Why win rate is not the edge

A 90% win rate can be worse than a 60% win rate.

It sounds counterintuitive, but it is one of the principles that shaped how we build systematic strategies at Montcrest Capital.

Most traders naturally want to win more often.

But win rate only answers one question:

How often are you right?

It says nothing about how much you make when you are right, how much you lose when you are wrong, or how much risk sits beneath the surface.

The illusion of a higher win rate

During the development of one of our Bitcoin models, we tested what happened when stops were progressively widened.

The result looked impressive.

The historical win rate increased from approximately 61% to 82%.

But the strategy became worse.

Its expectancy declined from approximately +0.23R to +0.05R per unit of risk.

We were winning more often, while extracting less value from the risk we were taking.

That distinction is fundamental.

A strategy does not create value because it wins frequently.

It creates value when the mathematical relationship between its winners, losers and risk remains favorable over time.

The 100% win rate problem

Take the idea one step further.

It is technically possible to construct a strategy with a historical win rate approaching 100%.

One method is remarkably simple:

Never realize a loss.

Remove the stop, keep losing positions open and wait for price to eventually return to the target.

The trade ledger can look almost perfect.

The portfolio does not.

In our Bitcoin research, a no-stop variation produced 17 winners out of 17 trades, while one position experienced an adverse excursion of roughly -40R before recovering.

The loss had not disappeared.

It had simply moved somewhere the win-rate statistic could not see it.

This is why we view unusually attractive metrics with skepticism rather than excitement.

What Montcrest optimizes instead

At Montcrest, our objective is not to build the algorithm with the most attractive headline statistic.

We optimize for something harder:

repeatable expectancy under controlled risk.

That means evaluating strategies across several dimensions:

• positive expectancy

• out-of-sample robustness

• drawdown

• execution

• scalability

• correlation with the rest of the portfolio

This philosophy also explains why Montcrest is not built around a single trading algorithm.

Our research architecture combines systematic strategies operating across different market horizons, from intraday execution to swing structures and longer-term Bitcoin cycles.

Each model is expected to contribute a different source of return rather than simply amplify the same underlying bet.

The edge is the system

A high win rate can be attractive.

A high return can be attractive.

An exceptional backtest can certainly be attractive.

But none of them, independently, constitutes an investment process.

The real edge lies in building strategies that survive scrutiny, combining them intelligently, and understanding exactly where their returns come from and where they can fail.

That is the philosophy behind the algorithms we are building at Montcrest Capital.

We are not trying to win every trade.

We are building a system designed to win over time.