Balance_Of_Power

Unguarded: PASSGuarded: FAIL · 3 failDirectionalgunbot-quant

BTCUSDT · 2026-08-13 01:50 → 2026-09-12 01:50 UTC (30 days) · 0.03% fee + 2 bps slippage · 1,000 quote, one figure for every cell

The shipped configuration, both ways

Both columns are the same strategy on the same month of BTCUSDT at 1m with no posture overrides — the configuration as it ships. The only difference between them is the break-even guard.

	As it ships	Break-even guard	Difference
Return	-23.62%	9.55%	+33.18%
Buy and hold, same windowthe benchmark, identical in both runs	21.88%	21.88%	—
Edge vs buy and holdwhat it did against simply holding the asset	-45.51%	-12.33%	+33.18%
Max drawdown	23.82%	5.57%	-18.25%
Round tripsclosed trades — how much it actually did	714	32	-682
Time in marketat 100% it is holding the asset, not trading it	10.35%	98.27%	+87.92%
Exits refused by the guardevery sell the guard blocked for being below cost	—	81,024	—
Realisedbanked by closed trades, in quote	-236	112	—
Unrealisedstill open at the end of the window, in quote	0	-16	—
Fees paid	147	15	-132
Closed at a lossthe guard cannot close a losing trade, so the loss stays open instead	694	0	—
Win rateof closed trades — under the guard nothing closes at a loss, so this is not an achievement	2.80%	100.00%	—
Profit factorgross win ÷ gross loss, and the gross loss is zero	0.12	undefined	—

What the guard is, and what a better number under it means. The break-even guard refuses any exit below average entry plus fees. It cannot improve a strategy's judgement — it can only stop it selling at a loss, which means holding the position instead. Across the pack that turns a strategy that churned into one that buys and waits: trade counts collapse and time in market rises towards 100%. In a window where the asset itself rose, that reads as a much better return while being strictly less of a strategy.

So read the return next to round trips, time in market and edge vs buy and hold. The last of those is the one that matters: not one strategy in this matrix beat simply holding the asset, in either run.

Ignore the win rate and the profit factor under the guard. Both count closed trades only, and the guard closes nothing at a loss — in all 101 guarded runs the number of losing closed trades is exactly zero. So the win rate goes to 100% and the profit factor becomes undefined for arithmetic reasons, while the losses sit in the position that is still open. The unrealised row is where they went.

All fifteen configurations

Return in every cell. medium at 1m — the highlighted cell — is the strategy exactly as it ships; calm and aggressive are the measured extremes of its own posture settings. One capital figure is used for all fifteen so they stay comparable, which means an aggressive posture can run into the cash wall — as a real operator would.

As it ships — the strategy's own exits, stops included
	calm	cautious	medium	bold	aggressive
1m	-6.34%	-11.53%	-23.62%	-31.24%	-22.66%
3m	-2.60%	-4.62%	-9.49%	-7.20%	-4.74%
5m	-1.65%	-2.21%	-2.79%	-7.45%	-2.75%

With the break-even guard — no exit below average entry plus fees
	calm	cautious	medium	bold	aggressive
1m	1.04%	1.20%	9.55%	14.17%	15.84%
3m	0.42%	2.82%	6.36%	14.10%	15.91%
5m	1.57%	2.93%	6.53%	12.38%	13.61%

The run as it ships

What was run


Timeframes	1m · 3m · 5m
Postures	calm · cautious · medium · bold · aggressive
Cells	15
Bars	1m: 45,801 · 3m: 17,000 · 5m: 11,241
Settings read	77, of which 40 are posture settings and 5 measurably move the run

Correctness checks

The invariants the matrix asserts. This is a correctness result, not a profitability one — a strategy can pass every check and still lose money.

Check	Result	What it means
`responds-to-config`	pass	each profile produces a distinct set of trades at every timeframe
`responds-to-timeframe`	pass	each timeframe produces a distinct set of trades
`posture-orders`	pass	mean deployed capital rises from calm to aggressive at every timeframe
`cost-independent`	pass	trade count is not an artefact of the fee model
`deterministic`	pass	re-running a cell reproduces it exactly

Every one of the 15 cells passed its own order-path checks.

How to read this

medium is the strategy exactly as it ships — zero overrides. calm and aggressive are the measured extremes of its own posture settings, and cautious and bold sit at the geometric midpoint either side of shipped. Which direction of a knob is the brave one was measured, not assumed.

One capital figure is used for all 15 cells, calibrated so the shipped configuration is fully deployed at its peak. That makes the cells comparable, and it means an aggressive posture can run into the cash wall — which a real operator would hit too.

A result here is a measurement on one window of one market. It is not a forecast, and it is not advice to run anything unattended.

Every number on this page is read from the run record that produced it — the audit pipeline writes the page and the record together, so the two cannot drift apart.

The run with the break-even guard