Validation

Monte Carlo

Why

Profit inside one history may be carried by a few fills in a row. The already locked fills are cut into blocks in time order and drawn with replacement, to see whether a poorer assembly still makes money.

Formula

What is drawn

Only the already locked fills on the discovery-sample user book are drawn. The sealed sample is not drawn. Each fill uses cash, not a percent.

Long cash = (exit price − entry price) × quantity × contract multiplier − fee cash − funding cash
The price term for a short is the opposite

The cost is already inside that cash: taker 0.05, maker 0.02 (percentage points), slippage of 10 basis points, and funding of 0.0001 per 8 hours.

Block length

n = number of discovery-sample fills
L = min( floor(n / 2) , max( 5 , floor(√n) ) )

The fills are cut in time order into non-overlapping blocks. A block has length L. The last block may be shorter. The order inside a block is kept. The last block is not wrapped onto the first.

Redraw

Draw these blocks with replacement
Assemble them back into n fills
Cash sum of this draw, C = sum of the cash of the n assembled fills

Changing only the order of the same fills does not change the sum. That is not this redraw. Breaking each fill apart and shuffling them is not this redraw either.

Pass

The 5th percentile of these C values > 0

When n < 10, this gate is not passed. Even if a 5th percentile can be computed from fewer than 10 fills, that number is not a pass. A larger or smaller L makes the lower band a different number.