Is ChatGPT a Free Alternative to the Bloomberg Terminal for Undergraduate Experiential Learning?

Abstract

This is a comparison of active learning-based assignments for basic stock analysis in an introductory finance class using the Bloomberg Terminal and ChatGPT from both faculty and student perspectives. I undertook this analysis at the end of Spring of 2024 after learning that my College would be terminating its Bloomberg contract at the end of semester. The results indicate that the DDM output can be replicated using Yahoo finances API and Python with coding assistance provided by ChatGPT, and doing so will likely improve students understanding of the learning objectives. However, ChatGPT/Python assignment will be significantly more difficult than the Bloomberg version, thus placing the weight of the cost savings on the shoulders of the weakest students.

Introduction & Literature Review

This analysis was carried out in the Spring semester of 2024 in the business department at Albright College, a small Liberal Arts College in Pennsylvania. The primary sample drew from students in the introduction to finance class required for all business majors which is known as “Financial Management. It is a 300-level class taken after students have completed prerequisites in economics, accounting, and statistics at the 100 and 200 levels. The course uses the 16th edition of “Fundamentals of Financial Management” by Brigham and Houston as the primary text. All 26 students in that course were offered the opportunity to participate in this study as extra credit, 12 (46%) chose to participate. An additional convenience sample of 10 students enrolled in upper-level finance courses also completed this comparison study for extra credit. In total 22 students completed the comparison study. This analysis reports both student and faculty feedback directly comparing an assignment based on evaluating Bloomberg’s DDM function with the steps necessary to mimic that output using Yahoo finances API and Python with coding assistance provided by ChatGPT.

Active and Experiential learning

Active learning-based assignments are common in both AACSB and ACBSP accredited programs. Experiential learning is defined as a structured educational process in which learners gain knowledge and skills by engaging in real-world or simulated activities that mirror professional scenarios, often followed by reflection, analysis, and integration of theory and practice (FINSIMCO, 2025). Examples include managing investment funds, participating in trading competitions, conducting client-based projects, and performing financial analyses for real companies (Elliot Davis, 2016). Active learning, on the other hand, emphasizes the importance of student engagement during the learning process, often through collaborative problem solving, discussions, applied exercises, group work, and use of technology-driven tools. Both approaches shift the focus from passive absorption of content to active knowledge construction, critical thinking, and hands-on skill development as promoted by ACBSP and AACSB. The original Bloomberg active learning assignment that this ChatGPT version is meant to replace was designed as a stepping stone to help students progress to experiential learning exercises in portfolio construction and evaluation. Many AACSB-accredited institutions have built finance labs equipped with industry-standard technologies such as Bloomberg Terminals, enabling students to participate in trading simulations, analyze real-time market data, and engage in investment competitions using professional tools (Bloomberg for Education, 2025). These labs offer scenarios where students manage investment portfolios, make real trades using notional capital, and compete in global trading challenges, thereby bridging the gap between theoretical knowledge and practical application (Bloomberg for Education, 2025).

Bloomberg’s integration into the finance curriculum

At the time that I was brought into the College in 2016, it was estimated that there were more than 6000 Bloomberg terminals at colleges and universities around the world (Athavale, Edwards & Kemper, 2016). The College had recently acquired a terminal, and I was hired with the intention of integrating the terminal into the undergraduate business curriculum. Over the next few years, I did so by developing an integrated series of active learning-based Bloomberg application assignments, across the entire finance curriculum as well as the capstone business required for all management majors.

The original Bloomberg DDM assignment that I focus on here was developed as part of this integrated series of projects in the introductory finance course. It was created based on several pedagogical papers including Lei & Li (2012), Johnson & Shagnea (2012), Scott (2010); Tan & Tuluca (2017). Over the years I experienced the same stumbling blocks as many of my colleagues at other colleges and universities regarding student “buy-in” and difficulty with the terminal interface. Payette & Libertella (2012) found that less than 1/3 of the students thought this software was easy to use. My students experienced similar difficulties. I do not have data on student feedback for this assignment alone, outside of this small comparison study, but over the past 16 semesters that this assignment was included as part of the series of project was run in the course, 40% percent of students on average complained about Terminal-based assignments in their native responses. However, student views of the Terminal seem to improve over time. In the most recent 3 semesters, there has not been a single negative narrative comment about the terminal from students in the Senior Finance seminar.

Despite some negative feedback from students, the overwhelming majority of finance faculty and industry professionals see the value of integrating Bloomberg Terminal applications into the finance curriculum. The Bloomberg education portal now includes 100 syllabi samples,19 terminal activities, 15 separate student assignments, and 13 professional guides, 8 terminal based case studies and three certifications offered via Bloomberg for Education.

ChatGPT’s integration into the curriculum

While the cost of a Bloomberg Terminal may be becoming an increasingly hard sell for today’s small liberal arts colleges, one can certainly not deny that these same colleges are grappling with the effects of free LLMs such as ChatGPT! As Dong et. al. (2023) notes, between Spring 2022 and Fall 2023,195 papers related to ChatGPT and other LLM’s were uploaded onto SSRN within the accounting finance and economic networks. Despite this there are relatively few papers on the use of ChatGPT for financial pedagogy and those papers that do exist tend to focus on its ability to write code for data analytics software such as Python and R. For example, Lui et. al, (2024) present two projects showing how ChatGPT can write code to access publicly available financial data through Yahoo finances API. This information is then used in sample assignments for portfolio optimization and the factor loadings for the Fama and the French three factor model. While the instructions for both assignments are very clear, they did not provide any evidence of students’ impressions or feedback. Lan (2023) also proposes Python based ChatGPT assignments, describing how to use API to download data and provides links via GitHub to programs which will help students analyze mortgages, determine which asset pricing model meets their requirements, evaluate their risk tolerance, and optimized portfolios. While the literature here is not as robust and does not include a good deal of student feedback, its simplicity and cost are very promising! This paper fills the gap by providing both student and faculty feedback directly comparing an assignment based on evaluating Bloomberg’s DDM function with the steps necessary to mimic that output using Yahoo finances API and Python with coding assistance provided by ChatGPT.

The remainder of this paper is organized as follows. First, the learning objectives of the assignment are presented, followed by the two versions of the assignment designed to meet those objectives. Next, the results of the student feedback questionnaires are evaluated. These questionnaires assessed students’ familiarity with both the Bloomberg Terminal, ChatGPT, and Python both pre and post assignment. They also asked students to rate how well they thought each assignment helped them to meet the learning objectives. Finally, I will share my thoughts from the faculty perspective.

DDM Assignment Learning Objectives:

After completing this active learning-based assignment, students will not only grasp how to use the Dividend Discount Model but will also develop critical thinking skills about its practical application and limitations in real-world financial analysis.

Project Learning objectives with formula references
Estimate Future Dividend Projections: Based on calculated growth rates, students should project future dividends, recognizing how changes in growth assumptions can affect stock valuation.
Determine the Required Rate of Return Using CAPM: Students must calculate the required rate of return using the Capital Asset Pricing Model (CAPM).
Understand and Apply the DDM: Students should be able to apply single and multiple stage Dividend Discount Model stock calculate the theoretical value of a stock as the present value of its future dividends.

Bloomberg DDM Assignment

Note: For brevity, only the Commands, calculations and analysis instructions are shown here.

Enter your Ticker <Go>, Use the DES (Description) page to find the current stock price, beta, most recent dividend, dividend growth rate and earnings per share.
1. Click on DVD to explore your company’s dividend history.
  1. How often does your company pay dividends?
    1. Calculate DIV0 (your firm’s most recent ANNUAL dividend)
2. What is your company’s 1 year dividend growth rate? (Look at “1 YR Dividend Growth”)
  1. Assuming your growth rate remains constant, use your current ANNUAL dividend and 1 year dividend growth rate to estimate the dividend next year (DIV1)
3. Has your company’s dividend growth rate remained stable over the past few years? (Compare dividend growth rates over the past 3 years)
  1. Based on this analysis, do you think it is reasonable assume that your DIV1 calculation above is an accurate estimate of next year’s dividend? Why or why not?
Type DDM to bring up Bloomberg’s theoretical price for your company using a 3-stage dividend discount model. The forecasted future growth rates, dividends and price in the horizon/terminal year have already been calculated for you in this model.
1. What required rate of return is Bloomberg using in its calculation?
  1. Use CAPM to calculate this based on the Bond rate (Rf), Country/Region Premium (Rm-Rf), and beta fount on the left-hand column on the DDM output)
2. Identify the components of Bloomberg’s multistage dividend discount model.
  1. How does your DIV1 estimate in part a above compare to Bloomberg’s “Dividends per share FY 1”?
  2. What is Bloomberg using as the “horizon year”?
  3. What is Bloomberg’s terminal growth rate?
3. Is your company currently over or undervalued by the market according to Bloomberg?

ChatGPT based DDM assignment.

The assignment motivation, learning objectives and formula references are the same.

Note: For brevity, only the ChatGPT prompts, abbreviated output/ directions, calculations and analysis instructions are shown here.

Download Python and install the Yahoo Finance API integration to allow you to download data directly from Yahoo finance into Python for analysis
1. ChatGPT Prompt: “Write Python code that will directly download monthly stock data for a list of stocks from Yahoo finance using the y finance library in Python. Assume that this is my very first time ever using Python. Please include instructions on how to download Python and install pandas.”
  1. Output from ChatGPT: To download monthly stock data using the yfinance library in Python, follow these step-by-step instructions, assuming this is your first-time using Python:
    1. Step 1: Install Python
    2. Step 2: Set Up Your Environment
    3. Step 3: Write the Python Script

You should see:

Once you see this, you will know that you have everything correctly set up.

Use ChatGPT and your knowledge of the dividend growth formulas to estimate DIV1 for your company.
1. ChatGPT Prompt “Write python code to find the annual dividends for (Insert your ticker here) for the past 5 years. Use that data to calculate the annual dividend growth rate for (Insert ticker here). Calculate next year’s expected dividend for (Insert ticker here) based on the most recent annual dividend and the growth rate calculated above.”
  1. What is your company’s 1 year dividend growth rate?
    1. Assuming your growth rate remains constant, use your current ANNUAL dividend and 1 year dividend growth rate to estimate the dividend next year (DIV1)
Use ChatGPT and your knowledge of the Capital Asset Pricing Model (CAPM) to calculate the required rate of return on your stock.
1. Ask ChatGPT to find the current risk-free rate (10-year Treasury rate) (No coding needed)
2. Ask ChatGPT to find the market risk premium (no coding needed)
3. ChatCPT Prompt: “Write Python Code to find the beta of (Insert Ticker here)”
4. ChatGPT prompt: “Write python code to calculate the required rate of return on (Insert your ticker here) when beta=(Insert reply from part 3c above), Rf=(Insert reply from 3a above) and Rmp=(insert reply from 3b above)”
Use ChatGPT to calculate the theoretical price of your stock according to the dividend discount model.
1. ChatGPT Prompt: “Write python code to calculate the theoretical price of (Insert Ticker Here) when Div1=(insert from part 2a above), g=(insert answer from part 2a above) r=(insert answer from part 3d above)”
  1. Is your stock currently over or undervalued according to the single stage dividend discount model?
Use ChatGPT to calculate the theoretical price of your stock using the 3-stage dividend discount model.
1. ChatGPT Prompt: “Write python code to calculate the theoretical price of a stock using the 3-stage dividend discount model.” This will produce generic code requiring you to input:
  1. D0- The most recent annual dividend
  2. g1 – The current growth rate
  3. g2- The transitional growth rate (which can be higher or lower than g1 based on the company’s growth prospects)
  4. g3 is the growth rate at maturity (which can be higher or lower than g1 and g2 based on the company’s growth prospects)
  5. T1- The length of time the company is growing at g1
  6. T2-The length of time the company is growing at g2
  7. R= required rate of return
    Thus far the only inputs you know from the analysis above are D0, g1 and r. You will need to predict the rest based on analyst reports.
2. Use ChatCPT to help analyze your company’s earnings announcements, 10Ks, and/or analyst predictions for your firm and industry. Based on this analysis provide estimates of g1, g2, g3, T1& T2.
  1. Estimate these values and explain your estimates. I am not expecting estimation quality on par with Bloomberg, but your inputs should at least be consistent (For example if your industry has been contracting and your firm isn’t touting a new product that will turn things around over the next few years than g1<g2<g3 and T1 and T2 may be only a few years apart, but if they’re reporting a new product currently under R&D which is expected to go into production in four years from now and lead to a large new full of clients over the coming years, then g3>g2>g1 and T2=4, T2=3? Etc.)

Analysis of Comparison results:

Study participants were recruited via Canvas message and announcements to complete this comparison analysis for extra credit. The primary sample pool was the 26 students enrolled in the introductory finance course during the Spring Semester of 2024. Unfortunately, only 12 students in the introductory finance course participated so the study was opened up to the two other finance upper level courses running that semester which resulted in 10 more participants. In total 22 students completed this comparison study. Since the submissions were not anonymous, this allowed me to also include the students’ scores in the introductory finance class for comparison purposes. Grades were coded 1 to 4 with corresponding to a D or F and 4 representing an A. Grades were well dispersed throughout the sample with 18% having earned either a D or F, 18% Cs, 36% Bs, and 27% As.

Student feedback on the two projects was evaluated using a questionnaire that assessed their familiarity with both the Bloomberg Terminal, ChatGPT, and Python both pre and post assignment and asked to rate how well they thought each assignment helped them to meet the learning objectives.

Student Feedback:

Sixty-eight percent (15) of them self-reported no experience with ChatGPT for Python coding prior to the start of the project. Likewise, 55% (12) of the subjects had no exposure to Bloomberg prior to the start of this semester, although at this point in the semester they had used it for several projects. Students were asked to rate their comfort with both technologies. Since the Bloomberg Terminal was used all semester, this comparison is based on recalling their initial comfort with the terminal at the start of the semester they took the introductory course versus rating their comfort at this point in the semester. For ChatGPT assisted Python coding, students were asked to rate their comfort with the process at the start and end of the assignment. The comfort rating scale was 1 through 5 where 1 is “very uncomfortable” and 5 is “very comfortable”. All students self-reported increased comfort levels with both technologies; though, the increase was most dramatic for the Bloomberg Terminal with 36% rating themselves “very comfortable” at the end of the semester, compared with only 9% ChatGPT/Python. However, this result should be taken with a grain of salt given that we are comparing use of the technology for the entire semester versus a single project.

Next the students were asked to evaluate the degree to which they believe each of the assignments met the learning objectives. As explained above, the first learning objective was to understand and apply the multistage dividend discount model to calculate the theoretical price of the stock as the present value of its future expected cash flows. The other learning objectives were to estimate future dividend projections based on projected growth rates, and to determine the required rate of return for the company using the capital asset pricing model. The rating scale was 1 through 5 with 1 being “I did not understand the learning objective any better after completing the assignment” and 5 “the assignment improved my understanding of the learning objective greatly.”

With respect to the projected dividends, 73% of respondents rated ChatGPT/Python as a 4 or 5, while 14% indicated that it did not improve their understanding (rating=1). The Bloomberg version faired a bit better, with 77% rating it was a 4 or 5 and only 9% rating it as a 1. The only substantive difference between the two versions for this learning objective was that students need to calculate the one-year growth rate using ChatGPT/Phyton, whereas it is reported directly on screen with Bloomberg.

For the CAPM calculation on ChatGPT/Python version, students were required to look up the values of the proxies for the market return and risk-free rates, calculate the beta and finally calculate the stocks estimated return using the model, whereas in the Bloomberg Terminal version, they simply needed to identify the variables on screen. These extra steps seem to have been beneficial for student understanding because 64% of students rated the ChatGPT/Python as a 4 or 5 vs. only 36% for the Bloomberg version.

The calculation of the stock price using the multistage dividend discount model is where the differences between the two projects are most visible. The Bloomberg Terminal calculates the intrinsic value of the stock automatically using the DDM command, so students are simply asked to identify the components of the model and compare the result with the current market price. Without access to the Terminal, the students need to take the role of the analysts and estimate all components. Theoretically, these extra steps should help students gain a deeper understanding of the model, this may explain why none of the respondents rated the ChatGPT/Python version as a 1 whereas 9% did so for the Bloomberg version. But the Bloomberg version received 55% 4 and 5 ratings vs. 41% for the ChatGPT/Phyton version.

Finally, students were asked, “In your opinion (based solely on this comparison assignment), is AI assisted financial data analysis using python and the Yahoo finance API an adequate substitute for the Bloomberg Terminal in BUS 345” Fifteen students shared their opinions. The most common negative sentiment was simple anger at the loss of the Terminal, for example:

“I know I can use the Terminl <sic> at work, but IDK if I will use Python so no if the point is getting me ready for a job”

Some students expressed frustration with the “learning curve” for Python even with ChatGPT’s assistance, although as others point out, this can be rectified with additional instruction and video examples.

“Getting started and asking questions on ChatGPT was faster than navigating through Bloomberg at first but then you get used to Bloomberg. Maybe I will get used to programming but I’m not yet and nothing in my other classes really prepared me to do it.”

“I think the python and yahoo would be adequate, I struggled with it but I feel like if we were to start off with this method, and had videos like we do with Bloomberg, it wouldn’t be too hard to grasp. Just because it is my first time using it do I think this.”

Faculty Assessment:

A passing score for this assignment was 75% or above on the assignment rubric. Since the overall purpose of the project is to evaluate a stock price using the dividend discount model, rubric placed higher weight on the final stock price calculation learning outcome. Eighty-Two percent (18) of the respondents “passed” the Bloomberg version of the project, compared to 50% (11) for the ChatGPT/Python version. None of the students who did not pass Bloomberg version were able pass the ChatGPT/Python version. The multistage dividend discount model learning outcome had the most dramatic difference in scores across the two versions of the assignment.

	GRADE>=75%	GRADE<75%		GRADE<75%
	BB (N=18)	ChatGPT/Python (N=11)	BB (N=4)	ChatGPT/Python (N=11)
DIV1	100%	100%	50%	36%
Ri	94%	82%	25%	18%
PO	100%	73%	25%	0%

Pearson correlations reveal that the student’s performance in the introductory finance class was positively correlated with their performance on both versions of the project, but only statistically significantly so for the ChatGPT/Python assignment even after accounting for prior use of the technology. This stark contrast in the ability of the students to both pass the project and score proficient on the primary learning objective between the two different versions is troubling. The syllabus for the introductory finance class as approved by the department and College’s curriculum development committee includes active learning assignments which constitutes 30% of the overall course grade. This component is currently being fulfilled by the Bloomberg assignments which are completed in groups because we have only one terminal on campus. To examine the impact of the switch to ChatGPT assisted Python assignments on student grades, I ran a logistic regression. Based on this limited sample, there is a 12.8% chance that a “C” student (Grade in 345=2) “passing” the ChatGPT/Python version, which is certainly concerning!

Dependent variable=ChatGPT_Pass

	Estimate	Std. Error	z value	Pr(>\|z\|)
(Intercept)	-6.5462	2.9749	-2.2005	0.0278 *
Grade_in_345	2.3169	0.9969	2.3241	0.0201 *
Signif. codes: 0 ‘ * ‘ 0.001 ‘ ‘ 0.01 ‘ * ‘ 0.05 ‘.’ 0.1 ‘ ‘ 1

Discussion, Limitations and Conclusions

The Bloomberg Terminal’s ease of use and presentation of data are highlighted in this analysis because instead of simply typing the Ticker, DVD and DDM into the Terminal, the steps necessary to mimic that output using Yahoo finances API and Python with coding assistance provided by ChatGPT are far more involved. These extra steps focused on writing code to first look up the data and then perform a calculation using one of the formulas contained in the learning objectives, which seems to have helped improve students’ self-assessed understanding of those objectives. On the other hand, this added complexity appears to have been “too much” for the weaker students. While 82% of the sample were able to score 75% of above on the Bloomberg Terminal version of the assignment, only 50% were able to do so on the ChatGPT/Python version. Furthermore, the results of logistic regression based on this sample predicted only a 12.8% chance that a “C” student would earn 75% or above on ChatGPT/Python version of the assignment. Since the approved syllabus for the course in question weighs the active learning assignment so heavily (30% of the overall course grade), moving to ChatGPT/Python based assignments will almost certainly negatively impact student’s course grades. Hopefully the student who said, “if we were to start off with this method, and had videos like we do with Bloomberg, it wouldn’t be too hard to grasp” is correct, and the impact of this transition on student grades truly can be mitigated by fully incorporating Python with coding assistance provided by ChatGPT and Yahoo Finances API into the course as I had done previously with the Bloomberg Terminal will be sufficient.

The most obvious limitation of this study is the sample size. Unfortunately, the termination of the Bloomberg contract prevented me from repeating this comparison study in subsequent semesters.

So, is ChatGPT a viable free alternative to the Bloomberg Terminal for undergraduate experiential learning? According to this analysis, yes but the cost savings will come at the expense of the lowest performing students. It is my hope that this analysis will help other faculty at small colleges facing similar financial constraints with their decisions to either reaffirm their commitment to the use of the Terminal or adjust their active learning-based assignments accordingly.

References

Athavale, M., Edwards, J., & Kemper, K. J. (2016). “Bloomberg 201: From Wall Street to University Avenue.” Advances in Financial Education, 34-50
Bloomberg for Education. (2025). Finance Labs Offer Real-World Learning | AACSB. https://www.aacsb.edu/insights/articles/2025/04/finance-labs-offer-real-world-learning
Davis, E. (2016). Experiential Learning in Business School Finance Programs – AACSB. https://www.aacsb.edu/insights/articles/2016/07/experiential-learning-business-school-finance-programs
Dong, M.; Stratopoulos,T.& Wang, V. (2023), “A Scoping Review of ChatGPT Research in Accounting and Finance”
Johnson, R. and Shagnea, M. (2012) “Bloomberg Exercises for Investment and Financial Market Classes,” Financial Education Association Annual Meeting, Charleston, S.C., September 2012
Lei, Adam YC, and Huihua Li. (2012) “Using Bloomberg Terminals in a security analysis and portfolio management course.” Journal of Economics and Finance Education 11.1, 72-92.
Scott III, R. H. (2010). “Bloomberg 101”. Journal of Financial Education, 80-88.
Sharma, A. (2015). “Use of Bloomberg Professional in support of finance and economics teaching”. Cogent Economics & Finance, 3(1)
Tan, X., & Tuluca, S. A. (2017). “Supply Chain, Financial Management and Bloomberg Terminals”. Financial Management and Bloomberg Terminals (September 30, 2017). Journal of Insurance and Financial Management, 3(4), 66-78.