How we Value Companies

On Sunday, we ran a survey to get feedback from you on how to improve Rebound Capital. The most surprising result? Vert few of you thought our valuation models were valuable to your investing process.

We reached out to a few of our readers, and the common feedback we heard was that the models felt too complicated.

So today, we will dive deep into how we create institutional-grade valuation models. We will go through each step of the process, why we do those steps, and their significance.

So get a cup of coffee — you just have to read through this once deeply, and you’ll come away with a solid understanding of how companies are valued and how analysts assign buy/hold/sell ratings on companies.

Welcome to Rebound Capital. If you are new here, we conduct deep research into beaten-down stocks and study companies that made a successful comeback. Subscribe for free and join 7,100 other investors to make sure you don’t miss our next briefing:

So why do we need a valuation model in the first place?

For example, during our Alphabet ($GOOG) analysis, we recognized that Search revenues were the most critical metric. We saw very clearly that the Search revenues were consistently growing. There was a lot of market chatter about the search revenues getting affected, but that was not reflected in the numbers. That gave us the confidence to go long Alphabet.

One can make the argument that future search revenues were expected to be affected (due to ChatGPT), and hence Alphabet stock sold off. But, here we will draw your attention to the most essential tenet for rebound investing: “follow the data”.

Many times, stocks will sell off based only on narratives. Keeping track of key KPIs that move the stock will help us develop conviction in our stock picks. This has worked time and again for us — Look at ASML, AMD, & GOOG.

While doing a deep dive into Alphabet, we didn’t come across any data point that would suggest a slowdown in search revenues. In fact, the management was consistently giving good counterarguments. This is how we generate stock ideas: a disconnect between the narrative and the underlying business metrics.

Once we recognized the key reason for the drawdown, we then did a quick DCF (discounted cash flow) analysis for Alphabet. The stock was trading materially below the intrinsic value. When weak stock sentiment meets an undervalued stock based on intrinsic value, that is when the odds are in favor of Rebound Investing.

A DCF model estimates what a company is worth based on the money it’s expected to make in the future. It projects how much cash the company can generate after expenses and then “discounts” it back to today’s value using a rate that reflects risk and the time value of money.

In short, it helps answer: “If I owned this business, how much would its future profits be worth today?”

Now, let’s review the steps required to create a DCF model and how we identify key KPIs that move a stock.

We will be using AMD for our example. Since our buy recommendation in July, the stock returned an incredible 61% in 3 months (part of the run-up was due to the OpenAI x AMD deal). Based on our updated valuation model, we issued a sell rating in Oct’25, locking in our gains. (This update & model were exclusive to our paid subscribers. Please consider upgrading to get all our latest reports and picks.)

Here’s why:

The first step is building the company’s revenue model. This is essentially our attempt to build how the company earns revenue bottom up.

In AMD’s case, we are summing up revenue from the different divisions of AMD’s business – Data Center, Client, Gaming, and Embedded Segment. In the Data Center, we sum up revenue from the CPU and GPU segments.

A good starting point to build up the models is to first check the company’s own guidance for different segments. Company guidance are the forward looking financial estimates (for revenue, earnings or margins) provided by management during quarterly earnings calls or investor day presentations (once a year presentation by companies to investors).

This guidance is very critical as analysts and investors track management’s guidance and compare the actual results against the guidance given to gauge management’s execution.

In AMD’s case they have been consistently breaking out revenue from different segments and giving guidance on the GPU and other segments. For example, in the Q3’24 earnings call, AMD’s management had raised guidance for GPU revenue for full year 2024 to $5B from $4.5B (in the previous quarter).

Given the Data Center is their most important division, we broke down the revenue from this segment into revenue from CPU and GPU segments. For the GPU segment, we estimated the number of units sold in 2024 basis the ASP (Average Selling Price) of the GPU chips sold (we get an idea about this based on searching the GPU prices online).

Since the CPU division is a legacy technology and not growing as rapidly – we decided not to break down its revenue further and instead took a simple CAGR to project CPU revenues out into the future.

For the GPU segment (all important) – we projected the number of units sold based on AMD’s deal with OpenAI and taking an assumption on how many GPUs AMD will sell to other customers. The ASP for the GPUs in the future years is estimated based on discussions on technology blogs about the price of the GPUs in the coming years.

Adding up the revenue from different segments, we get the total revenue for AMD.

This is a less involved process. Here, we have to take estimates for the Gross Margin(%) and the Operating Margin(%), or take estimates of the operating expenses in the coming years. These estimates can be made based either on the management’s commentary or our own analysis of the future trends in the industry.

In AMD’s case, the management has talked about both the effects of a growing data center revenue on the Gross Margin and Operating Margin:

We have taken our estimates largely based on AMD’s guidance and the comments from other management teams in the industry (Broadcom’s management talks about lower gross margins than corporate average but higher operating margins at scale).

To get the Free Cash Flow we look at historical data to understand how much of the Operating Income has historically converted to Free Cash Flow as a %. Then we make our own assumptions for how this % will change (if at all) in the future. Usually, we can expect the ratio to remain the same in the coming years. This conversion metric is important but usually the market doesn’t penalize a stock for lower conversion.

The next step is to use the Free Cash Flow (FCF) estimates we calculated in the previous step and discount them to get a discounted present value of the future cash flows.

This is more art than science and depends on the risk appetite of investors. At Rebound Capital, we

For AMD, after the Open AI deal, the company was trading at ~$240. This was more than 20% of our calculated intrinsic value of $198 for AMD and we issued the sell rating.

From our AMD update:

Given that we already had made 61% gain (that too over a 3 month period), it made sense to play it conservatively and allocate to a better opportunity (hello, Constellation!)

Please consider upgrading your subscription to get access to our exclusive Rebound portfolio.

As you might have realized by now, a DCF is just a tool with a lot of assumptions. Come tomorrow, if Google/Grok announces another blockbuster deal with AMD, we will have to reframe the assumptions and then probably assign a higher valuation to the company (This is exactly why analysts change their price target once new information comes out. This is also the reason why most analysts play catch up where they re-rate the company only after the stock price has gone up).

Excessively focusing on a DCF can lead us to miss out on the most disruptive technologies of the day. This is because new age business models invest in the business using their income statement and hence may have no Free Cash Flow to report.

The other extreme is when the market totally ignores the basic valuation for companies (since they are disruptive). Palantir is now trading at a phenomenal >200x Enterprise Value/Free Cash Flow. While it might work out for this company, if an investor consistently buys companies at such valuations and ignores the basic valuation principles, they are bound to get very poor investment results.

If you found this interesting, please share this report with a friend.

Rebound Capital’s work is provided for informational purposes only and should not be construed as legal, business, investment, or tax advice. You should always do your own research