bachelor of Applied Science in Honors Electrical Engineering from the University of Waterloo.