How To Consistently Outperform The Market?
No bullshit guide.
Over the last five years, we have outperformed the market every year with increasing margins.
We finished last year up 44% while S&P 500 delivered 17.9%. We are currently up by 57% over the last twelve months against just 26% of S&P 500.
I am not telling this to brag; I am telling this because I will explain how we have achieved this so far, as this is one of the questions I receive frequently.
As the market is at an all-time high, and attractive opportunities are getting rarer, I think this is a very good time to talk about the fundamentals, to understand where investment returns come from and how you generate them consistently.
Before getting into the meat of it, I have to say we practice a very specific way of investing. We have one goal—to outperform the benchmark over the long-term.
We don’t target a specific margin of outperformance. We just have a strategy that is optimized for long-term outperformance, and margins are just a byproduct. We may outperform by 5% or 35%, it’s a byproduct, nothing we shoot for. We apply the strategy, and the market decides the final performance.
Let’s call that margin of outperformance alpha.
If we were an investment firm and were about to teach how to generate that alpha to a junior portfolio manager, we would do it in four main steps:
Where to look for alpha?
How to spot a potential alpha?
How to enter positions and size them to capture alpha?
When and how to exit positions to lock in alpha before it erodes?
Thus, I’ll follow this same structure here as well and provide examples from my own investments when relevant.
Let’s cut the intro here and get to the meat of it!
Where To Look For Alpha?
You can do a quick research about the sources of alpha in the market, and I guarantee you’ll find hundreds of alternative interpretations. As someone who has spent more than a decade in the industry, I have already done this, and I guarantee that 99% of what you’ll find is noise.
After years spent reading and researching about this, I can say Michael Mauboussin provides the most comprehensive yet simple explanation I have seen.
He flags four sources of alpha—informational, technical, analytical, and behavioral:
However, this doesn’t automatically mean you can capture all these sources of alpha.
Think about information edge.
Thanks to the current technology, the sophistication of the capital markets, and investment research firms, all publicly available information is immediately priced into the stock, provided that the company gets even some institutional coverage.
Thus, there are only two ways to have an information edge:
You are an insider
You look at obscure stocks nobody follows
If you are an insider ot have an insider tip, you can’t trade on it, so rule that out. If you are operating in the land of obscure stocks, information discovery is a double-edged sword. Yes, you may discover something that others don’t know, but you may also never know things that would prevent you from investing if you knew.
So, in modern capital markets, information edge is largely irrelevant for investors.
What about the technical edge?
The technical edge is more real in the current markets compared to the information edge, but it’s excruciatingly hard to capture if you aren’t an institutional investor.
If you bank on a technical edge, you must have hundreds of pre-defined models to trade the inconsistencies before the opportunity window closes. You lose it if you stop to take a look at fundamentals. You must have models doing it on autopilot and it only gives you marginal gains as all institutions are doing it, narrowing the opportunity.
So, it exists, but it’s probably irrelevant for you if you aren’t an institutional investor.
Let’s turn to analytical edge, then, does it exist?
Analytical edge is basically you look at all the same information and technical set up every other market participant looks at, and you interpret it differently than the consensus.
Analytical edge is real because it’s your prerogative to determine what the material you are looking at means. It may be due to your personal background, knowledge base, biases, etc, but it’s always possible to look at the same material and have a different interpretation.
Aswath Damodaran famously valued Uber at $6 billion in 2014, assuming it would be an urban car-share company. Then Bill Gurley argued that its platform allowed Uber to expand to almost all verticals in mobility services. Then, Damodaran made valuations based on different views of the company:
As you see, looking at the same material, you could pick any one of these paths for the company. Damodaran, as a value-oriented financial mind, picked an urban car service scenario, and Bill Gurley, driven by his own background, picked the broader mobility services. At the end, Gurley came out right, and Uber is a $150 billion company now.
This shows you how important and decisive an analytical edge can be. It’s real, it doesn’t require much dwelling on details and it’s your prerogative.
When looking at this, most people think they needed to make the mobility services, strong network effects, double the market size assumptions to make money on Uber. No, it was way simpler than this. It was okay if you just thought it could easily expand beyond urban ride-hailing, while it was valued as just urban ride-hailing.
You didn’t need to know the peak optionality. You just needed to assess whether some optionality existed and how likely it was. The rest would be discovered by time.
Time brings us to the latest source of alpha—behavioral.
You don’t need to think a lot to understand that behavioral edge is real. It flows naturally from the analytical edge that we just proved to exist via the Uber example.
Let’s say you were Bill Gurley in 2014, who thought Uber would be a global mobility service, and you've just discussed your thesis with Aswath Damodaran, who saw it as an urban ride-hailing business. You needed time to come out right.
At the end of 2014, Aswath would look right, and even at the end of 2015, Aswath would still look right. But in 2023, Aswath would be ridiculed for his shortsightedness, and you would be praised for your correct analysis of the business opportunity. You needed to give time to the thesis to exploit your analytical edge.
Indeed, Uber took a decade of gradual expansions to reach the stage it is today:
If you weren’t patient, didn’t control your emotions, you wouldn’t exploit your analytical correctness. Analytical edge is meaningful only when it’s coupled by behavioral edge.
So, there are only two sources of alpha you can capitalize on—analytical and behavioral.
Analytical edge is the differentiator, and behavioral edge is what allows you to exploit that analytical advantage.
The natural next question is, where can you find the situations you may have that initial analytical edge? Where do you find that first potential alpha? You force it? You randomly turn over rocks?
No. It’s simpler.
How to spot a potential alpha?
Howard Marks is one of the greatest investors alive. He tells a striking anecdote:
He was talking about the state of the market with his son, who is also an investor. His son says, “Dad, you say don’t make market calls, but all your market calls have been correct so far.” Marks responds, “yes, because I have only made a few of them.”
Indeed, Marks called the market three times in the past 20 years, and he was correct in all of them:
What do you see in his calls above?
He only calls it when it’s at extremes. Low or high, when you call it at the extremes, your chances of success are way higher.
You may say something about the future direction of the market at 20x earnings, and it can go both ways. Going up to 22x earnings and dropping to 18x earnings are equally likely. But at 28x earnings, going down is much more likely, as the market has never traded above 28x earnings for more than several months in its whole history.
That also applies to seeking analytical alpha, an initial edge in individual equities.
When things go to extremes, an opportunity for analytical alpha arises. Remember Damadoran’s Uber valuation.
If Damodaran thought it was an urban car service with strong local network effects, which yielded around $7 billion. If Bill Gurley thought it was a logistics business with local network effects, that would yield ariund $14 billion valuation and analytical alpha wouldn’t be that big. He thought it was a mobility services business with global network effects, which yielded an over $90 billion valuation.
That tells you where to look—you have to look at things having swung to extremes.
As a long-only investor, I only care about extremes on the downside. There are two types of this extreme:
Extreme underappreciation of the current business
Extreme underappreciation of the potential.
These are the two types of extremes we capitalize on, and each one requires different disciplines. Let’s start with the first one, i.e., the extreme underappreciation of the current business.
Let’s start with a recent example: UnitedHealth.
It’s the dominant health insurer in the US, having grown revenues by around 10% annually in the long term. Its 10-year median P/E is 22:
On June 17, 2022, it was trading at $452 per share and 24x earnings.
What would it do next, provided that you knew nothing would change in the fundamentals of the business? You couldn’t have a strong opinion.
Indeed, it went to trade on 28x earnings a month later, but down to 20x earnings a year later. Both directions were equally possible, and the stock was driven largely by noise rather than definitive signals.
Fast forward to August 2025, and it was trading at 10x earnings.
Where would it go, provided that you somehow knew business fundamentals would remain intact in the long-term? This time the answer would be easy: Up, up, up.
As you see, in these types of extremes, the key phrase is “provided that you knew fundamentals would be intact.”
There is no way to know this definitively, but you maximize your chances when you pick a business that has some durable competitive advantage. If the business has this, fluctuations in performance tend to be temporary, and the stock is more likely to revert to its long-term median over time.
The core skill here is understanding competition dynamics—i.e. what makes a business durable?
This is a skill you need to develop, but principally, you can apply the “extremes” model here as well. Strongest competitive positions are obvious, like Costco (insane scale), Amazon (scale + network effects), Apple (brand+ecosystem), UnitedHealth (vertical integration at scale), American Express (brand+network effects), etc.
When businesses like these with durable competitive positions swing to extremes, you have a chance to assume that fundamentals will likely stay intact and the stock will revert to the median.
The second type is when the potential is extremely underappreciated.
Here, the most recent example from my portfolio is AMD.
It announced Q4 2024 results in early February 2024 and delivered $25.8 billion in revenue. The market didn’t like it, the stock started to sell off, and hit $85 levels in April 2025 as the broader market also sold off due to tariffs:
At that point, this was the picture I was looking at:
The stock was trading at 20x 2026 earnings and around 15x 2027 earnings.
Annual server GPU spending was expected to reach $700 billion by 2030.
AMD just had around 4% market share in server GPUs.
At that point, even if we assumed that AMD server GPU market share in 2030 would be just 5%, it would generate $35 billion in revenue just from server GPUs. In a more likely scenario, it could reach around 7-8% market share and generate $49-$56 billion just from server GPUs.
Assuming its $13.2 billion non-data center revenue in 2024 kept growing just 10% a year, it would reach almost $25 billion in 2030.
So, in the most conservative scenario, we were looking at a company that could triple its revenue in 5 years, trading at just 20x earnings. Even if we assumed stable margins and earnings multiple, that would be a 3x opportunity.
The market was underappreciating its potential without even considering the CPU demand boom thesis that we have today.
So, you didn’t need to be too visionary to foresee the CPU demand boom back then. It wasn’t a case, and it flew nowhere near my mind when I bought the stock in 2025. Its potential, based on the existing paradigms, was just extremely underappreciated.
So, these are the two cases of extreme dips we pursue:
Extreme underappreciation based on the assumption of the continuity of the historical performance.
Extreme underappreciation of the ability to capture future potential.
The critical analytical skill in the first case is competition, and it’s the market potential, growth, and assessment of capabilities in the second case.
Once you make these analytical decisions, the rest is behavioral. Whether you’ll be able to wait long enough, when you should exit, when you should double down, etc.
In this behavioral process, the first critical step is how to enter positions and decide on size.
How To Enter Positions And Decide On Size?
Once you make an analytical decision at the extremes I mentioned above, you have an analytical assumption.
The keyword here is assumption.
The most frequent mistake I see is that people behave as if their assumptions are the base case and everything except that, especially the bad things, are deviations. This is why they generally take big initial positions, driven by the fear of missing an opportunity, and when their assumptions don’t play out, they lose big.
No, we have to be humble. We just have an assumption, and it may not play out. So, the most important thing is limiting the downside. How can we do it? We do it by playing as safely as possible.
If you initially start with a big position, you undermine the safety. This is why I almost always start with a small position. Around 4-5% in cases where a business’s ability to sustain performance is underappreciated, and with 1-2% when its potential is underappreciated.
From there on, we initially have three and eventually two options:
Nothing is obvious. (This eventually evolves to one of the following)
It’s obvious that assumptions aren’t playing out.
Assumptions are obviously playing out.
When nothing is obvious, you keep holding until something is obvious, and make no decisions. When it’s obvious that assumptions were wrong, we exit fast. When they play out, we have two options:
The market bumps it up to fair value. → We do nothing, let that small position run
The market takes longer to recognize it → This is what we double down on
As you see, we are only interested in scaling the position when we get the confirmation of our assumptions from the business, but the market somehow fails to recognize value.
This is a very small subset of stocks, but they exist. And we operate here.
We don’t try to make money on everything; we are just interested in this subset. This allows us to make moves that are derisked by the signals given by the business itself.
An obvious example from my investments is SoFi.
I entered the stock in May 2023, at around $5 per share. At the time, the management had $0.50-$0.80 EPS guidance for 2026. Meaning it was valued at 6-10x 2026 earnings.
I thought this was too low, looking at the business fundamentals and its growth potential beyond 2026 and entered the stock.
From that time to September 2024, the business gave several signals of confirmation as it delivered only double beats of both analyst estimates and its own guidance.
The stock? It didn’t go anywhere:
This period allowed me to increase my bet while watching the solid execution and move towards the target step by step. The management has $0.60 EPS target for this year, and the market thinks it should be valued at 26x earnings despite increasing inflation and rate hike prospects.
The same thing happened with Amazon.
We had the cloud acceleration thesis since the beginning of 2024, and it actually played out as Amazon’s cloud revenues accelerated every quarter starting from Q1 2025, yet it dropped down below $200 twice anyway.
This was the perfect setup to keep betting on a scenario that actively got derisked:
In short, you should be aware of the fact that your presumed analytical alpha is just an assumption until it actually materializes, even though you maximize the chances by betting on the extreme cases.
If and when it materializes, the market recognizes it pretty fast in most cases, but in some small number of cases, it behaves pretty lazily. These are your chances to grow your bets with less risk, as you already receive some confirmation.
We are only interested in those cases.
The last step is determining when we capture that alpha for good, lock it in, and exit the position.
When And How To Exit Positions?
This is not about exiting an assumption that didn’t work. The right time to exit a position that didn’t work is right away, at the moment you understand it.
This is about an assumption that was validated, a thesis that actually worked.
We exit such positions in three cases:
When there is a better opportunity.
When the thesis starts cracking.
When it becomes egregiously overvalued.
The first one is pretty intuitive.
As I said, validated assumptions are divided into two groups:
Those that the market bumps up right away
Those that are rerated with a lag
And we are only interested in doubling down on the second group, which is a small subset. Thus, as we always start with small positions, over time, we are left with many small positions that are fully valued by the market and each making up 1-2% of our portfolio. So, these positions lack the ability to considerably move the needle for us.
We exit those positions to bet on new assumptions and grow the positions where the thesis has been validated, but the price lags.
A recent example from the portfolio is InPost.
We bought it for around EUR 10 and exited after it appreciated by 50% after it got an acquisition offer. It was 5% position, already appreciated by 50%, the acquisition was on the table, and it could fall back to old prices if the acquisition talks collapsed.
At that point, it made much more sense to exit it with profit and allocate it to other positions for better optionality rather than a potential rise in the acquisition price because it wouldn’t double, and anything less than that wouldn’t be material for us, given the position size.
The second case is when a working investment thesis starts to crack.
A recent example of this from our portfolio is Hims & Hers.
I bought this at $15 and rode it to $60s and didn’t sell even when it hit $70. Because the thesis didn’t crack. Then I sold it at a lower price, at $50 per share, because the thesis started to crack as the business posted its first QoQ decline, and margin pressure was obvious:
What changed? It stopped selling GLP-1 to broad audiences under the FDA’s 503B exemption, which applies when there is a shortage. After the shortage was resolved and it stopped selling to the broader public, we saw a significant slowdown. And there emerged a bigger threat of a lawsuit from Novo Nordisk, as Hims was still selling it under the personalization exemption.
There was no way it would reaccelerate the core business enough to compensate for all the revenue loss from GLP-1s. We saw this as growth declined to 3% in Q1 2026, and the stock declined to $20 levels.
The third case is when a position gets egregiously overvalued.
What I mean by this is that more than even the most optimistic future assumptions are currently priced into the stock.
Let’s go through the Nebius example, as I often get asked when I could completely exit Nebius.
It is committed to deploying 5 GW of Nvidia systems by 2030. There will likely be some missteps, but let us assume 5 GW will really be deployed because we want to see the optimistic outcome. Based on the current GPU/hour rates, 5 GW can translate to $50 billion of revenues. Assuming a standard 4x sales multiple, Nebius can be valued at $200 billion.
Discounting it back to today at 10% gives us a $124 billion business. If it reaches beyond this valuation, it means the most optimistic projections are already priced in, and there is nothing to gain by holding the stock. That’s when I would be willing to completely sell out.
This means double the current price:
These are the cases we exit, and if we don’t see one of these, we stay in the position.
🏁 Last Words
The market is at an all-time high, and it’s hard to find opportunities. Yet, I see that many people are forcing themselves to find opportunities to bet on.
That generally doesn’t end up well.
The best opportunities are often obvious, and they become obvious when we are at extremes. This is how we outperform the market.
The recipe is simple:
We believe only the edges we can bank on are analytical and behavioral.
Analytical edge can be banked on when we are at extremes.
We start small and double down only on a small subset of bets that got validated, but the stock price lags.
We stay in the position unless there are way better opportunities, an investment thesis starts cracking, or it becomes egregiously overvalued.
It’s not sexy, but systematic implementation of this pays off. It has paid off for us so far, and I am confident that it’ll continue this way.
Hope this helps.
That’s all friends!
Thanks for reading Capitalist-Letters!
Please share your thoughts in the comments below.
👋🏽👋🏽See you in the next issue!
Thank you very much, this article is a masterpiece.
One question: if the thesis is still intact and the opportunity is still there, but the position has already delivered a huge return and maybe has less upside left… do you make tactical reallocations to buy other businesses with better upside or a more attractive risk/reward profile?
Thanks again for the article.