Where the Money Actually Went

A strategy we audited lost $72,983 across 11,888 trades. It had a real edge.

Gross P&L at the price level, meaning exit minus entry times size on every fill, was positive $22,121. Costs across the book were $95,104. Net, minus $72,983.

It won 54.6 percent of its trades and averaged +$1.86 a trade before costs, then paid $8.00 a trade to express that. The edge is not missing. It is 4.3 times too small to pay for itself.

One decomposition explains the whole book

Once you see the gross and the costs separately, everything else about this book stops being mysterious.

The equity curve decays in an almost perfectly straight line across four years, because a constant per-trade drag produces a straight line. No blowup, no regime break, just arithmetic.

Longs lost $36,213 and shorts lost $36,771, which is nearly identical, because costs do not care about direction. If a book loses the same amount on both sides, the first suspect should not be the trader's bias. It should be the toll both sides are paying.

And 1,299 trades, 10.9 percent of the book, finished on the right side of their entry and still lost money, averaging minus $3.58 each. Fee-sized losses, not thesis-sized ones. Almost nobody tracks how many of their losing trades were actually losses and how many were fees.

The second win rate

There is a number hiding inside every book that costs money to trade: the gross win rate. This book wins 65.6 percent of the time at the price level and 54.6 percent net. That eleven-point gap is the cost drag expressed as a rate, and it is the same 1,299 right-but-lost trades seen from the other side.

Most analysis of a losing strategy stops at no edge, move on. That verdict would have been wrong here, and it would have thrown away a working signal because of an execution problem.

The actionable version

The useful output of this decomposition is not that the book is down. It is a target: break-even cost for this strategy is $1.86 a trade, it is paying $8.00, so it becomes profitable at 77 percent lower execution costs, or with 4.3 times more gross edge per trade. The levers are the usual three. A cheaper venue or fee tier. Maker instead of taker. Fewer trades with larger targets, so each toll is paid against a bigger move.

This decomposition now runs inside NoxarQuant as the cost diagnosis on the Portfolio page. Not whether you are up or down, but where the money actually went, and whether the thing underneath is real.

The disclosure

Costs on this book are modelled at $8.00 a trade flat on fixed $10k notional, which is 0.04 percent per side. That is realistic taker territory, but it is a model, not reconstructed exchange fills. The conclusion is robust to the assumption: even at half those costs, the edge is still 2.2 times too small. Funding on held positions is not included, so the true gap is wider than stated, not narrower.