Better Sharpe Ratio is a plumbing problem

Why investment returns are ultimately governed less by the quality of the trades an investor finds, and more by the number of times they can afford to place one.

Imagine I offer you a game.

You pay 50 cents to flip a fair coin. Heads, you get $1.50. Tails, you get nothing. That is a 50% expected return, which is obscene. People have started wars over less attractive economics.

But if I only let you play once, there is a 50% chance you walk away broke.

Now imagine I let you play 100 times.

Your expected return per flip is still 50%. Nothing has changed about the quality of the bet. But the probability of ruin (of flipping tails 100 times in a row) is 0.5^100, a number so small the universe would get bored and collapse before you finished writing it out.

The mean hasn't moved, but the tails have been pruned. Same coin. Same odds. Radically different outcome, just because you flipped it more.

Same edge, more bets, higher Sharpe

This is, I think, under-appreciated in the business of running money.

People talk endlessly about finding better trades - better coins. They talk much less about the infrastructure that determines how many times you get to flip. But the Sharpe ratio does not care about this distinction. It is roughly your expected return divided by the volatility of that return, and the math is relentlessly clear: same edge, more bets, higher Sharpe. Not because you got smarter. Just because you did it more.

This is, incidentally, the entire secret of Renaissance Technologies.

“Jim Simons did not find a magic coin. He built a machine that could flip one thousands of times a day.”

The cost of curiosity

So what determines how many times you flip?

For most fundamental investors, it is the cost of curiosity.

An investor reads something interesting about, say, the relationship between container shipping rates and semiconductor inventories. She thinks: huh, I wonder if that is true, and if so, whether it is priced in. What happens next is the part nobody romanticizes.

She needs data from multiple providers, in different formats, with different quirks. She needs to figure out if these data sets even describe the same universe. Then she has to clean everything, merge it, model it, and debug the model when it breaks.

This takes weeks, and the whole time she doesn't know if the question was worth asking.

The cost of curiosity is not just high. It is front-loaded. You pay the full price before learning if there was anything to find.

This is the bottleneck. Not "what do I think" but "can I afford to find out."

So the curiosity dies on the vine. She does her 10 trades a year. They are good trades. But she walked past 90 others that might have been just as good, because the data work was too expensive.

She is the person flipping the coin 10 times when she could have been flipping it 100.

What we’re building at Reflexivity

I want to be precise about the claim. The claim is not "we make investors smarter." The claim is "we make it radically cheaper to be curious."

Most approaches to making data easier give you a chatbot on top of a database. You ask a question, it writes a SQL query, and maybe the answer is what you wanted. But this fails when the investor doesn't yet know what to ask. When the question requires understanding that shipping rates from Provider A and inventory data from Provider B are talking about the same supply chain from different angles, using different taxonomies.

This is what Reflexivity's Knowledge Graph solves. A Knowledge Graph doesn't just store data. It stores relationships between data.

It knows that Company X is a subsidiary of Company Y, a customer of Company Z, which ships through Port A. The relationships are first-class citizens, not afterthoughts.

So when our investor wonders how shipping rates connect to chip inventories, it is not a data engineering project. It is a query.

But a map without a guide is just a poster.

Reflexivity's reasoning engine is the thing that takes a loosely formed question, navigates the graph, and comes back not just with data but with analysis and, often, with three related questions the investor should have asked but didn't. The Knowledge Graph makes the data connectable. The reasoning engine makes it navigable. Together, they collapse the cost of curiosity from weeks to minutes.

And that changes the math. Remove the data plumbing constraint and a 10-flip-a-year investor becomes a 100-flip-a-year investor. The edge hasn't changed. The process hasn't changed. But the variance around her expected return has compressed, dramatically, because she expressed that edge enough times for the law of large numbers to show up and do its thing.

Making data analysis easy is not a productivity story.

It is a risk-adjusted return story.

The coin is the same coin. The only thing that changed is how many times you get to flip it.